Also known as the App Store Accountability Act, SB 2420 is currently facing a major constitutional challenge before the 5th U.S. Circuit Court of Appeals. The consolidated cases—Students Engaged in Advancing Texas (SEAT) v. Paxton and Computer & Communications Industry Association (CCIA) v. Paxton—pit challengers against Texas Attorney General Ken Paxton and place the intersection of free speech and government regulation of technology platforms squarely before the court. 

SB 2420 requires app stores to verify the age of every user and mandates that minors obtain individualized parental consent before downloading or purchasing any app.

Texas argues that the law merely strengthens parental authority. The U.S. District Court for the Western District of Texas was unconvinced. As the district court explained: 

The Act is akin to a law that would require every bookstore to verify the age of every customer at the door, and for minors, require parental consent before the child or teen could enter and again when they try to purchase a book.

The case is now on appeal before the 5th Circuit. The International Center for Law & Economics (ICLE) filed an amicus brief supporting the plaintiffs and arguing that SB 2420 violates the First Amendment. The brief’s distinctive contribution is to connect First Amendment doctrine with an underlying law & economics framework, building on my prior ICLE issue brief, “A Coasean Analysis of Online Age-Verification and Parental-Consent Regimes,” as well as several earlier Truth on the Market posts examining app-store age-verification requirements. 

The Wrong Gatekeeper

ICLE’s brief centers on the concept of the “least-cost avoider.” In law & economics, the principle holds that liability should generally fall on the party best positioned to prevent harm at the lowest overall social cost. As I have noted elsewhere: 

From the perspective of law & economics, the most important question in disputes such as these is to identify the lowest-cost avoider of harms, while bearing in mind relevant transaction costs. It is not necessarily the case that the answer is always the user. It could be that online intermediaries are best-positioned to monitor and control harms to those users. But among the relevant social costs we must consider in this example is the risk of collateral censorship.

That risk is central to SB 2420. When the law makes app stores responsible for preventing potential harms to minors, it creates powerful incentives to overrestrict access to lawful content. To avoid liability, app stores must erect broad age-verification and parental-consent barriers around vast amounts of harmless, educational, and expressive material. The result is collateral censorship: protected speech becomes harder to access, not because it is harmful, but because platforms seek to avoid legal risk. 

App stores are poorly positioned to make these judgments. Requiring them to police every download and purchase imposes substantial economic friction while restricting access to First Amendment-protected expression.

Under SB 2420, minors cannot access a large swath of the modern marketplace of ideas without first obtaining a government-mandated digital permission slip. A minor could be blocked from freely accessing: 

• Current events through The New York Times app;
• Local weather information through The Weather Channel app;
• Vocabulary definitions through the Merriam-Webster Dictionary app;
• Scripture through the YouVersion Bible app; or
• Educational and creative content on platforms such as YouTube and Instagram.

Yet the Supreme Court has repeatedly recognized that minors enjoy substantial First Amendment protections. The government lacks a “free-floating power” to dictate which ideas children may encounter. SB 2420 nevertheless places state-mandated barriers between minors and a wide range of constitutionally protected speech. 

Parents and minors, by contrast, are often the true least-cost avoiders. They already have access to voluntary tools—including content filters, parental controls, and built-in app-blocking features—that allow families to tailor restrictions to their own preferences and circumstances. That reality also aligns with a core principle of First Amendment doctrine: when less restrictive alternatives exist, the government cannot justify burdening protected speech with broad, one-size-fits-all regulations. 

Calling Speech a Contract Doesn’t Make It One

On appeal, Texas has tried to minimize the First Amendment issues by recasting SB 2420 as a mere contract regulation. Because app stores require users to accept terms of service and often monetize user data, the state argues that it may freely regulate a minor’s ability to enter those arrangements. 

The 5th Circuit accepted a version of that argument when it stayed the district court’s preliminary injunction, treating the law as a regulation of “commercial speech” and subjecting it to the less demanding standard of intermediate scrutiny. 

Federal courts around the country, however, have repeatedly rejected similar arguments. As ICLE’s brief explains, there are at least two reasons why. 

First, speech does not lose First Amendment protection simply because someone earns money from it. A bookstore does not forfeit its constitutional rights because it sells books for profit, and an app store is no different. The Supreme Court has long recognized that entities engaged in expressive activity retain First Amendment protection even when they operate as businesses. 

Second, any commercial aspects of app-store transactions are “inextricably intertwined” with fully protected speech. App stores are multi-sided platforms that connect developers and users. Many apps are offered for free or at reduced cost because developers and platforms rely on advertising revenue generated through user data. The economic transaction and the dissemination of speech are therefore inseparable. 

For that reason, SB 2420 cannot be treated as a simple commercial regulation. Because the law restricts access to fully protected speech based on content, the district court correctly concluded that it must satisfy strict scrutiny—the most demanding standard of constitutional review. 

Parental Controls, Not State Controls

To survive strict scrutiny, a law must be the least restrictive means of advancing a compelling government interest. Even assuming the state has a compelling interest in protecting minors from potentially harmful content, SB 2420 fails that test because it is both overinclusive and underinclusive. 

The law is overinclusive because it restricts access to vast amounts of harmless, First Amendment-protected speech. It also forces minors—and adults who value their privacy—to surrender sensitive age-verification information merely to use an app store. For many apps, parents would likely have no objection to a download. Yet in a world with real transaction costs, requiring parental approval for every app and purchase means some speech will inevitably be restricted regardless of parental preferences. 

At the same time, the law is underinclusive. A minor can bypass the app store entirely by opening a pre-installed browser such as Safari or Chrome and accessing the same allegedly harmful content directly on the web. 

SB 2420 fares no better under intermediate scrutiny. The law burdens substantially more speech than necessary to achieve its stated objectives. There is no sound basis for restricting access to speech across every app simply because a small subset of apps may contain objectionable content. 

Helping parents protect their children online is a legitimate and worthwhile goal. The problem is that Texas chose one of the most speech-restrictive ways to pursue it. Parents already have access to voluntary tools, including content filters, parental controls, and app blockers. Market demand has given app stores strong incentives to provide those tools, and families remain free to use them as they see fit. 

The 5th Circuit should affirm the preliminary injunction. The First Amendment problem with SB 2420 is not that it seeks to protect children. It is that they try to accomplish that goal using the wrong gatekeeper. The least-cost avoiders are parents and minors—not the app stores that stand between them and the modern marketplace of ideas.

The post Texas Wants to Check Your App Store Papers appeared first on Truth on the Market.

The Act was designed to speed wireless deployment while preserving local control over routine land-use decisions. Section 332(c)(7) reflects that compromise. It preserves state and local authority over the “placement, construction, and modification” of wireless facilities, while imposing a handful of federal constraints to ensure Americans receive the benefits of timely wireless-service deployment. 

The most important of those constraints is the effective-prohibition clause, which provides that local regulation “shall not prohibit or have the effect of prohibiting the provision of personal wireless services.” The key words are “or have the effect of prohibiting.” That language extends beyond outright bans to government actions that, whatever their form, leave an area without wireless service. For nearly 30 years, courts have wrestled with a recurring question: How far does that functional phrase reach? 

Faced with a statute that condemned effects without defining them, the federal courts of appeals developed a framework to fill the gap. Beginning with the 2nd U.S. Circuit Court of Appeals in Sprint Spectrum, L.P. v. Willoth (1999), and eventually adopted by nearly every circuit to consider the issue, courts converged on the “significant gap” test. Under that approach, a denial has the effect of prohibiting service when it leaves a significant gap in a carrier’s coverage and the carrier’s proposal is the least intrusive means of closing it. The test gave concrete meaning to the statute’s “effect of” language, tied liability to real-world coverage rather than the label a locality attached to its decision, and used the no-alternatives requirement to supply the causal connection implied by the word “effect.” 

Last month, the 11th U.S. Circuit Court of Appeals broke from that consensus. In T-Mobile South, LLC v. City of Roswell, the court held that the effective-prohibition clause governs only the regulation of siting—that is, control through generally applicable rules—and therefore cannot be invoked to challenge the denial of a single permit application. 

That reading is difficult to square with the statutory text. The phrase “effect of prohibiting” is at least as naturally read to reach functional prohibitions as formal ones. The court’s narrower interpretation also carries consequences that cut against the deployment Congress sought to accelerate. Under the 11th Circuit’s approach, a locality can keep wireless facilities out indefinitely by denying applications one at a time, each on seemingly site-specific grounds, without ever adopting a rule that a court could invalidate. The result is a moratorium in all but name—effectively insulated from challenge because no one put it in writing. 

The Consensus Nobody Argued About

For nearly 30 years, most federal courts of appeals operated on a shared—and largely unexamined—premise: a city’s denial of a single wireless-facility permit can “have the effect of prohibiting” service under 47 U.S.C. § 332(c)(7)(B)(i). The debate was rarely whether an individual denial could violate the statute. The real question was how courts should determine when it does. 

The framework emerged almost immediately after enactment of the Telecommunications Act of 1996. In Sprint Spectrum, L.P. v. Willoth (1999), the 2nd U.S. Circuit Court of Appeals held that a denial effectively prohibits service when it prevents a carrier from remedying a significant gap in coverage and the proposed facility is the least intrusive means of closing that gap. 

Just as important, the court rejected the argument that Section 332 reaches only general bans on wireless service: 

Absent an explicit policy banning personal wireless services, the Board contends, courts can only consider whether in aggregate a town’s repeated denials of applications have the effect of a general ban. Since Ontario does not have a general ban in effect, as evidenced by its earlier approval of Frontier’s application and professed willingness to accept some level of PCS service, the Board insists its actions must necessarily be in conformance with subsection B(i)(II). We disagree with this reasoning.

Not every circuit followed the same path, but most arrived at the same destination. The 4th U.S. Circuit Court of Appeals initially suggested that the effective-prohibition clause targeted only general bans. See AT&T Wireless PCS, Inc. v. City Council of Virginia Beach (1998). Yet the court soon recognized that an individual permit denial could itself amount to an effective prohibition when it prevents a carrier from remedying a significant coverage gap and no reasonable alternatives exist. See 360 Communications Co. of Charlottesville v. Board of Supervisors of Albemarle County (2000). The court reaffirmed that understanding in T-Mobile Northeast LLC v. Fairfax County Board of Supervisors (2012), while emphasizing that the statute does not guarantee carriers flawless coverage. The burden may vary across jurisdictions, but the basic premise remained the same: individual siting decisions can be challenged under Section 332. 

Stripped to its essentials, the resulting “significant gap” test asks two questions. First, does a significant gap in wireless coverage exist in the relevant area? Second, do reasonable alternatives to the carrier’s proposed facility mean that service can be provided without the denied application? That two-step inquiry became the dominant framework for effective-prohibition claims nationwide. 

Over the next decade, most circuits adopted some version of the significant-gap approach. The 7th U.S. Circuit Court of Appeals did so in VoiceStream Minneapolis, Inc. v. St. Croix County (2003); the 9th U.S. Circuit Court of Appeals in MetroPCS, Inc. v. City & County of San Francisco (2005); and the 6th U.S. Circuit Court of Appeals in T-Mobile Central, LLC v. Charter Township of West Bloomfield (2012). 

If anything, more recent developments expanded the reach of the effective-prohibition clause rather than narrowing it. In a 2018 declaratory ruling, the Federal Communications Commission (FCC) recast the inquiry around what it called “material inhibition.” Under that standard, a state or local requirement effectively prohibits service whenever it materially inhibits a provider’s ability to offer a covered service. The FCC explained that the standard extends beyond coverage gaps to include barriers to network densification, capacity upgrades, and deployment of new technologies such as 5G. A locality need not erect an insurmountable obstacle; a material one is enough. 

The FCC derived the test from its 1997 California Payphone decision and applied it to the parallel effective-prohibition language found in both Sections 253 and 332. Although the ruling focused primarily on broad regulatory barriers, the 3rd U.S. Circuit Court of Appeals applied the standard to an individual permit denial in Cellco Partnership v. White Deer Township Zoning Hearing Board (2023). The court expressly abandoned its earlier significant-gap framework and instead asked whether, under the totality of the circumstances, the denial prevented the carrier from providing service without unreasonable cost. Where the significant-gap cases at least attempted to cabin liability, the material-inhibition standard lowered the bar further. 

The 11th Circuit Changes the Question

Last month, the Eleventh Circuit decided that the widespread application of section 332 to individual siting decisions was incorrect. Instead,  the court held that the statute’s bar on conduct that has “the effect of prohibiting” wireless service does not reach an individual permit denial at all. It limits only “[t]he regulation of” siting — and “regulation,” the court concluded, means control by rule, not the disposition of a single application.

Last month, the 11th U.S. Circuit concluded that nearly three decades of effective-prohibition jurisprudence had gone off track. In Roswell, the court held that Section 332’s prohibition on conduct that has “the effect of prohibiting” wireless service does not reach the denial of an individual permit application. Instead, it applies only to “[t]he regulation of” wireless-facility siting. And, in the court’s view, “regulation” means control by rule, not the disposition of a single application. 

The court’s reasoning is relentlessly textual. Because “regulation” derives from the verb “regulate,” the court began with contemporaneous dictionary definitions, which generally described regulation as controlling or directing conduct “according to rule, principle, or law.” A locality that denies a permit, the court reasoned, is not regulating property through a rule. It is merely refusing to grant one applicant an exception to rules already in place. The actual constraints on where a wireless facility may be built come from the ordinance itself, not from any individual permitting decision. 

The court found further support in Section 332’s structure. Subsection (c)(7)(A) preserves local “authority . . . over decisions” regarding wireless-facility siting. The limitations that follow in subsection (c)(7)(B), meanwhile, divide neatly into substantive and procedural categories. 

The substantive limits—the prohibitions on unreasonable discrimination and effective prohibition in subsection (B)(i), as well as the radiofrequency-emissions provision in subsection (B)(iv)—speak in terms of “regulation.” The procedural limits—the requirement to act within a reasonable time in subsection (B)(ii) and the requirement that denials be supported by substantial written evidence in subsection (B)(iii)—speak instead of a “decision” to “deny a request.” 

For the court, that distinction mattered. Congress had available a term broad enough to encompass both generally applicable rules and individual permit denials: “decisions.” Yet it chose “regulation” for the statute’s substantive constraints. When Congress uses materially different terms, courts generally presume it intends materially different meanings. On that logic, “regulation” is a subset of “decisions,” not a synonym for them. 

The implications are sweeping. The court held that the significant-gap test “fails at the threshold” because it focuses on the wrong object. The relevant question is not whether a permit denial leaves a coverage gap, but whether a governing rule has the effect of prohibiting service. The panel was openly skeptical of the significant-gap framework’s pedigree, describing it as the product of judicial “common-law rulemaking” untethered from statutory text and suggesting that its widespread adoption was evidence that “something has gone badly wrong.” 

Although the court framed its criticism in terms of the significant-gap test, its reasoning extends much further. The FCC’s material-inhibition standard likewise asks whether government action functionally impedes deployment. If the statute reaches only rules and not individual denials, that approach appears vulnerable for the same reason. As the court put it, its interpretation is “irreconcilable with any version of the significant gap test.” 

Still, the court did not eliminate effective-prohibition claims altogether. Providers may continue to challenge local regulations that operate as barriers to wireless deployment. A carrier that can show ostensibly permissive siting rules function as a de facto ban, or that a consistent pattern of denials reflects an unwritten policy against wireless facilities, may still prevail. What the carrier may not do, under the 11th Circuit’s approach, is treat the denial of a single permit application as the prohibited act itself. 

The Evasion Problem

The significance of Roswell lies not only in its sharp break from existing precedent, but also in how that break cuts against the purposes of the Telecommunications Act. If the decision stands, its effects may extend well beyond the 11th Circuit, creating new obstacles to wireless deployment and increasing uncertainty for providers nationwide. 

As a textual matter, the court’s critique of the significant-gap test is less persuasive than it first appears. Congress did not merely prohibit regulations that “prohibit” wireless service. It prohibited regulations that “prohibit or have the effect of prohibiting” the provision of wireless service. An outright ban already “prohibits.” The additional phrase must therefore do some independent work. 

The most natural reading is that Congress sought to reach government actions that, whatever their form, make service unavailable in practice. Viewed through that lens, the significant-gap test is not free-floating common law. It is an attempt to give operational meaning to the statute’s “effect of prohibiting” language. The test’s significant-gap requirement tracks the statutory concern with the provision of service, while its no-reasonable-alternatives requirement supplies the causal connection implied by the word “effect.” Without the alternatives inquiry, a court cannot determine whether a denial actually caused a service gap or merely foreclosed one of several viable paths to coverage. 

Indeed, even the 4th U.S. Circuit’s decisions suggest that the significant-gap framework can be used to determine whether an ostensibly individual siting decision is functionally equivalent to a broader policy of exclusion. Under that approach, a permit denial can violate Section 332 when it effectively operates as a prohibition, regardless of whether the locality formally labels it as such. In that respect, the significant-gap test is not obviously inconsistent with the logic that Roswell claims to embrace. 

The practical consequences of the decision are more troubling. Under Roswell, a single denial can never establish that a locality has effectively prohibited wireless deployment. Instead, a provider must identify a written or unwritten rule that forecloses service. That approach creates an obvious avenue for evasion. 

A locality determined to keep wireless facilities out need not enact a moratorium or adopt an exclusionary ordinance. It can simply deny applications one by one, each supported by seemingly site-specific findings about aesthetics, neighborhood character, or the particular parcel at issue. So long as no formal rule emerges, the locality may achieve the same result while avoiding scrutiny under the effective-prohibition clause. The result is a moratorium in all but name. 

That concern is hardly hypothetical. In its 2018 Small Cell Order, the FCC concluded that de facto moratoria—local actions that effectively halt deployment without formally banning it—can themselves constitute effective prohibitions. Roswell points in the opposite direction. Rather than looking at whether deployment has been blocked in practice, it rewards localities that accomplish the same objective through a series of individualized denials.

The panel acknowledged this concern but offered only a limited response. Providers, the court explained, remain free to challenge unwritten rules or to show that ostensibly permissive ordinances are merely a “fig leaf” for a prohibition. Yet the court also described that burden as “heavy.” The problem is that proving an unwritten policy is most difficult when intervention is most valuable: early in the process, before enough denials have accumulated to reveal a pattern. By the time a provider can demonstrate a systematic practice of exclusion, deployment may already have been delayed for years.  

The court’s fallback answer is that Congress remains free to amend the statute. But that response assumes the statute does not already address the problem. A stronger reading is that Congress addressed it directly. The phrase “have the effect of prohibiting” instructs courts to look beyond form and examine function. The significant-gap test, whatever its imperfections, represented a serious effort to do exactly that. Roswell replaces that inquiry with one focused largely on whether the locality had the foresight to avoid putting its prohibition in writing. 

The decision may also inject uncertainty into wireless deployment far beyond the 11th Circuit. The court did not reject the significant-gap framework on its own terms. Instead, it held that providers must first show a broader regulatory policy before the framework becomes relevant. Other courts could adopt that threshold requirement without formally repudiating their existing precedents. 

That uncertainty matters. Wireless infrastructure projects require substantial upfront investment, and providers must assess regulatory risk before committing capital. If carriers can no longer assume that an unlawful denial can be challenged promptly under Section 332, some projects—particularly those with more modest expected returns—may never move forward. The result would be slower deployment, weaker coverage, and fewer options for consumers at a time when demand for wireless capacity continues to grow. 

How to Ban Towers Without Banning Towers

Roswell marks a sharp break from nearly 30 years of Telecommunications Act jurisprudence. By insisting that only a “rule” can violate the effective-prohibition clause, the 11th U.S. Circuit removed the most common effective-prohibition claim from federal court and handed localities a roadmap for resisting wireless deployment without adopting a policy providers can challenge directly. 

That result is hard to square with the Act’s text or purpose. Congress barred not only regulations that prohibit wireless service, but also those that “have the effect of prohibiting” it. The significant-gap test was imperfect, but it took that command seriously by asking whether local action actually prevented service from being provided. 

Roswell trades that functional inquiry for formalism. Under its logic, the problem is not whether a locality blocks deployment, but whether it does so through a rule rather than one denial at a time. 

Congress enacted Section 332 to keep local authority from becoming a bottleneck to wireless deployment. The 11th Circuit risks turning that bottleneck into a blueprint.

The post The Roswell Loophole: How to Stop Wireless Deployment One Permit at a Time appeared first on Truth on the Market.

Sen. Chuck Grassley (R-Iowa) and Sen. Amy Klobuchar (D-Minn.) “introduced” AICOA last week. Co-sponsors include Sens. Dick Durbin (D-Ill.), Josh Hawley (R-Mo.), Sheldon Whitehouse (D-R.I.), and Cory Booker (D-N.J.). Here is the bill’s text, at least as introduced. 

“Introduced” is technically correct as a matter of process. And, as far as I know, this precise text string has not been introduced before. But we have seen pretty darn similar bills, under the same title, from Klobuchar before. AICOA appeared as S. 2992 in the 117th Congress (we’ll call that AICOA 1.0) and S. 2033 in the 118th Congress (AICOA 1.1). There have been changes along the way, but the essentially bad idea remains, in essence, bad. 

I don’t know whether AICOA 1.2’s bite at the apple will lead to anything more than the others did. There’s rather a lot going on, and I think I read something about an election to be held in November of this year. My best guess is that this version, too, will not pass. 

Then again, I lack a crystal ball. And there is bipartisan interest—not least among populists on the right and the left—in doing something to the sector. So I’m less sanguine about all this than I’d like to be. Is the third time the curse? 

Same Same But Different, And Not Better 

My International Center for Law & Economics (ICLE) colleague Geoffrey Manne has a helpful initial summary under the equally helpful title, “Revised AICOA Doubles Down on Flawed Antitrust Shortcut.” Manne’s top-level observation is this: 

The latest version of [AICOA] fails to fix the bill’s central legal and economic flaws—and in several ways makes them worse. 

That seems right, as do Manne’s more specific objections:

AICOA’s economic assumptions remain flawed. The bill treats vertical integration, self-preferencing, and default settings as suspect, even though those practices often make products better, safer, and easier to use. …

It lowers the competition-harm threshold to anything above de minimis. …

[It] raises the burden for key safety, privacy, and security defenses. …

The bill also replaces the earlier enforcement-guidelines process with expedited litigation provisions that direct courts to prioritize these cases and seek final judgment within one year. …

While the bill lowers the maximum civil penalty from 15% to 10% of U.S. revenue during the violation period, it adds a 1% floor once penalties are imposed. That is not leniency . . . It is a new floor for enormous fines, even in close cases involving unsettled legal questions. …

To that I’d add, among other things, a basic “big is bad” assumption lifted straight from the neo-Brandeisian playbook. See, for example, Tim Wu’s “The Curse of Bigness” and, for one possible if counterproductive implementation, the European Union’s Digital Markets Act. 

Analyses of AICOA 1.0 and AICOA 1.1 remain unfortunately relevant. Geoff’s list of prior ICLE scholarship is therefore useful as a current resource:

• “10 Things the American Innovation and Choice Online Act Gets Wrong” — Dirk Auer (Jan. 18, 2022)
• “How Tech Startups Could Be a Casualty of the War on Self-Preferencing” — Geoffrey A. Manne (Jun. 3, 2022)
• “AICOA Is Neither Urgently Needed Nor Good” — Thomas A. Lambert (July 25, 2022)
• “The Catch-22 of AICOA’s Guidelines” — Geoffrey A. Manne (July 25, 2022)
• “Reining in Digital Competition to No Good End: Will AICOA and OAMA Rise from the Grave?” — Daniel J. Gilman and Lazar Radic (Mar. 8, 2023) 
• “Regulate for What? A Closer Look at the Rationale and Goals of Digital Competition Regulations” — Lazar Radic, Geoffrey A. Manne, and Dirk Auer (Apr. 1, 2025) 

Many of our concerns were widely shared. Reviewing AICOA 1.0—the version from the 117th Congress—in the Michigan Technology Law Review, Herbert Hovenkamp summarized his view neatly: “AICOA was a bill that deserved to die.” 

His concerns were many, and familiar. Noting AICOA’s general hostility to self-preferencing—when a firm favors its own products or services on its own platform—Hovenkamp observed that “[s]elf-preferencing is an essential tool of competition, which has never imposed a requirement that people must sell other people’s merchandise.” He also objected that AICOA’s prohibitions applied “to products and services over which the seller has little or no market power. As a result, its substantive requirements are egregiously mistargeted.” 

Hovenkamp also anticipated that AICOA’s “gatekeeper” approach to competition policy would resurface—as indeed it has—despite the bill’s failure to gain traction in the 2021-2022 session: 

The issue … will almost certainly be considered again by Congress. When it does so, the “gatekeeper” approach to competition policy should be abandoned. It is too narrow because it ignores the conduct of firms that are not designated as gatekeepers, including offline sellers who are not included no matter what their size. It is too broad because it overreaches, perhaps egregiously, to condemn competitively harmless conduct by firms defined as gatekeepers.

That was hardly a defense of the antitrust status quo. To the contrary, Hovenkamp suggested that “[u]nderenforcement is a serious problem.” Set aside debates about the degree or locus of such problems. Hovenkamp recognized AICOA’s fundamental flaws as a vehicle for competition-policy reform—one with consequences likely, in some cases, to be unpredictable and, in others, “just plain bad.” 

Similarly, New York University’s Daniel Francis presented testimony on AICOA 1.1 before the Senate Judiciary Subcommittee on Competition Policy, Antitrust, and Consumer Protection in 2023. Francis, a former deputy director in the Federal Trade Commission’s (FTC) Bureau of Competition, also advocated antitrust reform and, not incidentally, increased funding for the federal antitrust-enforcement agencies. At the same time, like Hovenkamp, he was crystal clear that he did not “recommend enacting AICOA.” 

His detailed critique—more than 80 pages of his testimony focused on AICOA—covered ground that should be familiar to readers of Truth on the Market. For example, sections of his testimony carried subheadings identifying several problems with AICOA’s approach to self-preferencing: 

‘Self-Preferencing’ Includes Many Desirable Practices

Banning Self-Preferencing Would Inflict Consumer Harms

A Ban Would Deter Product Improvements

A Ban Would Deter Platforms from Protecting Consumers

A Ban Would Challenge Some Free-to-Use, Ad-Supported Services

A Ban Would Threaten Closed Ecosystems

Francis also argued that AICOA’s scope appears arbitrary. And he raised concerns about the bill’s limits on data use, its access and interoperability requirements, and its “no-conditioning rule.” On the latter, he noted that the “provision amounts to a per se rule against technological product tying of a kind that modern antitrust has long—and very wisely—left behind.” 

AICOA’s Vertical Leap of Faith

Part of what’s striking about the new AICOA is how many deeply flawed assumptions it retains from versions that failed to gain traction in earlier Congresses. Part of what’s also striking is how arbitrary some of its departures from the alpha and beta test versions appear to be. 

For example, the new AICOA’s general hostility to vertical integration, self-preferencing, and default settings is not new. Neither are its interoperability and data-portability requirements. Interoperability means designing a system so that it can interoperate–at the least, exchange data–with other systems; data portability means enabling users to move “their” data from one service or platform  to another. Both can be useful. But the decades-long push to encourage, and now require, interoperability in electronic medical systems is an object lesson in potentially confounding factors, even in a context where there are clear policy benefits to interoperability and there is public funding behind its development. Mandatory interoperability and data portability restrictions can prove costly, risky, or counterproductive, as can other data restrictions contemplated by AICOA. They do not even guarantee the effective flow of information, much less competitive benefits.   

Each of these issues could warrant a much longer critique. For more—much more—on self-preferencing, I’ll simply steer readers to ICLE’s Issue Spotlight on “The Case for Self-Preferencing,” with work from ICLE scholars, academic affiliates, and others.  

As a related matter, we’ve had quite a bit to say about vertical integration, but the basic point starts with the Supreme Court’s stepwise recognition—from Continental T.V. Inc. v. GTE Sylvania Inc. in 1977, which rejected per se liability for vertical nonprice restraints, through Leegin Creative Leather Products Inc. v. PSKS Inc. in 2007, which rejected per se liability for vertical price restraints—that vertical integration is not generally anticompetitive, even if it can prove anticompetitive under particular facts and circumstances. 

Directionally, at least, that trend in antitrust law has followed the economic literature. For example, James C. Cooper, Luke M. Froeb, Dan O’Brien, and Michael G. Vita reviewed the theoretical and empirical literature on vertical integration and vertical restraints and found “a paucity of support for the proposition that vertical restraints/vertical integration are likely to harm consumers.” They recognize that harm is possible—that is, the welfare effects of vertical integration are theoretically ambiguous—and argue that empirical evidence is therefore critical to sound competition policy. Their review of that empirical literature “suggests that vertical restraints are likely to be benign or welfare enhancing.” 

Similarly, Francine Lafontaine and Margaret Slade reviewed the literature on exclusive contracts and vertical restraints and found that, when firms adopt vertical restraints, they typically improve product quality and service, benefiting consumers as well as producers. 

The point is not that vertical restraints should be per se lawful. They remain subject to antitrust scrutiny under the rule of reason—the fact-specific legal test courts typically use to weigh competitive harms against benefits. Rather, as Cooper et al. observe, novel legal presumptions against vertical restraints are likely to produce too many false positives: condemning conduct that is actually benign or beneficial. Nothing in the literature, or in the “fact sheet” accompanying AICOA, suggests that the bill’s vertical restrictions will fare any better. 

What’s New? The Target—and the Hammer

So what’s at least somewhat new? Start with the target: Which firms would be subject to the AICOA’s restrictions?

The bill’s reach has been somewhat redrawn, although AICOA 1.2 retains the big-tech-is-bad approach of versions 1.0 and 1.1. It keeps the DMA-style “gatekeeper” model Hovenkamp criticized in assessing AICOA 1.0, along with the scope restrictions Francis found arbitrary in his testimony on AICOA 1.1. 

The new target is any “systemically important platform,” along with the firms that own or operate such platforms—or that hold, control, or benefit from at least a 25% share of them. It’s still Big Tech, but with a somewhat different metric for “big.” 

Under the bill, “big” turns partly on the scale of the parent or “operator”—the firm that owns or controls the online platform. AICOA targets firms with “average annual gross revenues of not less than $175,000,000,000,” as adjusted. It also turns on the platform’s popularity: “monthly active users in the United States equal to not less than 34 percent of the population of the United States over the age of 12, as determined by the most recent decennial census of population conducted by the Bureau of the Census.” Alternatively, the bill reaches platforms with monthly subscriptions held by the same percentage of U.S. households. 

To be sure, the revenue number is large. So is the monthly-active-user threshold. The most recent decennial census was conducted in 2020. It found roughly 280 million people over age 12. Multiply that by 0.34, and the threshold comes out to about 95.2 million monthly users. 

That is not just the GAFAM firms—Google, Apple, Facebook, Amazon, and Microsoft—and their platforms. Walmart and others would seem to fit the bill. Still, AICOA would reach a relatively small number of firms that operate online platforms. 

But why gross revenue? More broadly, why these large firms and these restrictions? Once again, there is no good answer—or even an intelligible one. Certainly, and consistent with earlier versions of AICOA, there is no market-power or monopoly-power requirement for any product or service market. 

Another new wrinkle is the bill’s civil-penalty provision: 

to be deposited in the Treasury of the United States, in an amount not greater than 10 percent, and not less than 1 percent, of the total United States revenue of the person for the period during which the violation occurred. 

That is 10% of revenue, not any measure of profits. And it is the revenue of the parent firm, not the “systemically important platform” itself. The 1% floor is still revenue, and it is still the parent firm’s revenue. As Manne says, that is “a new floor for enormous fines, even in close cases involving unsettled legal questions.” 

More than that, the bill contemplates civil penalties wholly untethered from any assessment of the harm supposedly caused by the prohibited conduct. Out the window goes the possibility of efficient remedies—that is, remedies that force firms to internalize the harms they cause and thereby give them incentives to avoid causing such harms in the future. 

Instead, we get runaway penalties that might be appropriate for unequivocally harmful conduct, where courts and enforcers have little concern about overdeterrence or penalties that ultimately do more harm than good to competition and consumers. That is not the world we are in here. None of this is about unequivocally harmful conduct. Compare and contrast, for example, Steven Shavell with Louis Kaplow and Robert Cooter. 

What of Manne’s twin complaints that AICOA “lowers the competition-harm threshold to anything above de minimis,” while increasing “the burden for key safety, privacy, and security defenses”? The first objection is lifted straight from the bill’s text. Key provisions—such as the self-preferencing provision—address conduct that “would materially harm competition.” And the definitions section sharpens the point: Under AICOA, “materially harms competition” means “any actual or reasonable risk of lessening competition or impairing the competitive process that is more than a de minimis amount.” 

Couple that very low bar with AICOA’s burdens on defendants seeking to establish an affirmative defense after the government pleads any risk of anything more than de minimis harm. For the prohibitions on “preferencing, limiting, and discrimination,” a defendant must show, by “clear and convincing evidence,” that the conduct was necessary either to comply with federal or state law or to “protect safety, user privacy, the security of nonpublic data or of the platform, or any other significant cybersecurity risk, or to prevent fraud or spam.” The defendant must also show that the conduct was “narrowly tailored in scope” and “could not be achieved through less anticompetitive means.” 

How are we to read “less anticompetitive means” against a standard triggered by any risk of anything more than a de minimis amount of harm? 

What’s more, any such affirmative defense would rely on the defendant’s production of ordinary-course documents that:

i. describe the specific purpose for which the conduct was undertaken; and

ii. identify the material risks or harms the conduct was intended to address.

For other prohibitions in the bill, the defendant would have to establish, “by a preponderance of the evidence that the conduct has not materially harmed and would not materially harm competition.” But again, material harm to competition means “any actual or reasonable risk of lessening competition or impairing the competitive process that is more than a de minimis amount.” 

These are not minor adjustments to the rule-of-reason scrutiny typically applied to allegedly anticompetitive restraints, and especially to vertical restraints alleged to violate Section 2 of the Sherman Act, at least since the Supreme Court’s 1977 decision in Continental T.V. Inc. v. GTE Sylvania Inc. 

Rather, for conduct likely to be procompetitive or benign, AICOA presumes illegality—setting the plaintiff’s burden very close to ground level. It then shifts the burden to the defendant, where the bill’s sponsors seem to have in mind “ain’t no mountain high enough”—a ’60s classic, to be sure, but not a theory of harm to competition or consumers. 

Third Time’s No Charm

The sponsors of AICOA 1.2 seem to have made at least a passing attempt to revise the details of versions 1.0 and 1.1. What drove those editorial choices is anyone’s guess. The key faults remain, and where the bill does change, it is hard to see a net improvement. 

This version flunks the “no economic sense” test, just like the last one, and just like the one before that. Herbert Hovenkamp said that “AICOA [1.0] was a bill that deserved to die.” That assessment was strict but fair. Entirely fair. 

So, with apologies to Sens. Klobuchar and Grassley: This one, too, deserves to die. Third time’s no charm.

The post AICOA Rises from the Grave, Still Looking for a Theory of Harm appeared first on Truth on the Market.

Last month, Brazil’s antitrust authority, the Administrative Council for Economic Defense (CADE), entered that debate. It issued part of its long-awaited decisions on a batch of AI partnership agreements and “acqui-hire” deals between Big Tech firms and AI startups. The practical result was modest. The doctrinal signal was not. 

CADE dismissed three cases: NVIDIA/Run, Microsoft/Mistral, and Google/Character.AI. It ordered ex post notification in one case, Microsoft/Inflection, and opened two new administrative proceedings involving Google/Windsurf and Google/Hume AI. One final case, Amazon/Anthropic, had been scheduled for decision at the same May Tribunal session, but Commissioner-Rapporteur José Levi Mello do Amaral Júnior withdrew it from the agenda, citing “notorious facts” that warranted further investigation. 

The headline result was prudent. Three cases escaped formal notification, and the one that did not—Microsoft/Inflection—had already been cleared on the merits by the United Kingdom’s Competition and Markets Authority (CMA). But beneath that restrained bottom line, the Tribunal opened two doctrinal doors that deserve careful scrutiny. 

The first is substantive. CADE held that “reverse acqui-hires”—bundles of nonexclusive licensing, key-team hiring, and substantial payments—can qualify as “concentration acts” under Article 90, II of Brazil’s Competition Law, even when no shares or direct assets change hands. In plain English, CADE signaled that a deal need not look like a conventional merger to be treated like one. 

The second is procedural. CADE invoked its “call-in” power under Article 88, §7º to require ex post notification of a below-threshold deal, even though the agency still lacks clear, predictable criteria for when it will exercise that exceptional authority. 

Both moves echo the July 2024 joint statement by the U.S. Justice Department (DOJ), Federal Trade Commission (FTC), European Commission, and the CMA on competition in generative-AI foundation models and AI products. CADE has, in effect, decided to surf the same wave. 

This post explains how it got there. It starts with the basics of Brazil’s merger-notification regime, then examines how the Tribunal characterized “acqui-hire” and “reverse acqui-hire” arrangements as “concentration acts.” It then asks what distinguished Microsoft/Inflection from Google/Character.AI on the “call-in” question, summarizes the outcome of each case, and closes with a substantive critique of the killer-acquisition theory CADE has imported. 

The question, in the end, is whether Brazil’s Competition Law has the doctrinal board to ride that wave safely. 

The Merger Review Before the Merger Review

CADE opened the five APAC proceedings—short for Apuração de Ato de Concentração, or investigation of a concentration act—in the second half of 2024, shortly after the joint statement from the four other leading enforcers. 

A brief procedural note helps. APACs are governed by CADE Resolution 24/2019. They are used mainly in gun-jumping cases—that is, cases involving transactions that should have been notified before closing but were not—and to determine, at a threshold level, whether a transaction qualifies as a “merger” under Brazil’s Competition Law. 

An APAC does not involve a substantive competition analysis of the transaction. At the end of an APAC, CADE’s Tribunal may order formal notification of the merger, which triggers the Ato de Concentração (AC) procedure. That is the proper vehicle for substantive legal and economic review. Under the AC procedure, CADE’s General Superintendence conducts the competition analysis and, depending on the case, refers it to the Tribunal for a final decision. 

That means none of these AI-related proceedings has yet received a meaningful substantive competition review. The reason is simple: an APAC is not built for that job. 

Instead, CADE’s Tribunal faced three preliminary questions. First, does each transaction qualify as a “merger” under Brazil’s Competition Law? Second, if so, does it meet Brazil’s mandatory notification thresholds? Third, if it does not meet those thresholds, should CADE nonetheless require formal notification under its “call-in” power?

Even without a full merger analysis, the APAC proceedings surfaced a long list of questions and hypothetical concerns worth examining. They also share one central legal fact: CADE opened them ex officio—on its own initiative—because none of the parties self-notified. And none self-notified because none of the deals met Brazil’s mandatory filing thresholds. 

The Thresholds Are the Point

Before turning to the specific cases, it is worth pausing over Brazil’s merger-notification system under Brazilian Competition Law (BCL) Law 12.529/2011. 

Under Article 88 of the BCL, a transaction must be notified to CADE before closing only if it first qualifies as a “concentration act” under Article 90. That category includes mergers, acquisitions of control or parts of companies—including tangible or intangible assets—and the contested category of “associative contracts,” which mainly covers joint ventures and similar arrangements.

Article 90’s concept of a “concentration act” is broad by design. Congress did not want merger control to be trapped by the narrower labels of business or contract law. Still, the category does not cover ordinary day-to-day commercial contracts between firms. A supply agreement is not a merger merely because the parties shook hands with gusto. 

Nor is classification as a “concentration act” enough to trigger CADE review. The transaction also must meet the revenue thresholds in Article 88. Mandatory notification arises only when both conditions are met: one party’s Brazilian group—typically the acquirer—recorded revenue of at least 750 million Brazilian reais (about $144 million) in the year before the transaction; and the other party’s group—typically the target—recorded revenue of at least 75 million Brazilian reais (about $14.4 million). If those thresholds are not met, the transaction is not subject to mandatory notification. 

The thresholds are jurisdictional. Without them—and absent a “call-in” decision—CADE has no authority to review the transaction at all. In that respect, Brazil is closer to the European Union Merger Regulation and its jurisdictional rules than to the U.S. Hart-Scott-Rodino Act regime, which is reporting-based rather than jurisdiction-conferring. 

Article 88, §7º creates the exception: CADE’s so-called “call-in” power. The statute provides that, “within one year from the date of consummation,” CADE may require notification of a transaction that otherwise falls below the thresholds. CADE has used that power sparingly. One study covering 2011—when the current BCL took effect—through 2023 found only seven uses. For perspective, CADE’s public dataset records at least 4,358 merger filings from 2015 through 2023 alone. Even that understates the full picture, because it excludes 2011 through 2014. 

Until now, the call-in power has been understood as a narrow corrective, reserved for deals whose economic significance slips through the revenue test. The five AI cases test just how far that correction can stretch before it becomes something else. 

In each of the four AI deals decided at the Tribunal’s May session, CADE’s General Superintendence issued its opinion in March of this year. It took nearly two years for the Superintendence to conclude that none of the transactions was subject to mandatory notification under Brazil’s merger-control regime. The acquirers—NVIDIA, Microsoft, and Google—comfortably cleared the 750 million Brazilian reais threshold. The targets—Run, Mistral, Character.AI, and Inflection—generated no meaningful revenue in Brazil. 

That resolved the second question identified above: The transactions did not meet the mandatory notification thresholds. The Tribunal then turned to the remaining two questions. First, do these transactions qualify as “mergers” under Article 90 of the BCL? Second, as the Article 88, §7º inquiry was reframed, should the targets’ lack of Brazilian turnover be overridden because AI startups may have strategic value?

Those are interesting questions. But especially on CADE’s call-in power, they raise a more basic one: Are these the right cases in which to answer them? And is this the best use of CADE’s resources when no agency has found evidence of actual harm to competition in these deals? 

The Amazon/Anthropic case adds another wrinkle. As noted above, CADE postponed that case for further investigation because of “notorious facts” that were not disclosed publicly during the Tribunal session. Why? Which facts or conditions justify different treatment? 

Other agencies have not found concrete competitive harm from the Amazon/Anthropic partnership. The FTC and the CMA both examined the deal. The CMA cleared it outright. The FTC’s report—conducted under the Biden administration and approved 5-0—identified potential risks, including lock-in, input foreclosure, and information asymmetry, but stopped short of enforcement action. 

When a Hiring Spree Starts Looking Like a Merger

In short, CADE’s Tribunal concluded that acqui-hires and reverse acqui-hires can qualify as mergers under Article 90 of the Brazilian Competition Law and may therefore require mandatory notification when Article 88’s revenue thresholds are met. That conclusion is significant. But the reasoning behind it—and the factual elements needed to turn a bundle of contracts into a “concentration act”—remains somewhat murky. 

As Selcukhan Ünekbas has explained, “acquihires are transactions aimed primarily at acquiring a firm’s workforce, rather than its products or other assets. They differ from ‘license-and-hire’ agreements, in which the acquirer also licenses the target’s technology.”

The controversy is not new. It traces at least to a paper by John Coyle and Gregg Polsky published more than a decade ago. So the debate over how Big Tech firms acquire, absorb, or otherwise siphon talent from startups—and what that means for competition law—has been around for a while, at least in the literature.

The newer label, or perhaps the newer lens, is the “reverse acqui-hire.” A traditional acqui-hire involves the direct acquisition of a small startup, with the workforce as the main prize. A reverse acqui-hire tries to reach much the same practical result without formally acquiring the company.

Reverse acqui-hires typically combine three sets of obligations, often spread across several contracts:

(1) some compensation paid to the team—conventionally structured to create incentives for the team to stay with the acquirer for a longer period of time; (2) a non-exclusive license to the startup’s intellectual property; and (3) some kind of waiver from the target company that it will not sue the acquirer for poaching the talent.

Unlike a traditional acqui-hire, the startup generally survives the deal: “In contrast to traditional acquisitions or acqui-hiring, RAH transactions neither involve equity transfer nor full buyout of technology nor full takeover of the people.” The startup usually reshapes its business afterward, but it remains a distinct entity, in some form, from its would-be acquirer. 

In Google/Character.AI, Commissioner Camila Alves examined the hallmark features of a reverse acqui-hire and how they bear on Article 90’s definition of a “concentration act.” In her view, the transaction should not be assessed as a pile of disconnected pieces, but as an integrated whole. Licensing intellectual property, hiring part of a technical team, terminating a prior investment, and making a substantial payment to the company might each fall outside Article 90 if viewed in isolation. But when bundled together as “part of a single economic package” (§ 59), CADE may treat the operation as a concentration act under Article 90, II of the BCL—either as the acquisition of “intangible assets” or as the acquisition of a “part of a company” (§ 72). 

Alves stressed that this is not a broad rule making every nonexclusive technology license or hiring decision notifiable. Rather, in her view, what matters is the coordinated structuring of those elements into a single negotiated package. Put less delicately: one contract may be a hire, another may be a license, and a third may be a payment. Together, CADE may see a merger wearing a fake mustache.

Commissioner José Levi reached a similar conclusion in Microsoft/Inflection. Microsoft argued that none of the deal’s individual elements—the hiring of Inflection co-founders Mustafa Suleyman and Karen Simonyan, the hiring of a substantial part of Inflection’s team, the licensing arrangements, and the related corporate contracts—amounted on its own to the acquisition of Inflection’s business or assets. José Levi agreed that the pieces should not be viewed in isolation. Taken together, he concluded, they fit Article 90, II.  

For José Levi, Microsoft/Inflection matched the reverse acqui-hire pattern described in the literature (§§ 46-51): leading technology firms effectively absorb a startup’s business activity by hiring its key personnel and compensating investors through substantial licensing payments, thereby producing the economic effect of a conventional acquisition outside the formal channels of corporate law. Because these arrangements replicate the economic logic of a conventional acquisition, CADE cannot decline jurisdiction merely because the transaction used an unconventional form. 

CADE therefore characterized Google/Character.AI and Microsoft/Inflection as reverse acqui-hire transactions based on the combination of elements in each deal, even though neither involved a direct acquisition of shares or assets. 

The other two cases differed. NVIDIA/Run was a straightforward full acquisition of Run by NVIDIA. Microsoft/Mistral involved a €15 million investment in instruments convertible into shares that, upon conversion, would represent 0.31% of Mistral’s equity. Under Articles 9 and 10 of CADE Resolution 33/2022, acquiring more than 5% of a company’s shares—combined with the Article 88 revenue thresholds discussed above—triggers mandatory notification. The acquisition of call options is treated as equivalent to a direct acquisition of shares for merger-control purposes. Those two cases therefore did not raise the acqui-hire questions that drove the analysis in Google/Character.AI and Microsoft/Inflection. 

To be sure, Article 90, II of the BCL is exceptionally broad, as noted above, and deliberately so. It provides: 

Art. 90. For the purposes of Article 88 of this Law, a concentration act shall be carried out when: […] II—one (1) or more companies acquire, directly or indirectly, by purchase or exchange of stocks, shares, bonds, or securities convertible into stocks or assets, whether tangible or intangible, by contract or by any other means or way, the control or parts of one or more companies.

Given that language, it is defensible to read reverse acqui-hire arrangements—combining personnel hiring, nonexclusive licensing, and monetary payments—as a single concentration act when those elements occur together, even though none would qualify on its own. 

The harder question is what, exactly, the contracts must contain before CADE can call the package an acqui-hire and, by extension, a concentration act. Is hiring one person enough, if that person is the chief executive officer? Must the acquirer hire part of the team, a substantial share of the team, or nearly everyone? What obligations must the nonexclusive licensing agreement include? Is an upfront payment necessary? If no compensation changes hands, can the transaction still be an acqui-hire? Must the contracts include a no-sue waiver by the target for talent poaching? And if one element is missing while the others are present, does the transaction still count? 

Those questions remain unresolved. That uncertainty is the predictable consequence of defining the concept by combining elements that, standing alone, would not trigger merger review. CADE’s Tribunal will need to refine these factors in future cases if companies are to have the legal certainty they need when deciding whether a transaction qualifies as a merger under Brazilian competition law. 

CADE Calls One In, Lets the Others Surf By

With the first question resolved (reverse acqui-hires can qualify as “concentration acts” under Article 90 of the Brazilian Competition Law) and the second settled as well (none of these AI partnership transactions triggered mandatory notification because they failed to meet Article 88’s revenue thresholds) the analysis turns to the third: CADE’s “call-in” power under Article 88, §7º. 

In the end, CADE required formal notification only in Microsoft/Inflection. The remaining cases were not called in. Why? 

One caveat: The analysis that follows focuses mainly on Microsoft/Inflection and Google/Character.AI, because the opinions in the other cases were not yet public at the time of writing. 

Article 88, §7º does not give CADE a roving license to second-guess every transaction—or every “concentration act”—that falls below the mandatory revenue thresholds. As both rapporteurs acknowledge, invoking the provision requires a juízo de conveniência e oportunidade: a discretionary assessment of whether a call-in is convenient and opportune in the specific case. 

Commissioner Alves frames the requirement clearly in Google/Character.AI: 

The provision (call-in power) functions as an exceptional mechanism for correcting the mandatory notification system. The revenue criteria confer objectivity, predictability, and administrative rationality on merger control, but they do not exhaust all hypotheses of possible competitive relevance. […] This possibility does not authorize the conversion of Article 88, §7º into an ordinary instrument for broad review of every non-notifiable operation. Its application requires caution, specific reasoning, a demonstration of competitive plausibility, and observance of the legal limits. It is an exceptional jurisdiction to be exercised when the circumstances of the case indicate that the ordinary revenue criteria may not have adequately captured the economic relevance of the operation. (§§ 21-22)

Alves reiterates the point in her conclusion: Article 88, §7º must be preserved “as an exceptional instrument for correcting the revenue-based notification system, and not as a mechanism of ordinary use in any atypical operation” (§ 133). On the standard itself, Alves (§§ 21-22 and § 133) and Commissioner José Levi (§ 60) appear to agree: The call-in power is exceptional; the transaction must present a plausible competitive concern; and the Tribunal must show that the ordinary thresholds failed to capture the deal’s economic significance. 

So what made Microsoft/Inflection convenient and opportune in a way that Google/Character.AI was not? Reading the two opinions side by side, three distinguishing features emerge—though the line they draw is hardly neon-bright. 

The first is the intensity of the team transfer. Alves flags the distinction: “the precedent (the Microsoft/Inflection case analyzed by the CMA in 2024) is not identical. In Inflection, the team transfer was more intense, involving almost the entire team of the company; in Character.AI, Google itself reports that [REDACTED] employees were released, in addition to a non-exclusive license over certain technologies” (§ 68). 

José Levi makes a similar point in Microsoft/Inflection, relying on the CMA’s parallel findings: “from the data in the CMA’s decision on the same operation, it is possible to infer that the employees hired by Microsoft corresponded, in fact, to almost the entire team of Inflection” (§ 72). The redacted portion of Camila’s analysis suggests Google hired a meaningful subset—but not the whole team—of Character.AI. The scale of the team transfer therefore appears to have mattered to CADE’s decision to require formal notification in Microsoft/Inflection. 

The second distinction is what happened to the target’s product strategy after the deal. This is the point on which José Levi rests most heavily. Before the transaction, he explains, Inflection focused on developing foundation models—general-purpose AI models that can support many downstream applications—and used them as the basis for Pi, its consumer-facing chatbot. After the team responsible for growth in that segment departed, Inflection allegedly shifted its focus to customized generative AI models for corporate clients (§ 75). 

José Levi treats that pivot from business-to-consumer to business-to-business as “circumstantial evidence that the operation could have produced a reduction of rivalry in the segment of FM development and consumer-facing chatbots” (§ 76)—a segment where Microsoft operates through Copilot. Character.AI, by contrast, continued operating its consumer-facing chatbot platform after the Google deal. As Alves notes, it “preserved formal independence” and “continued its downstream application” (§ 140). 

The third distinction is horizontal overlap. José Levi’s emphasis on Microsoft Copilot and the consumer-facing foundation-models segment frames Microsoft/Inflection as a deal that may have reduced current rivalry, not merely future rivalry, in a market where the acquirer already competes (§ 77). Alves’ analysis of Google/Character.AI leans more heavily on potential competition and innovation trajectory—the idea that Character.AI might have become a more significant rival in the future (§§ 81-82, 138-139). Whether the Tribunal believes this distinction matters under Brazilian Competition Law remains unclear. 

Despite these differences, Alves is explicit that the dismissal of Google/Character.AI rests not only on the absence of competitive plausibility, but also on timing and proportionality. She writes:

I understand that it is neither convenient nor opportune to determine the notification of the operation. This conclusion does not derive exclusively from the time elapsed since consummation, although that element is relevant in the exceptional design of Article 88, §7º. It also weighs the convenience of an ex post requirement at this moment, the already advanced stage of implementation of the operation, the absence of consolidated administrative criteria for reverse acqui-hires, the proportionality of the measure, and the possibility of directing institutional action toward more recent analogous operations, in a specific and timely procedure. (§ 141).  

José Levi acknowledges similar factors in Microsoft/Inflection but reaches the opposite conclusion. He leans instead on the killer-acquisition and talent-hoarding literature (§§ 65-69), the intensity of the team transfer, and the alleged reduction of rivalry in the consumer-facing foundation-models segment. 

A skeptical reader is entitled to find the resulting line less than crystalline. Both deals were announced and consummated in 2024. Both involved nonexclusive licenses, substantial payments, and the hiring of part—or nearly all—of a target’s technical team. Both were treated as “concentration acts” under Article 90, II. And under Alves’ opinion, both were at least plausibly relevant from a competition standpoint. 

The different outcomes therefore seem to rest less on a clear legal rule than on a bundle of factual gradations and ex post implications: how many team members were hired, what happened next at the target, and how closely the target’s business overlapped with the acquirer’s existing products. 

That may be enough to decide these cases. It is less helpful for the next company trying to plan around CADE’s call-in power. 

Based on the decisions so far, it remains difficult to articulate a clear legal test for when CADE will require formal notification. For companies assessing whether their own license-and-hire arrangements may attract scrutiny, the emerging criteria remain opaque. CADE’s Tribunal will need to clarify them in future cases—and it will soon have the chance to do so in Google/Windsurf and Google/Hume AI. 

CADE Catches the AI Wave—Now Comes the Hard Part

After this brief tour of the cases, the outcomes were as follows. 

In NVIDIA/Run, Commissioner-Rapporteur Carlos Jacques Vieira concluded that Run does not operate in Brazil, that the transaction does not produce effects in Brazil, that the parties lack market power in the country, and that there was no harm—or risk of harm—to competition sufficient to justify invoking Article 88, §7º. The European Commission reviewed the same case in December 2024 and approved it, finding that it raised no serious doubts and would not harm consumers. Like the other cases discussed below, NVIDIA/Run was decided unanimously by the four current members of CADE’s Tribunal. 

In Microsoft/Mistral, Commissioner-Rapporteur José Levi found that Microsoft’s stake in Mistral fell below the 5% threshold, that the partnership agreement did not confer control, and that there was no evidence of competitive harm in the relevant market. He therefore dismissed the case and declined to require formal notification under Article 88, §7º. That outcome tracks the analysis and predictions advanced by Dirk Auer and Mario Zúñiga when the investigations were first opened. The CMA had reviewed the same transaction in May 2024 and similarly concluded that, given the deal’s specific features, “the CMA therefore does not have jurisdiction to review the Partnership in its current form.” 

In Google/Character.AI, Commissioner-Rapporteur Alves voted to dismiss but wrote a substantial opinion on AI partnerships and acqui-hiring agreements. As discussed above, Alves treated the combined deal—a nonexclusive technology license, the “release” of part of Character.AI’s technical team for Google to hire, the cancellation of a prior investment, and a payment of roughly $2.7 billion—as a “concentration act” under Article 90, II of the BCL. She nonetheless declined to require formal notification after weighing the costs and benefits of an ex post filing requirement. 

In the same decision, Alves ordered CADE’s General Superintendence to open two new APAC procedures involving the more recent Google/Windsurf and Google/Hume AI cases. CADE’s debate over whether acqui-hiring arrangements may qualify as concentration acts under Brazilian merger-control law is therefore very much alive. 

Finally, in Microsoft/Inflection, José Levi required formal notification. As discussed above, he treated the transaction as following the same basic template as Google/Character.AI: the hiring of nearly the startup’s entire team, combined with a nonexclusive license and a substantial upfront payment. He found that package to be a “concentration act” under Article 90, II of the BCL. Microsoft must now formally notify the transaction under Article 88, §7º within 30 days. 

CADE will therefore have to assess the transaction’s substantive effects, including the relevant market, the degree of market concentration, and the potential for unilateral and coordinated effects. It will also likely have to confront, at least to some extent, the “killer acquisition” theory of harm—the claim that an incumbent buys, absorbs, or neutralizes a nascent rival to prevent future competition. 

That is notable because CADE required formal notification even though the CMA had already cleared Microsoft/Inflection on the merits in September 2024, finding no realistic prospect of a substantial lessening of competition. Microsoft/Inflection will thus become the first AI partnership agreement to receive full review under CADE’s regular Ato de Concentração merger-control procedure. 

CADE’s first wave of AI partnership decisions produced a restrained practical outcome but an ambitious doctrinal framework. The Tribunal expanded the interpretation of concentration acts to encompass reverse acqui-hires and signaled a greater willingness to use its exceptional call-in power for transactions involving AI startups. 

Those uncertainties are not merely procedural. They reflect a deeper substantive debate now emerging internationally and increasingly shaping Brazilian competition law. The next question is whether the theories driving this enforcement agenda—especially the theory of digital killer acquisitions—rest on solid empirical ground. 

That question is the focus of the second article in this series. For now, CADE has paddled into the AI wave. The harder part is proving it can steer.

The post Brazil Catches the Acqui-Hire Wave appeared first on Truth on the Market.

To many observers, this looks wasteful at best and sinister at worst. Why would a profit-maximizing firm pay enormous sums for talent it seemingly has no intention of using?

Ronald Coase had a wry answer for moments like this. “[I]f an economist finds something—a business practice of one sort or other—that he does not understand,” he observed in 1972, “he looks for a monopoly explanation. And as in this field we are very ignorant, the number of ununderstandable practices tends to be rather large, and the reliance on a monopoly explanation, frequent.”

A new working paper by Shaolong Wu of Harvard Business School and Zefan Qian of Georgetown, “Talent Hoarding and Upstream Innovation: Labor Market Distortions by Large Incumbents,” supplies precisely that monopoly explanation, complete with a formal model and an empirical test. Large incumbents, the authors argue, sometimes hire and retain frontier researchers not to put them to work, but to keep rivals from doing so. Because top AI-research talent is scarce, every researcher a dominant firm keeps “on the bench” is one a challenger can’t hire. The incumbent protects its existing profits, and society loses the discoveries that researchers would have produced elsewhere.

The paper closes with policy recommendations to match: limits on garden leave, narrower noncompete agreements for publicly funded researchers, and institutional pressure to keep frontier talent “actively deployed.”

The policy audience is already primed for this argument. The Federal Trade Commission (FTC) has announced its intention to scrutinize acquihires—transactions structured around hiring a startup’s employees rather than acquiring the company outright—to ensure they aren’t used to evade merger review. Talent-centered theories of competitive harm are rapidly becoming the next front in the broader campaign against large technology firms. All the more reason to get the economics right.

It’s a clever paper, and perhaps a more careful one than many in this genre. Its headline claim, however, substantially outpaces its evidence. What the data actually show is that one group of software firms retained more skilled employees after the Supreme Court weakened their patent protections. Everything beyond that—the “idle benches,” the foreclosed rivals, the lost innovation, and the social harm—comes not from the data but from assumptions built into the model.

More importantly, nearly every one of those assumptions rules out, by construction, a far more ordinary explanation: that the same behavior reflects good management rather than anticompetitive conduct.

The Model Does the Hoarding for You

In the paper’s model, an incumbent firm earns profits from a legacy technology, while an entrant may develop a replacement. The key input into discovering that new technology is a fixed pool of researchers (an assumption we will return to) who possess knowledge of the incumbent’s technology. Hiring one of those researchers does double duty: It increases your own odds of a breakthrough while reducing your rival’s. There is, by assumption, no one else to hire. 

The model also imposes another crucial condition. A researcher generates innovation only when paired one-for-one with “innovation capital”—new equipment, computing power, organizational capacity, and similar inputs. A retained researcher who lacks that matching investment is “benched”: employed and paid, but contributing nothing to the only outcome the model treats as socially valuable—the discovery itself. The model does allow benched workers to perform ordinary legacy work—it simply assigns that work no social value.

An additional assumption drives the paper’s headline results. Those results emerge from what the authors call the “no self-replacement” case, in which a breakthrough is assumed to be worth no more to the incumbent than the legacy product it already sells. The incumbent, therefore, gains nothing from innovation itself. In such a world, where the breakthrough has no upside for the incumbent, the only reason to retain a researcher is to keep her away from a rival. 

Notably, retention doesn’t require leaving the researcher idle. Matching her with capital and putting her to work also reduces the entrant’s chances of success. Under the model’s assumptions, even productive employment can function as a form of exclusion. Whether she sits on a bench or works at a desk, the point is the same: she isn’t helping a competitor build the next generation of technology.

The paper thus assumes, quite literally, that an incumbent’s only reason for incurring the cost of retaining a knowledgeable researcher is to deny that researcher to a rival. Under those conditions, if a successful entrant would destroy the incumbent’s profits, the incumbent will rationally pay to keep researchers in-house even when it places no value on anything they might produce.

On the empirical side, the paper exploits the U.S. Supreme Court’s 2014 decision in Alice Corp. v. CLS Bank. The Court held that implementing an abstract idea on a generic computer is not, without more, patentable. The decision thus made broad software patents substantially harder to obtain and enforce. For firms whose competitive position depended on such patents, Alice increased the risk that rivals could imitate or displace them—what the paper terms “external displacement risk.” Talent hoarding, in turn, became a potentially more attractive way to protect an incumbent position.

The authors find that software firms holding pre-2014 software patents increased their net hiring after Alice by roughly 0.34 percentage points per month relative to software firms without patents—a not-insubstantial effect, about 63% of the sample average. 

Several additional findings are noteworthy:

• The increase came almost entirely from reduced employee departures rather than increased recruiting.
• The effect was concentrated among workers with advanced degrees.
• It was not accompanied by measurable growth in assets or other complementary capital.
• A companion analysis of worker résumés found that science-and-research employees at exposed firms advanced more slowly up the occupational ladder in the years following the decision.

In short, the study finds: more high-skilled workers retained, no corresponding increase in complementary capital, and slower career progression among the researchers who stayed. The authors interpret this pattern as evidence of their central hypothesis: the defensive retention of underutilized talent.

The Data Are Not the Dispute

To give the paper its due, the distinction between retaining a worker and deploying one is genuinely useful. Too much commentary on tech hiring blurs the difference. The research design is also serious: exposure is measured before the shock, the shock itself is a judicial decision no firm controlled, and the authors are unusually candid about their limitations. Most importantly, the central empirical finding appears real. Patent-exposed software firms reduced employee separations after Alice, and those reductions were concentrated among workers with knowledge likely to be valuable for innovation.

The dispute isn’t over the finding, but over its interpretation. The talent-hoarding story depends on a series of modeling choices, each of which rules out a competing explanation under which the same data reflect efficient, rather than anticompetitive, behavior.

Foreclosure by Construction

The paper’s most significant assumption is one most readers will never notice. It has two parts. First, the incumbent and the entrant are assigned identical discovery technology, meaning a researcher generates exactly the same probability of a breakthrough wherever she works. The incumbent firm itself contributes nothing distinctive, except already being there. Second, as noted above, in the headline case, the incumbent places no value on producing the breakthrough at all. 

Taken together, the model effectively assumes the incumbent is the worst possible home for research talent. It’s no better at research, and it’s uninterested in the results. The only remaining explanation for retention is blocking. 

The authors would fairly answer that this isn’t a claim about the world. They set the new technology’s value to zero on purpose, to strip out the incumbent’s offensive motive—wanting the breakthrough for itself—and study the defensive motive in isolation. Fair enough. Isolating one mechanism is what models are for.

But here, the model removes the efficient reason to retain talent before the welfare verdict is rendered. Then that verdict gets carried over to a world where incumbents usually do want the breakthrough, and where retention is often offensive as well as defensive. The conclusion isn’t “found” in the research; it’s built into the model. 

And that assumption elides a great deal. Firms aren’t equally good at converting researchers into discoveries. They differ enormously in management quality, in the data and infrastructure researchers can use, in their ability to select promising projects and kill bad ones, and in the colleagues a new hire works alongside. A large incumbent may get more out of a marginal researcher than a startup can, because she joins teams with experienced managers who have already solved a thousand deployment problems, and because she can work with proprietary data and installed systems no entrant possesses. 

In that model—the real world—scarce talent flowing toward incumbents isn’t foreclosure. It’s the market allocating a scarce input to its highest-valued use. Eye-popping compensation is the competitive price of that input, captured by the workers themselves.

The paper’s model rules out this more realistic dynamic, and the empirical work can’t restore it. A finding that exposed firms retained more researchers is equally consistent with “those firms denied rivals an input” and “those firms are where the input is worth the most.”

An economist might respond that, in the model, the incumbent’s willingness to pay exceeds the worker’s deployment value. That’s what makes it hoarding. While true in this context, that result follows from the equal-capability assumption. Relax that unrealistic assumption, and the wage premium may reflect a productivity difference rather than a blocking premium.

That means the paper’s welfare conclusion turns on a parameter it never measures.

The Salop Problem Returns

The claim that a dominant firm will rationally outspend any challenger to preserve its position is one of the oldest moves in the antitrust playbook. A recent version of it comes from Steven Salop, who argued—in a paper pointedly titled “Potential Competition and Antitrust Analysis: Monopoly Profits Exceed Duopoly Profits”—that acquisitions of potential or nascent competitors by dominant firms raise inherent anticompetitive concerns. Because keeping a monopoly is worth more than sharing a duopoly, the incumbent will always pay more to eliminate a threat than the threat is worth to anyone else. 

Wu and Qian have, in effect, transported this logic from the market for startups to the market for researchers. Here, the thing the incumbent supposedly overpays to control isn’t a nascent rival firm, but the researcher who might help create one. Same wine, different bottle.

The argument fails in the labor-market setting for many of the same reasons it fails in the acquisition context—reasons Dirk Auer, Brian Albrecht, Eric Fruits, Daniel Gilman, Lazar Radic, and I discussed in the International Center for Law & Economics’ (ICLE) comments on the FTC and U.S. Justice Department’s (DOJ) draft merger guidelines, and that Albrecht has explored at length here at Truth on the Market.

The first problem is arithmetic. Buying off a single potential entrant may be profitable when monopoly profits exceed duopoly profits. But once one challenger is paid to stand down, the next stands to enter as a duopolist, and the next after that, and so on. Each must be compensated at roughly the duopoly level, not some small fraction of it. With enough potential challengers, the cost of paying them all off exceeds the value of preserving the monopoly, and the strategy collapses. If we nonetheless observe the conduct, that suggests something other than monopoly maintenance may be at work.

The same arithmetic applies to talent. The paper’s model contains one incumbent, one entrant, and one closed pool of researchers, so cornering the input requires outbidding exactly one rival. The real frontier-AI labor market looks nothing like that. There are multiple deep-pocketed AI labs, a venture-capital ecosystem aggressively funding startups to compete for the same people, and a university pipeline continuously producing new talent.

To foreclose discovery through retention, an incumbent would have to outbid all of those rivals, for every pivotal researcher, indefinitely. The predictable result isn’t foreclosure, but an auction. And in that auction, the owners of the scarce input—the researchers themselves—capture much of the surplus. That looks less like market failure than competition doing exactly what it is supposed to do.

Two features of labor markets make the strategy even less durable than its acquisition counterpart. First, workers, unlike startups, can’t be bought outright. Because “human capital is inalienable,” a retained researcher stays only as long as she chooses to stay. Maintaining exclusivity is therefore perpetually expensive and contractually fragile. 

Second, when employees leave large firms, they disperse. They don’t march single-file to the one rival most capable of threatening their former employer. As Kevin Murphy has observed, departing workers spread across many employers, with only a small fraction joining any particular competitor and many leaving the industry altogether. The model’s premise that every researcher released by the incumbent flows directly to the one firm poised to destroy it bears little resemblance to how labor markets actually function.

The second problem is the one discussed above. Salop’s result requires the incumbent to be at least as capable as any challenger—an assumption he relegates to a footnote, while acknowledging that “monopoly profits are not always higher” when an entrant has lower costs or a better product.

As Dirk Auer, Sam Bowman, and I have written:

Although it is convenient in theoretical modeling to assume that similarly situated firms have equivalent capacities to realize profits, in reality firms vary greatly in their capabilities, and their investment and other business decisions are dependent on the firm’s managers’ expectations about their idiosyncratic abilities to recognize profit opportunities and take advantage of them—in short, they rest on the firm managers’ ability to be entrepreneurial.

Once that realistic possibility is admitted, differences in firms’ willingness to pay no longer demonstrate preemption. They may simply reflect differences in productivity.

Again, the empirical prediction that retention rises after Alice doesn’t depend on the equal-capability assumption. But the conclusion that retention is driven by exclusionary motives does. The empirical test therefore confirms the portion of the model that doesn’t require the assumption, while leaving untested the assumption that does most of the work in generating the paper’s policy recommendations.

No New Assets, No New Ideas?

The paper’s cleverest empirical move is to infer foreclosure from the combination of rising headcount and flat capital investment. At first glance, the logic seems straightforward: if firms were truly putting these researchers to productive use, we should see corresponding investment in the inputs they need.

But the argument rests on a strong assumption: that a researcher paired with anything less than a full new unit of “innovation capital” produces nothing. In software, of all industries, that is a peculiar picture of the production process.

The complementary inputs that matter—data, codebases, proprietary tools, accumulated organizational knowledge, and much of the underlying computing infrastructure—are often already in place. An additional researcher can use those resources without the firm having to book a single new asset. Much of this capital is also non-rival within the firm. Another engineer working with the company’s data doesn’t consume that data. Even rivalrous inputs, such as computing power, are often available within existing capacity, with enough slack to absorb another researcher without generating a measurable increase in asset purchases.

A firm whose competitive advantage consists precisely of these intangible assets can productively absorb additional researchers with no observable capital response at all. Put differently, the paper’s headline empirical finding—more labor, flat balance-sheet assets—is exactly what we would expect if incumbents are where marginal researchers are most productive. 

The authors also argue that complementary capital “can often be scaled within just days and weeks,” making its absence particularly informative. If firms wanted to deploy these researchers, they argue, they could have done so cheaply and quickly.

But that cuts both ways. If complementary capital is as inexpensive and scalable as the authors suggest, then the absence of a detectable increase in assets is weak evidence of non-deployment. Cheap capital may not show up as a measurable balance-sheet change even when researchers are fully deployed. And if computing capacity can be expanded on short notice at low cost, then computing power was never the binding constraint. Talent was.

In that case, scarce talent flowing toward the firms willing to pay the most for it is not evidence of benching. It is what efficient allocation looks like.

When the same evidence supports opposite welfare conclusions, it ceases to be especially informative.

Nor does the authors’ fallback diagnostic—that patent output per employee fell at exposed firms—break the tie. That claim is perilously close to circular. The Supreme Court had just made it more difficult for those firms to obtain patents. Of course patenting declined. That was the treatment. It is not evidence that researchers were working less effectively or producing less valuable output.

The People Are the Patents Now

The most natural reading of the paper’s central finding requires no “benches” or anticompetitive animus. Alice didn’t conjure displacement risk from nowhere; it weakened the specific legal instrument—patents—that exposed firms had used to protect their innovations. Economists have long understood patents and trade secrecy as substitute ways to appropriate the returns to invention. And trade secrets don’t live in filing cabinets—they live in employees’ heads.

When the Court devalued these firms’ patents, the rational response was to lean harder on the other instrument: retaining the people who embody proprietary knowledge.

That reading fits the paper’s evidence rather neatly. It predicts retention rather than recruitment, because the point is to protect knowledge the firm already has, not acquire more of it. It predicts concentration among advanced-degree workers, because they are most likely to carry the relevant know-how. And it predicts no corresponding capital expansion, because nothing about the firm’s investment program or resource allocation needs to change. 

The paper’s own marquee example points in exactly this direction. Describing Adobe—the lead illustration—the authors write that, once Alice weakened its patents, “the strategic asset that mattered more was not just the code or the patent portfolio, but the people who embodied the know-how.” That’s a description of a firm switching from one appropriability mechanism (patents) to another (retaining employees who hold trade secrets). The paper states the benign reading in its own words, in its own flagship example, and then subsumes it under “foreclosure.”

But those are not the same thing. Preventing your own know-how from walking out the door to a competitor is a legitimate interest that trade-secret law has protected for well over a century. On the most plausible reading, it’s also welfare-enhancing here: retention substitutes for the appropriability that Alice weakened, preserving at least some of the innovation incentives the decision would otherwise have eroded. 

Indeed, the paper’s own framing concedes the point without quite noticing it. The workers in the model are valuable to the entrant because of their “incumbent-specific knowledge.” What the rival wants from them, in other words, is largely the incumbent’s own proprietary information.

The Missing Foreclosed Rival

Even granting the retention findings, the paper’s welfare conclusion—that society loses when incumbents retain these workers—is built into the model rather than derived from the evidence. The social planner against whom the incumbent is judged is defined to value “the availability of the innovation” while placing no weight on the incumbent’s existing profits. By assumption, entrant innovation is socially valuable. 

But a challenger’s incentive to enter often includes what economists call “business stealing.” Much of the entrant’s prospective profit comes from taking customers and profits from the incumbent. From a social perspective, that is largely a transfer rather than a gain. Indeed, the economic literature on entry has long recognized that private incentives to enter can sometimes exceed the socially optimal level for precisely this reason. 

One could just as easily write down a model with the opposite welfare weights and conclude that the entrant’s poaching is the problem. Neither stipulation would constitute evidence.

Meanwhile, the foreclosure side of the story—the part that carries the antitrust implications—is never tested at all. Foreclosure requires, well, foreclosure: rivals must be meaningfully constrained in their ability to compete because they can’t access a necessary input. Yet the paper offers no evidence that startups in Alice-affected fields were starved of talent, hired fewer workers, grew more slowly, or innovated less than they otherwise would have. Perhaps they did, but the paper neither tests nor demonstrates it.

What evidence we do have points in the opposite direction. The authors themselves note that the cost of software experimentation was falling during this period, as cloud computing and related technologies made entry cheaper and more accessible. 

Readers familiar with the killer-acquisitions debate will recognize the problem. Even in pharmaceuticals—the sector with the strongest evidence for the theory—the most generous estimates suggest that genuine “killer” acquisitions account for only about 5% to 7% of deals. The evidence in digital markets is considerably weaker—which is to say, nonexistent.

A foreclosure mechanism that leaves no observable trace in the supposedly foreclosed market may be an interesting theoretical possibility. It is not an empirical finding.

A Raise Is Not a Market Failure

The paper’s career-trajectory evidence is also clever, but again curiously interpreted. The slowdown in advancement for workers in the “science and research” category at exposed firms is real but tiny—about 0.018 points on a 2-to-5 occupational scale. Indeed, the scientists who are the paper’s central concern show the smallest of the highlighted effects. 

The largest decline in the paper’s table of “strong negative effects” belongs instead to workers in “education and training,” at a rate of ?0.068—nearly four times the science-and-research estimate. That is difficult to square with a story about the “benching” of frontier talent. 

The next-largest slowdown, among workers in “law, compliance, and public safety,” is easier to explain. The authors themselves attribute it to a collapse in demand for patent-related legal work—a direct consequence of weaker patents, not talent hoarding. And in a separate “difficult to interpret” category, “healthcare and maintenance” workers at exposed firms also show larger slowdowns than science-and-research employees. 

The heterogeneous responses across job categories are supposed to support the hoarding story. But doing so requires emphasizing the category with the smallest effect while explaining away the larger effects the theory can’t account for.

There is also a simpler, and perhaps more important, problem: the data contain job titles but not wages. A worker who stays in the same role because her employer pays her substantially more hasn’t suffered a policy-relevant harm. She has been compensated. Indeed, the very study the authors cite on inventors moving to large firms finds that earnings rise by double digits after those moves.

A flatter title progression accompanied by a fatter paycheck isn’t a labor-market distortion. It’s known as a raise.

Before We Ban Good Management 

If we step back and ask what conduct this paper actually condemns, the normative implications become difficult to sustain. What the paper identifies is lower employee turnover, the payment of retention premiums, and efforts to keep proprietary knowledge from reaching competitors. That describes competent management far more readily than monopolization. Every well-run firm in every industry does these things. The conduct labeled anticompetitive is observationally indistinguishable from conduct we generally want firms to undertake. And the paper’s empirical design can’t tell the difference, because the assumptions that do the distinguishing work—equal firm capability, one-for-one capital matching, and a social planner indifferent to incumbent investment incentives—are imposed rather than tested.

The paper’s reception has outpaced its findings. One economist described it as “a dark theory (with evidence)” of Big Tech hiring, in which incumbents “leave some on the bench” to “slow rival innovation.” But the bench is precisely what the paper never observes. The authors are candid that they can’t see deployment worker by worker. Benching is inferred from the absence of a capital increase, not directly measured. “With evidence” is doing a great deal of work for a mechanism the study assumes rather than demonstrates.

Despite that, the paper advances policy recommendations: limits on garden leave, restrictions on noncompete agreements for publicly funded researchers, and institutional expectations that frontier talent remain “actively deployed.” These proposals arrive amid a broader labor-antitrust push whose evidentiary foundations are already remarkably thin, resting on a small number of studies with heavily qualified findings. 

Yet the mechanisms the paper would restrict perform real economic functions. Even the FTC, in the rulemaking that banned noncompetes, acknowledged evidence that such agreements can increase investment in workers’ human capital, physical capital, and research and development. Firms invest in training and entrust employees with valuable knowledge when they have some confidence that those investments won’t immediately walk across the street. 

Weakening those arrangements based on a possibility theorem and an empirical pattern that admits several efficient explanations is a recipe for getting error costs backward. It risks condemning ordinary retention practices in the hope of catching the occasional genuine “idle bench.”

It is also worth remembering which firms the talent-hoarding story is meant to indict. The paper takes its empirical cue from a 2014 patent decision, but its stakes—and its policy recommendations—are aimed squarely at today’s AI incumbents. Those firms are not behaving like hoarders stashing talent on idle benches. From 2019 to 2025, the largest tech firms—Amazon, Alphabet, Meta, Microsoft, and Apple—added nearly 1 million employees while increasing capital expenditures from $77 billion to $370 billion, according to their 10-K filings. That is not a picture of firms accumulating talent they have no intention of equipping. It is a picture of firms racing to deploy labor and capital as quickly as they can acquire them. In other words, it looks like competition, not foreclosure. 

The paper’s actual contribution is narrower, but still interesting—if perhaps not as headline-grabbing. It suggests that when courts weaken patent protection, firms respond by retaining the people who embody their proprietary knowledge. That’s a finding about the substitutability of appropriability mechanisms, and it counsels caution both about Alice-style doctrinal shocks and about labor-market interventions that would weaken the substitutes firms turn to in response.

Before talent retention becomes the next frontier of antitrust, we should demand evidence of the thing that matters: not that firms kept their researchers, but that those researchers sat idle, and that someone else would have put them to better use.

The post Gotta Catch ’Em All? Antitrust and the AI Talent Wars appeared first on Truth on the Market.

On April 28, 2026, the European Commission published the review required under Article 53 and declared the DMA “fit for purpose.” The report credits the law with “a tangible positive impact” and sees no need to revise the list of core platform services. In the Commission’s telling, the machinery is working; where the evidence remains thin, time will supposedly fill the gaps. 

I do not doubt the Commission’s sincerity. I doubt that the exercise could ever have produced a different result. 

The DMA’s evaluative architecture is designed in a way that makes failure effectively impossible to demonstrate. Not because the Commission is uniquely stubborn or populated by self-serving sycophants. Rather, the regime’s foundational premises determine what counts as evidence in the first place. In a  new white paper, I describe this phenomenon as autopoietic regulation: a system that reproduces its own assumptions rather than testing them. The DMA’s first review confirms that diagnosis precisely because it passes. 

The Review That Couldn’t Fail

Start with a simple question: What finding would have led the Commission to conclude that the DMA, or some core part of it, was a mistake? 

Not “needs more time.” Not “compliance is incomplete.” A genuinely negative verdict: that an intervention’s costs exceeded its benefits; that a prohibition targeted conduct that was making consumers better off; or that the regulation’s underlying premise was wrong. Any review worthy of the name should be able to answer that question. An evaluation that can return only “working” or “working, but not yet finished” is not really an evaluation at all. 

The DMA’s review cannot return such a verdict, and I argue that the reason is structural. The regulation’s recitals present its obligations as simultaneously good for end users, business users, and innovation. In the DMA’s self-conception, there are no tradeoffs (except gatekeepers’ losses, which, as I have argued, are the point). 

No recital seriously entertains the possibility that mandated interoperability could reduce quality, that a prescribed fee structure could distort investment incentives, or that redistributing rents from platforms to business users could leave consumers—or other business users—worse off. If the statute does not acknowledge such costs, the review designed to assess the statute has no category in which to record them. 

So watch what happens when those costs show up anyway. 

When Bad News Doesn’t Count

The review did not lack for adverse evidence. The accompanying staff working document notes that advertisers, despite gaining new data-access rights, still face “persistent issues in obtaining sufficiently granular and comparable data.” It reports that alternative app stores and payment methods have produced “mixed results,” hampered by technical and contractual barriers. Gatekeepers and others also warned that interoperability mandates carry high compliance costs and “can harm customers, stifle innovation, and raise security and privacy concerns.” 

In an ordinary oversight regime, observations like these would mark the beginning of an inquiry. Has the intervention made its intended beneficiaries better off? Are the benefits worth the costs? Under the DMA, they function largely as consultation inputs: recorded, acknowledged, and then set aside. 

The recurring move is to treat disappointing results not as evidence that an obligation may be misconceived, but as evidence that it has not yet had time to work. App-distribution rules underperforming? Too soon to judge. Messaging interoperability barely adopted? Too soon to draw conclusions. The asymmetry is striking: success counts immediately, while failure is always provisional. 

The clearest illustration comes from the one company that does not fit the DMA’s template. Amazon—the only retailer designated under a framework built largely for search engines, app stores, and operating systems—used the review’s publication to argue publicly that the assessment “fails to address the fundamental design flaws that make the DMA unfit for retail,” and that compliance has “stalled investment, damaged the customer experience, and not made services more affordable.” 

One need not take Amazon’s word for any of this. The point is what the review does with such claims. It records them as stakeholder grievances. The possibility that a 40%-50% decline in a retailer’s organic traffic might reflect a defect in the regulation, rather than transitional friction on the way to a better equilibrium, lies outside the frame the Commission brought to the exercise. 

A Machine That Eats the Evidence

Here is where the autopoiesis point—big word, I know—connects to the architecture of the Digital Markets Act (DMA), and to two ideas I develop at length in the paper: the structural-unfairness axiom and the conflation of public and private gain. 

The DMA does not ask whether a gatekeeper’s conduct harms competition to the detriment of consumers. It begins from the axiom that designated markets are structurally unfair—that gatekeepers, by virtue of their position, “obtain a disproportionate advantage.” Unfairness is not something the Commission must demonstrate case by case. It is a premise the regime carries in. 

That kind of premise is self-sealing. Whenever an asymmetry between gatekeepers and business users persists—whenever rivals have not gained share, or users have not migrated—the axiom supplies a ready explanation: The unfairness has not yet been corrected. There is no result the premise cannot absorb, because the premise was never stated in a form that data could contradict. As I argued in the companion post on consumer choice, stubborn loyalty to a gatekeeper’s product is read not as a verdict to respect, but as a wall yet to be torn down. 

The second idea explains why the review hears the costs it does and misses the ones it does not. The DMA equates business-user grievances with the public interest. Its objectives are framed around benefits to business users—overcoming barriers to entry and redressing the “imbalance” between gatekeepers and their business users—and its enforcement metrics ask whether those imbalances persist.

That is a structural choice. It erases a distinction competition law has worked hard to preserve: the difference between harm to competitors and harm to competition. Once business-user satisfaction becomes a proxy for the public interest, the review will systematically overweight the voices of organized complainants who want compliance never to be declared complete. It will underweight the millions of consumers who benefit from integrated products and have no standing committee at the Commission’s compliance workshops. The feedback the regime is wired to receive is precisely the feedback that confirms it. 

Put the two together, and the result is a regulation impervious to disconfirmation. Unfairness is structural, so it is always present. After all, when will rights and obligations be in perfect balance? Exactly: never. Gatekeepers are always the beneficiaries of a lopsided distribution, so their losses never count as costs. And because unfairness is structural and axiomatic, there is no counterfactual against which to judge whether the distribution of rights and obligations is “fair.” Compared to what?

Given the conflation of public and private interests, the public interest is likely to become whatever organized business users say it is. Given the assumption that any obligation imposed by the DMA is achievable, any failure to reach that benchmark will be read as either an infringement or malicious compliance. 

A regime constituted on those premises is not self-evidently “working” just because its review says so, any more than a bus headed for a cliff is working because it is still moving. As I have repeatedly stated in tweets I now cannot unearth: Movement is not progress. 

The One-Way Ratchet

None of this would be alarming if the DMA were cheap to get wrong. It is not. The conduct it prohibits per se is precisely the category for which the error-cost literature counsels caution, because the welfare effects are context-dependent and false positives are difficult to reverse. A wrongly acquitted monopolist may still be disciplined by entry. A wrongly condemned product design is abandoned, and the foregone innovation is never observed. 

A regime that cannot perceive its own false positives will keep producing them. Worse, it will mistake the absence of complaint from the people it has overridden for evidence of success. 

The Commission describes the DMA as an “iterative process,” a reassuring phrase that implies learning. But iteration only produces learning if the loop is closed—if some outcome can send the signal: stop, reverse, this was wrong. 

The DMA’s loop runs in one direction. Each iteration adds obligations, monitors compliance, and treats shortfalls as grounds for more intervention. The first review was supposed to be the moment when the architecture faced an external test. Instead, the enforcer reviewed its own enforcement, declared the results positive, and concluded that what the regime needs is not greater scrutiny, but more vigorous application of the same approach—with cloud computing and artificial intelligence next in line. 

That is the real significance of the review. An instrument confident enough to expand before demonstrating that it can succeed is not being disciplined by evidence. It is being driven by an axiom. 

And the question the DMA’s evaluative architecture is structurally incapable of asking—whether the costs of intervention exceed its benefits—remains unanswered because it cannot be asked. The possibility that the premise itself is wrong has been designed out of the system. 

A review that cannot find error is not evidence that the policy works. It is evidence that the policy has stopped looking.

The post Fit for Purpose, Unfit for Review: Why the DMA’s First Evaluation Could Never Find Failure appeared first on Truth on the Market.

Last week, a federal court in São Paulo suspended the daily fine that Brazil’s Administrative Council for Economic Defense (CADE) had imposed on Meta for refusing to open WhatsApp to rival AI chatbots. On June 5, Turkey’s competition authority announced both an abuse-of-dominance investigation into the same conduct and an interim measure giving Meta one month to admit third-party AI assistants. Three days later, the Italian Competition Authority (AGCM)—the first agency anywhere to order interim relief against the WhatsApp restrictions—closed its case in deference to the European Commission, which had expanded its own proceedings to cover Italy. Then, on June 9, the Commission adopted interim measures of its own, giving Meta five working days to restore rival assistants’ access across the European Economic Area (EEA). Africa’s Common Market for Eastern and Southern Africa (COMESA) Competition Commission, for good measure, is investigating, too. 

Meta is not the only company in regulators’ crosshairs. At its developer conference on June 8, Apple announced that its new Siri AI features will not launch in the European Union with iOS 27, citing the Digital Markets Act’s (DMA) interoperability requirements. The next day, the Commission reportedly rejected Apple’s request for an 18-month exemption, characterizing the company’s decision as a business choice. Earlier this month, meanwhile, the UK Competition and Markets Authority (CMA) imposed its first AI-related conduct requirement on Google under Britain’s new digital-markets regime, governing how publisher content may be used in AI Overviews. 

One week. A half-dozen authorities. Four continents. 

And a common thread: nearly all of this activity is occurring before any agency has demonstrated, in a final appealable decision, that the challenged conduct actually harms competition. Interim measures, preventive suspensions, and ex ante mandates have become the enforcement tools of choice in AI markets. Whatever else one makes of these interventions, they share a defining feature: they front-load the costs of being wrong. 

The Error-Cost Framework Strikes Back

The heightened scrutiny is understandable. AI technologies are likely to become important inputs across much of the economy, making concerns about future market power inevitable. Yet competition policy risks repeating mistakes from earlier periods of technological change. Regulators may be intervening before they understand how competition in AI markets actually works.

The error-cost framework associated with the Chicago School, and most prominently with Judge Frank Easterbrook’s seminal article “The Limits of Antitrust,” emphasizes the dangers of premature intervention under conditions of uncertainty. Because markets are dynamic and often self-correcting, conduct that initially appears exclusionary may ultimately prove efficient, innovative, or beneficial to consumers. Easterbrook therefore warned that false positives—mistakenly condemning procompetitive conduct—can be especially costly.

His prescription was not inaction. Rather, it was a set of sequential filters—market power first, followed by a plausible profit-from-harm theory—to screen out cases where the expected costs of error exceed the expected benefits of intervention.

Once antitrust law prohibits a beneficial business practice, the resulting loss of innovation and competition can persist for years. By contrast, many instances of market power are disciplined over time by entry, innovation, and technological change. The asymmetry is institutional. Erroneous condemnations become embedded in precedent and, if corrected at all, are usually corrected only through slow appellate or legislative processes. Erroneous acquittals remain subject to market discipline.

Put differently, Type II errors come with a built-in correction mechanism. Type I errors do not. The lesson is not that antitrust should never intervene. It is that intervention warrants particular caution when regulators have limited confidence about how markets will develop.

Many contemporary policy discussions implicitly assume that AI markets are becoming concentrated and therefore require early intervention. Current market realities suggest a far more fluid competitive environment. 

OpenAI, Anthropic, Google, xAI, Meta, DeepSeek, Alibaba, Mistral, and numerous other firms continue to compete intensely across multiple dimensions, including model performance, cost, deployment strategies, and ecosystem development. Leadership at the technological frontier has changed hands repeatedly over the past three years. The DeepSeek episode—in which a comparatively small lab matched state-of-the-art performance at a fraction of the presumed cost—should have put to rest the notion that today’s leaders are already entrenched, at least for now. 

Ironically, the partnerships between cloud-computing incumbents and AI startups that most concern regulators have arguably financed much of this competitive entry. Notably, every merger authority to examine those partnerships reached the same conclusion. The CMA closed all five of its AI-partnership reviews without remedies. After a 15-month investigation of Microsoft/OpenAI, the agency concluded that even the deepest of those relationships fell short of conferring control. 

Competition also increasingly takes place across several interconnected layers of the AI stack, including semiconductors, cloud infrastructure, foundation models, application-layer services, and deployment environments. Success in one layer does not necessarily translate into durable dominance in another. As one of us has argued at length elsewhere, the most important competitive bottlenecks in AI may emerge at deployment interfaces—the points where users and businesses actually access AI services—rather than within foundation models themselves. 

This dynamism should give regulators pause before assuming competitive outcomes that have yet to emerge. 

Maybe Data Moats Aren’t Moats After All

A second feature of AI markets deserves attention, though it warrants more caution. 

It is increasingly claimed—by both AI enthusiasts and critics—that AI systems improve through continuous deployment and feedback. The claim is most plausible in robotics and other physical AI applications. Unlike traditional software, which is updated periodically, these systems may evolve through ongoing interactions with users, environments, sensors, and operational constraints. 

Industrial robots, autonomous systems, logistics platforms, and enterprise AI applications generate enormous amounts of operational feedback. On this account, that feedback becomes an input into future improvements. Learning is not confined to the development stage but continues after deployment. Japan’s Fair Trade Commission (JFTC) devoted a new chapter of its generative-AI market study to autonomous driving this spring for precisely this reason. 

The result, at least in theory, is a recursive learning process. Deployment generates data. Data improves performance. Improved performance encourages further deployment. Additional deployment creates new learning opportunities.

The key question is how powerful these feedback loops actually are. Having enough data typically matters far more than having the most, and the returns to additional data often diminish rapidly. The same dystopian narrative that once predicted insurmountable data moats in search and social media is now being replayed for foundation models. Whether AI deployment loops will prove more durable or consequential than their search-engine predecessors remains an open empirical question, not a premise to be assumed.  

Yet these feedback loops sit at the heart of virtually every theory of harm currently advanced in AI markets. The Federal Trade Commission’s (FTC) Section 6(b) report on AI partnerships relies heavily on them. So do the remedies in the U.S. Google search case, which were expressly extended to generative AI to prevent Google from converting purported data advantages into AI dominance. 

Authorities that invoke these loops to justify intervention must also accept the corollary. If the loops are real and economically significant, interventions that sever them may destroy far more value than static analysis suggests. If the loops are weak, many of the underlying foreclosure theories collapse on their own terms. Either way, the case for aggressive early intervention appears weaker than it first seems. 

The Cost of Being Early

Traditional antitrust interventions often assume relatively stable market structures and that corrective measures can be imposed without significantly affecting innovation. AI markets may be different. 

If technological progress depends on recursive learning processes, premature intervention may disrupt the very mechanisms through which innovation occurs. Restrictions on product integration, limits on deployment strategies, or mandatory redesigns imposed before competitive harm is established may reduce experimentation, slow learning, and ultimately diminish innovation. 

The costs of false positives would then extend beyond conventional efficiency losses. They could alter the trajectory of technological development itself. 

That possibility makes the current enthusiasm for interim measures especially striking. Under Article 8 of Regulation 1/2003, interim relief requires only a prima facie finding of infringement and a risk of serious and irreparable harm to competition—a standard so exceptional that the European Commission invoked it only once during the regulation’s first two decades. By design, it is the enforcement tool most exposed to error costs because it rests on predictions rather than demonstrated effects. 

Yet in the WhatsApp saga, interim measures have become the tool of choice. Italy used them in December 2025. Brazil followed in January 2026. Turkey and the Commission acted this month. Authorities across four continents are now rewriting the terms on which a private messaging platform deals with AI developers in real time, before any of them has established that consumers were harmed. 

The concern becomes even more significant in physical AI systems. Robotics, autonomous vehicles, industrial automation, and other deployment-intensive applications depend on continuous real-world experimentation. Learning cannot be fully replicated in laboratories or regulatory sandboxes. It emerges through actual operation. 

A mistaken intervention that disrupts those learning processes may therefore have consequences that persist long after the underlying enforcement error becomes apparent. 

Gatekeepers, Special Responsibilities, and Other Status Crimes

These concerns are particularly relevant to the growing tendency to apply doctrines that focus more on firm status than demonstrated competitive effects. 

One example is the European concept of “special responsibility,” first articulated in Michelin v Commission and steadily expanded since. Under the doctrine, firms deemed dominant often face obligations that go beyond ordinary competition-law standards. 

When detached from a concrete assessment of exclusionary effects, special responsibility begins to function as a status-based constraint on conduct. Competition authorities and courts may come to treat product integration, ecosystem design, or self-preferencing as presumptively suspect simply because a successful dominant firm engages in them. That approach departs from competition-on-the-merits principles. 

Nor is this logic confined to Article 102 case law—or to Europe. It is hard-coded into the DMA, whose obligations attach to designated gatekeepers regardless of demonstrated effects. The result is the spectacle of European consumers losing access to Siri AI while Brussels and Cupertino argue over whose “choice” that was. 

The same logic animates the Commission’s pending Article 6(7) specification proceedings, which concern how Google must provide rival AI services with “equally effective” access to Android’s hardware and software features. One of us has criticized that interoperability theory as more slogan than substance. Similar thinking underlies the CMA’s new conduct requirements under the United Kingdom’s strategic-market-status regime. It also echoes through the WhatsApp interventions in Brazil and Turkey. 

AI may therefore become the first major technology to be born into a status-based regulatory environment, rather than growing up under effects-based antitrust scrutiny. 

The same impulse is beginning to reshape adjacent doctrines. Last month, Brazil’s CADE ordered the retroactive notification of Microsoft’s Inflection acqui-hire, a transaction that fell below every applicable filing threshold, on the theory that talent and licenses can substitute for an acquisition of control. The agency simultaneously opened new probes into Google’s Windsurf and Hume AI arrangements. Whatever the merits of treating acqui-hires as concentrations, the direction of travel is unmistakable: jurisdictional doctrines are being remodeled to reach AI transactions before their effects can be observed. 

This danger is particularly acute in AI markets. Product integration is often not a strategy for exclusion, but a mechanism for improving performance and, if the deployment-learning thesis is correct, accelerating learning. 

Integrating AI assistants into operating systems, embedding models within enterprise software, or optimizing interactions between hardware and software may generate substantial efficiencies. Those efficiencies often arise precisely because integration enables faster feedback, tighter coordination, and more effective learning. 

Condemning such conduct as self-preferencing without a concrete demonstration of exclusionary effects may amount to penalizing successful innovation rather than protecting competition. Treating integration as unlawful merely because it advantages a firm’s own products departs from competition-on-the-merits principles and risks undermining the very innovation process that competition policy is supposed to protect. 

Back to First Principles

None of this implies that AI markets should be exempt from antitrust scrutiny. 

Exclusionary conduct remains possible. Firms that control critical infrastructure, computing resources, data assets, or distribution channels may engage in practices that genuinely foreclose rivals. If durable market power emerges and is used to restrict competition, antitrust law should respond. 

The point is not that intervention is never warranted. It is that intervention should remain grounded in evidence of competitive harm, rather than assumptions based on firm size, ecosystem integration, or speculative fears of future dominance. 

Nor is such restraint merely theoretical. In the U.S. Google search case, Judge Amit Mehta rejected both structural divestiture and a proposed requirement that Google provide advance notice of its AI investments beyond existing Hart-Scott-Rodino obligations because the evidentiary record could not support either remedy. Taiwan’s Fair Trade Commission reached a similar conclusion from the opposite direction. After a year of consultation, it determined that generative AI requires no new enforcement instruments. The existing Fair Trade Act, applied through an issue-driven rule-of-reason framework, is sufficient. 

Effects-based enforcement is not a euphemism for inaction. It is a methodological commitment to identifying actual competitive harm before imposing remedies. 

The fundamental question remains the same one that has guided sound antitrust policy for decades: does the conduct harm competition, or merely competitors? 

The Machines May Be Learning. Regulators Should, Too.

The emergence of AI does not undermine the traditional error-cost framework. If anything, it reinforces its central insight. AI markets are characterized by extraordinary uncertainty, rapid technological change, and intense competition among alternative technological pathways. Under such conditions, regulators are especially likely to mistake efficient conduct for anticompetitive conduct. 

That danger is magnified to the extent that AI systems—as many enforcement theories themselves assume—improve through recursive learning, deployment-driven experimentation, and continuous feedback from real-world use. Interventions that disrupt those processes may do more than reduce static efficiency. They may alter the trajectory of innovation itself. 

In AI markets, the costs of false positives therefore extend beyond the immediate consequences of a particular enforcement decision. They may affect the pace and direction of technological development for years to come. 

For that reason, the error-cost framework deserves renewed attention. Far from rendering Judge Easterbrook’s insights obsolete, AI may strengthen the case for caution when competitive effects remain uncertain. 

The events of the past week suggest that enforcement is moving well ahead of that insight. Competition authorities should remain vigilant against genuine exclusionary conduct. They should also recognize that preserving experimentation, deployment, and innovation is itself a core objective of competition policy. 

In markets defined by continuous learning and technological change, the competitive process is often best protected through patient, effects-based enforcement rather than remedies imposed before the evidence is in. 

After all, if AI’s defining feature is that it learns from experience, competition policy should do the same.

The post Act First, Learn Later: AI Antitrust and the Error Costs of Regulation at Machine Speed appeared first on Truth on the Market.

Late last month, the California Assembly passed the COMPETE Act, based on recommendations from the California Law Revision Commission. The bill is now under consideration in the California Senate. The act would significantly expand liability under the Cartwright Act, California’s antitrust statute, to such an extent that it would likely become the most interventionist state antitrust law in the country.

As currently drafted, the statute could place a wide range of common business practices at legal risk. That would make California an antitrust outlier and could discourage investment and economic growth. At the same time, there is no assurance the statute would benefit consumers and, in some circumstances, it could even leave them worse off.

Antitrust by Guesswork

The statute’s key feature is a new cause of action that would allow suits against businesses for anticompetitive conduct by a single firm, as distinct from collusion among multiple firms, which the law already covers. The federal Sherman Act also provides a single-firm cause of action, but the California statute makes a critical change.

The statute would make firms liable for conduct that “monopolizes” or “unreasonably restrains trade,” rather than only conduct that “monopolize[s],” as the federal statute provides. That added phrase makes all the difference.

Monopolization suits inherently require courts to identify exclusionary conduct without punishing business practices that are simply part of vigorous competition. Over more than a century, federal courts have developed—through trial and error—a structured framework to meet that challenge. By contrast, there is no settled understanding of what it means to “unreasonably restrain trade” in the single-firm context.

Businesses therefore would have little guidance on compliance, especially because the statute also provides that courts should treat federal antitrust precedent as “at most, instructive.” Guess wrong, and the result could be treble damages or civil or criminal penalties, which were increased by legislation enacted in December 2025. While the statute excludes independently owned, California-based businesses with $10 million or less in annual “gross receipts” and 100 or fewer employees, it still leaves a wide range of medium and large businesses—including franchisees of large national businesses—exposed to significant liability.

The risks of overenforcement are compounded by the statute’s instruction that courts interpret the Cartwright Act “liberally” to “maximiz[e]” effective deterrence. That contrasts with the more balanced approach in federal antitrust case law, which strives to deter anticompetitive conduct without suppressing procompetitive practices.

The statute also pursues a wide range of policy objectives, such as “free and fair competition,” protecting “all trade participants,” and preserving “democratic, political, and social” institutions. That broad remit strains the rule of law and could yield enforcement actions and judicial decisions that depart from antitrust’s traditional focus on preserving competition and protecting consumers.

Taking the Guardrails Off

The risk of enforcement overreach is heightened by the statute’s unusually specific instructions that courts need not follow several doctrines developed in federal monopolization cases. These doctrines share a common purpose: reducing the risk that antitrust law mistakenly condemns procompetitive business practices. The result could be enforcement outcomes that protect certain competitors at consumers’ expense.

Predatory Pricing. The statute disclaims Supreme Court precedent—most notably, Brooke Group Ltd. v. Brown & Williamson Tobacco Corp.—requiring predatory-pricing claims to show both below-cost pricing and a reasonable prospect that the defendant can later recoup its losses through monopoly pricing. Courts developed these requirements to avoid discouraging price cuts that benefit consumers. Without them, businesses unable to match a rival’s prices could seek relief through antitrust litigation instead.

Multi-Sided Platform Efficiencies. The statute disclaims Supreme Court precedent—specifically, Ohio et al. v. American Express Co. (2018)—requiring courts to consider whether alleged anticompetitive harms on one side of a multisided platform are offset by procompetitive benefits on another side. A multisided platform is a business that serves distinct groups of users who interact through the platform, such as advertisers and consumers using a search engine. Without that framework, a court could condemn certain practices in digital advertising markets without considering how those same practices may benefit consumers through zero-price search, messaging, or other complementary services.

The “As Efficient” Principle. The statute disclaims certain federal appellate decisions requiring plaintiffs in monopolization cases—particularly those involving bundled discounts—to show that the challenged conduct impedes rivals that are at least as efficient as the defendant. The principle reflects the view that monopolization law should not protect less efficient competitors from more efficient ones. Abandoning it could discourage practices such as bundled discounts and other package deals that often benefit consumers.

If You Can’t Win in Court…

Viewed more broadly, the statute’s interventionist approach reflects a shift from the federal to the state level of a vigorous campaign—in scholarship, advocacy, and policymaking—to move U.S. antitrust law away from protecting the competitive process and toward achieving redistributive outcomes by limiting firm size and market concentration as ends in themselves. As I have shown elsewhere, federal courts have largely rejected legal theories motivated by these redistributive principles, which generally conflict with governing case law grounded in the consumer-welfare standard and the effects-based rule-of-reason methodology. Advocates have therefore turned to state legislatures to implement this alternative model of antitrust law.

Supporters of this state legislation rely heavily on the now-popular view that purportedly lax antitrust enforcement has increased market concentration and entrenched incumbents. Yet the real world is more complicated.

There is no scholarly consensus that the U.S. economy is unusually concentrated by historical standards or that antitrust policy was the determining factor in those digital market segments where concentration is high. Sectors with high concentration can still feature intense competition on price, quality, and innovation, as the current rivalry among major cloud-computing services and artificial-intelligence model developers illustrates. Moreover, it is well-established that size and concentration within a certain range can benefit consumers when economies of scale or scope—cost savings or product improvements that come from producing more goods or offering complementary services—yield gains in cost, quality, or convenience that are passed on to consumers.

The pending statute departs from core principles of antitrust law and economics to address a competition problem that has not been soundly demonstrated. Its loosely defined cause of action, expansion of policy goals beyond protecting competition, and rejection of doctrinal guardrails against overenforcement may encourage litigation targeting practices that pose no reasonable risk of anticompetitive harm. The statute arguably goes beyond even the broader liability standards of European competition law, which antitrust reformers often laud as a model, because it is enforceable through class actions and treble damages—plus attorney’s fees—that are uncommon in Europe.

The Golden State’s Self-Inflicted Wound

The combination of expansive liability standards and private enforcement could significantly alter California’s legal environment for the medium and large businesses—and small franchisees—that fall within the statute’s scope. Capital is mobile, and competitive pressures require firms to avoid regulatory environments that place them at a disadvantage.

If enacted in its current form, the statute may lead national businesses to shift investment and employment to other states. It may also discourage discount pricing, package deals, and other business practices that benefit California consumers—especially lower-income consumers—but cannot be matched by less efficient rivals.

In the name of protecting competition, California risks making itself less competitive. The ultimate losers may be the very consumers antitrust law is supposed to protect.

The post California’s Other Wealth Tax appeared first on Truth on the Market.

When Mozilla reports that Firefox is now selected through a DMA browser-choice screen once every 10 seconds—more than 6 million selections in total, with daily active iOS users 113% higher in the European Union than they would have been absent the regulation—the result is treated as proof that the DMA is working. Consumer choice has spoken. 

When fewer than 1% of Meta users chose its new paid, ad-free subscription option—introduced largely to comply with DMA edicts—over the existing free, ad-supported alternative, that result was treated as proof that the DMA needs to go even further. Critics pointed to the low uptake as evidence that Meta had priced the subscription to deny users a “genuine” choice. Consumer choice, it seems, needed a chaperone. 

Notice the asymmetry. In the first case, consumer choice is a verdict to be respected. In the second, it is a symptom to be cured. The common thread is not consumer welfare, but direction of travel. Choice matters when it moves users away from a designated gatekeeper. When it does not, regulators and commentators look for reasons to discount it. 

I develop this argument at greater length in a new International Center for Law & Economics (ICLE) white paper. It nonetheless merits separate treatment because it captures something about the DMA that more familiar critiques—focused on per se rules, limited judicial review, or the absence of an effects-based analysis—do not fully reach. 

The DMA does not ignore consumer choice. It instrumentalizes it. Choice enters the analysis on one condition: it must support more intervention. 

Choice for Me, but Not for Thee

Competition law, for all its faults, treats consumer behavior as informative. The entire consumer-welfare framework rests on the premise that what consumers actually do—what they buy, switch to, or stick with—tells us something about whether conduct is helping or harming them. 

If consumers continue choosing an integrated product after being shown alternatives, that is evidence. It may suggest the product is better, cheaper, or more convenient. It may suggest switching costs are real, but that the benefits of switching are modest. Effects analysis exists precisely to sort among those possibilities, rather than assume the answer in advance. 

The DMA dispenses with that inquiry. Its objectives of “fairness” and “contestability” are defined, by the Act’s own terms, independently of “the actual, potential or presumed effects of the conduct of a given gatekeeper … on competition on a given market.” In other words, the DMA’s core concepts do not depend on showing that conduct harms competition or consumers. 

Recital 33 defines fairness only in negative terms, with “unfairness” deemed to be “an imbalance between the rights and obligations of business users where the gatekeeper obtains a disproportionate advantage.” There is no benchmark, no counterfactual, and—crucially—no role for revealed preference. What consumers actually choose places no meaningful constraint on intervention. 

What emerges is a regime with a built-in direction, but no clear destination. The premise of structural unfairness is that designated markets produce unfair outcomes because they are structurally unfair. Whenever an asymmetry remains—whenever users have not migrated, whenever a gatekeeper’s product remains popular—the diagnosis is the same: not enough has been done yet. 

The logic is self-reinforcing. Persistent gatekeeper success is not considered evidence that might falsify the premise. It is taken as evidence that the premise has not yet been fully implemented. 

That helps explain why the DMA struggles to process the most mundane explanation for why people keep using gatekeepers’ products: fully informed consumers may simply prefer them. A framework that took that possibility seriously would have to treat sustained uptake of an integrated product as a reason to stop intervening—or at least to reconsider whether the intervention is working as intended. The DMA is structurally inclined to reach the opposite conclusion. Heads, the regulator wins; tails, the consumer switches. 

Fairness Isn’t Free

None of this would matter much if intervention were costless. It is not. Reshaping how consumers interact with core platform services is not a side effect of the DMA. It is the mechanism. And that reshaping carries costs that fall, at least initially, on the very consumers in whose name the regime ostensibly operates.

Some of those costs are already visible. Mozilla’s browser-choice data came with an acknowledgment that users now face more friction when navigating online. Apple has delayed or withheld a series of features in the European Union—including iPhone Mirroring, Live Translation for AirPods, and, most recently, its revamped Siri AI—each time citing the difficulty of complying with interoperability mandates without exposing user data.  

One can debate, as the European Commission and Apple do, how much of this reflects genuine engineering constraints and how much amounts to strategic foot-dragging. The debate itself nonetheless concedes the point. The regime is changing the consumer product and, at least in some cases, making it less convenient or less capable. 

There is a respectable case for accepting those costs. A DMA defender can argue openly that some consumer convenience, quality, and integration are worth sacrificing to achieve less concentrated markets and greater competition over time. Some proponents say exactly that: short-term inconvenience is the price of long-term contestability. That is a coherent position. 

It is also an inherently political one. It reflects a choice to benefit one set of interests—organized business users, rivals, and complementors—at some expense to another: consumers who prefer the status quo. 

The trouble is that the DMA rarely presents the tradeoff in those terms. Instead, it is packaged in the universal language of “fairness,” a word that suggests nobody loses. 

Yet the Organisation for Economic Co-operation and Development (OECD) has acknowledged that fairness in this context is “strongly tied to redistribution.” The Commission’s own expert panel similarly described the fairness objective as achieving “a higher standard of fairness in the distribution of the social value generated by large platforms.” 

Redistribution may be a defensible objective. But redistribution from consumers to business users, carried out in the name of consumers, is something else entirely. 

My colleagues and I have made a version of this argument before. The Google and Apple designation decisions illustrate how little users actually matter under the DMA, even as policymakers invoke their interests at every turn. 

The Difference Between Policing and Steering

That sleight of hand is easier to see once we are precise about the kind of regime the DMA creates. The problem is not merely that the DMA sometimes discounts consumer preferences or understates consumer costs. It is that the DMA treats consumer behavior differently from competition law because it is doing a different job. 

The DMA is not simply an “oversight” regime. It is an “ordering” regime. And that distinction is the whole ballgame. 

An oversight regime—such as competition law in its Article 102 Treaty on the Functioning of the European Union (TFEU) form—polices markets from the outside. It treats consumer behavior as a partly exogenous input: something to be measured, explained, and respected, even when the explanation is inconvenient.

If consumers keep buying the integrated product, the enforcer must grapple with that fact. The burden is to show harm to consumers, and persistent consumer uptake is evidence that demands an explanation. A diffuse and numerous consumer class cannot easily be conscripted into any particular firm’s commercial agenda. 

An ordering regime works in the opposite direction. It begins with a vision of what the market should look like—more rivals, more switching, more disaggregation—and treats consumer behavior as an outcome to be steered toward that vision. 

Choice architecture ceases to be something regulators should approach neutrally and becomes a policy tool. The question is no longer, “What do consumers reveal that they prefer?” It becomes, “Are consumers moving in the right direction, and if not, what additional steering is required?” 

Consumer choice is not ignored. It is repurposed—from a constraint on the regulator to an instrument of the regulator. 

That repurposing helps explain why the regime is resistant to its own failure. Every outcome can be interpreted as validation. Significant migration to rivals vindicates the intervention. Persistent loyalty to gatekeepers justifies intensifying it. 

The European Commission’s first review of the DMA reflects precisely this disposition. It treats designations, proceedings, and enforcement outputs as evidence of success, while treating reports of consumer harm as consultation inputs rather than findings that might call the underlying premise into question. 

A bus headed for a cliff is not “working” simply because it is moving. 

What If They Already Chose?

There is an older intellectual lineage here, though not the one usually invoked in the DMA’s defense. 

Ordoliberalism, the tradition that shaped postwar European competition governance, is generally remembered for its commitment to an “economic constitution“: a rules-based order insulated from day-to-day political pressure and policed by a neutral guardian, rather than steered toward preferred outcomes. On that telling, ordoliberalism is the opposite of market ordering. 

But the tradition always carried a more directive strand. Franz Böhm, one of ordoliberalism’s founders, described the “true core of an economic constitution” as “a steering norm, which guides economic happenings in a politically desirable direction.” He also accepted that the state should suspend “indirect economic steering by legally-governed competition” whenever “direct market steering by methodical means of command” seemed more appropriate. 

The competitive order was to be tended by a nobilitas naturalis—an “aristocracy of the public spirit” trusted to read political needs and pull the right lever. 

The DMA is that directive strand, shorn of the constitutional constraints that were supposed to discipline it. The European Commission is cast as the aristocracy of the public spirit, charged not with refereeing the competitive process, but with deciding what “fair” digital markets should look like and steering consumers toward them. 

Once the regulator’s role is to define desirable outcomes rather than police a competitive process, consumer choice loses its standing as an independent fact. It becomes one more input for the steering norm to interpret—respected when it cooperates, corrected when it does not. 

That is the deepest sense in which the DMA instrumentalizes choice. The problem is not merely that the Commission reads the data tendentiously. Under an ordering logic, there is no other way to read it. A regime built to redesign markets around a preferred template cannot also treat consumers’ revealed preferences as a verdict on whether that template was worth pursuing. It can only treat them as a progress report. 

That leaves the question the DMA is structurally unequipped to ask: What if consumers do not need steering? What if they already chose? 

The post If at First Consumers Don’t Switch, Regulate Again appeared first on Truth on the Market.

That day, the U.S. government cut non-Americans off from access to two of the world’s most advanced AI models. Brussels’ flagship response to precisely this scenario—the Cloud and AI Development Act (CADA)—would not have helped. 

As I argue in “Europe’s Sovereignty Stack: CADA, Compute, and the Limits of Autarky,” Europe needs to do two things instead: remove legal barriers and build serious computing capacity on the continent. That strategy would allow Europe to pursue two complementary goals. 

First, it would give Europe leverage through mutual dependence, strengthening its hand in negotiations over access to cutting-edge AI capabilities that it has little realistic chance of developing domestically in the near term. Second, it would reduce dependence on foreign providers for “good enough” AI systems, which will remain economically important across a wide range of applications that do not require state-of-the-art models. 

The Alliance, the Act, and the Kill Switch

A U.S. export-control directive ordered Anthropic, a leading American AI lab, to suspend non-U.S. access to its two most capable models—Fable 5 and Mythos 5—after a jailbreak exposed a security vulnerability. The company complied within hours, pulling the models globally for non-U.S. users while disputing the order’s proportionality. Its other models remained available. 

That was not an isolated event. It followed a remarkable 10-day run. 

On June 2, the White House issued an executive order granting the federal government up to 30 days of pre-release access to “covered frontier models”—advanced AI systems at the edge of current capability—and a role in choosing the “trusted partners” who get them first. 

On June 3, the European Commission proposed CADA, the centerpiece of its Technological Sovereignty Package. 

That same day, EU member states moved to join “Pax Silica,” the U.S.-led chip alliance, under an understanding to buy at least $40 billion in American AI chips. 

So the European Union decided to bind itself more tightly to the American stack—the layers of chips, cloud infrastructure, and models that make modern AI possible—while CADA proposed walling its most sensitive workloads off from that same stack. Then the kill switch fired. 

Open to the World, Closed to the Providers

CADA would create four “assurance levels” for public-sector cloud and AI services. The sensible part is that the framework is risk-based and applies only to the public sector. 

At the highest assurance levels, though, CADA seeks to shield European buyers from third-country interference by effectively excluding even American-owned providers. 

Henna Virkkunen, the European Commission vice president responsible for technology sovereignty, made that goal explicit. She told reporters that no provider handling critical workloads should have a “kill switch,” that “we want to make sure that our most critical sensitive data is stored in Europe,” and, according to press accounts, that U.S. firms would struggle to qualify for the highest sovereignty tier because of the U.S. CLOUD Act. 

CADA contains a single statutory escape hatch from its ownership restrictions, and it does not apply at the highest assurance level. A provider’s home country can be recognized as an “associated third country,” but the criteria appear tailor-made to exclude the United States. One provision would disqualify countries that maintain measures that “impede the provision of state-of-the-art technologies.” 

Read literally, that would disqualify a country for imposing the very export controls that underpin the Western technology alliance. It would exclude a partner for participating in the same system the European Union joined through Pax Silica during the very week CADA was unveiled. 

In a communication accompanying the proposal, the Commission insisted that technological sovereignty “does not mean isolation, protectionism, or tech decoupling,” and promised that the European Union would remain “open to the world.” That assurance is difficult to reconcile with both the text of the proposal and Commissioner Virkkunen’s public statements. 

The Problem Isn’t Dependence. It’s Leverage.

Perhaps the biggest flaw in CADA’s approach is that it effectively gives up access to the best AI capabilities precisely where they matter most: defense, intelligence, and other national-security applications. 

To be sure, CADA includes a safety valve that permits exceptional purchases when “no adequate or reasonable alternative” exists. But the exemption comes with conditions—such as prior unsuccessful tenders—and appears deliberately designed to be difficult to use in practice. That is a curious approach given the Commission’s own analysis, which acknowledged that some degree of dependence is “inevitable.” 

No amount of wishful thinking—or attempts to redefine the term—can change the reality that frontier AI is available only from the leading U.S. labs. Europe cannot develop comparable capabilities quickly enough to capture both the economic and military-intelligence benefits of being at the frontier. 

In highly competitive domains such as cybersecurity, even a delay of a few months can matter. Nor is there an obvious alternative source. The strongest non-U.S. AI labs are largely based in China, not Europe, and most assessments place them at least several months behind the leading American firms, with larger gaps in some areas. 

The June 12 export restriction exposed a deeper problem: the European Union has little leverage when it comes to negotiating access to state-of-the-art AI capabilities. Building that leverage—not restricting access to the technologies Europe already lacks—should be the real focus of European policy. 

Build the Bargaining Chip, Not the Fantasy

The immediate answer is not a sovereign European frontier model. That capability is not arriving on any meaningful timeline, and focusing exclusively—or even primarily—on that goal would waste scarce time and political attention. 

The better answer is to build the one thing Europe realistically can: compute capacity on European soil, measured in tens of gigawatts. What matters most is not who owns the accelerators or whose models run on them. It is where the metal sits. 

European compute could provide the leverage Europe lacks. Demand for computing capacity will outstrip even American supply. A European Union with large, modern data-center capacity to offer—and a willingness to host U.S. firms’ workloads—would create the kind of mutual dependence that makes a future cutoff too costly to contemplate. If U.S. firms run not only European workloads but also U.S. workloads in European data centers, then cutting off Europe would not merely sacrifice non-U.S. revenue. It would directly harm U.S. customers. 

To build that capacity—and win enough of the market to create real dependence—Europe needs to move fast. It needs permitting, including environmental permitting, and grid connections quick enough to make investing in the European Union more attractive than waiting elsewhere. In that legal environment, public subsidies may not be necessary. Without those changes, even large subsidies will not be enough. 

On security, Europe should draw the line along the right axis. Instead of treating any third-country control as suspect, CADA should target adversary control. And the recognition decision should become an openly political judgment about allies and adversaries, not a legalistic trapdoor for excluding partners. 

The main security burden should fall on measures that depend far less on the provider’s nationality: customer-held encryption keys, protection for data while it is being processed, vetted and security-cleared staff, anti-tamper controls, and portability. 

Bolted to Our Floor

In CADA, the Commission got one thing right: the need to accelerate permitting for data centers and related infrastructure. It just was not nearly ambitious enough. The proposal pairs a modest fast-track with promises of insulation from foreign control that its own impact assessment all but concedes are unattainable, given Europe’s dependence on American frontier AI. 

The report “Europe 2031,” written by researchers who want a more competitive Europe, sketches where a CADA-like strategy could lead. Public institutions most committed to “Buy European” become “the ones paying the ransoms,” outmatched by competitors using better foreign tools. The compute gap never closes. Washington rations access to frontier AI by alliance tier. Economic and political pressures mount as Europe finds itself with fewer options and less leverage. 

The report’s prescription is simple: make sure the data centers that get built are “bolted to our floor.” 

That is the real lesson of June 12. Europe does not need the illusion of autonomy. It needs leverage. And leverage comes not from pretending dependence can be wished away, but from building assets others cannot afford to lose.

The post Bolted to the Floor, Not Written in the Clouds appeared first on Truth on the Market.

That is the problem at the center of the current debate over free speech in the digital age. For years, the public argument focused on whether private technology companies were moderating too much content. A different concern has now moved to the forefront: “jawboning,” the practice of government officials using informal pressure, implicit threats, or regulatory leverage to induce private intermediaries to suppress speech on the government’s behalf.

Late last week, Sens. Ted Cruz (R-Texas) and Ron Wyden (D-Ore.) introduced the Justice Against Weaponized Bureaucratic Overreach to Networked Expression Act, or JAWBONE Act. The bipartisan legislation seeks to bring jawboning into the open and provide Americans with a legal mechanism to challenge government efforts to coerce private actors into censoring lawful speech. 

The proposal has much to recommend it. At the same time, it should be viewed as only a first step toward curbing the federal government’s ability to conduct coercive speech-suppression campaigns. To be fully effective, it should be paired with structural reforms that limit government officials’ discretion to pressure online platforms over speech decisions. 

How to Censor Without Looking Like It

What the JAWBONE Act gets right is its recognition that modern censorship rarely takes the form of a direct government order to stop speaking. More often, it operates through censorship by proxy. 

As I argued in a recent paper on jawboning, much of today’s speech ecosystem depends on multisided platforms—businesses that connect content creators, audiences, advertisers, payment providers, and other participants. Because these platforms often depend on government licenses, regulatory approvals, or antitrust clearance, they can be particularly vulnerable to government pressure. 

When officials exert pressure on one part of that ecosystem, the effects can spread throughout the network. If the Federal Communications Commission (FCC) hints that a broadcaster’s license may be at risk, or if the Federal Trade Commission (FTC) opens an investigation into an advertising agency or fact-checking organization, the consequences can extend well beyond the immediate target. Those actions can alter incentives across the platform and, in turn, affect what speech reaches the public. 

The result is a marketplace of ideas shaped less by voluntary choices and more by the preferences of those who hold government power. Because these interventions often occur indirectly, they can evade meaningful First Amendment scrutiny and leave the public unaware of the government’s role in influencing speech outcomes. 

Everyone Loves a Shortcut

The JAWBONE Act’s bipartisan sponsorship reflects an important reality: jawboning is a tool of institutional power, not a uniquely partisan tactic. Without stronger accountability mechanisms, whichever party controls the executive branch will have incentives to use government leverage to influence speech indirectly. 

The statements Sens. Cruz and Wyden issued alongside the bill underscore this point. Republicans have argued that the Biden administration used the Cybersecurity and Infrastructure Security Agency (CISA) and other federal agencies to pressure major technology platforms to suppress content related to COVID-19 and election integrity. Democrats, meanwhile, have criticized actions by the Trump administration and FCC Chairman Brendan Carr, alleging that threats of regulatory scrutiny were used to pressure broadcasters over disfavored late-night programming and political candidates appearing on programs such as The View. 

My paper identifies additional examples. When the FTC investigates fact-checking organizations such as NewsGuard or pressures Apple over editorial decisions in news curation, the underlying mechanism resembles earlier efforts by the Obama and Biden administrations to encourage banks and payment processors to cut ties with disfavored businesses or political causes. The targets may differ, but the basic dynamic remains the same: government officials pressure private intermediaries, which then shape what speech, information, or viewpoints reach the public. 

That dynamic can produce a form of backdoor censorship that distorts the marketplace of ideas while obscuring the government’s role in the process. 

Giving the First Amendment a Remedy

The JAWBONE Act is designed to target the legal barriers that have historically allowed jawboning claims to evade judicial review. In cases like Murthy v. Missouri, plaintiffs have faced significant hurdles in proving standing, and courts often dismiss cases as moot when policies change or officials leave office. 

The bill addresses those structural barriers in several ways. First, it creates a federal cause of action that allows individual Americans to sue government agencies or employees who coerce platforms into suppressing protected speech. It also permits plaintiffs to seek monetary damages, which may help address some of the standing problems presented in Murthy. 

Second, the bill would impose liability for attempted coercion. A violation occurs regardless of whether the censorship succeeds. That feature directly addresses the chilling effect of backdoor censorship efforts: once an official exerts coercive pressure, the legal line is crossed, even if the platform refuses to comply. 

Finally, the bill would require federal agencies to submit disclosures to Congress regarding certain communications with private platforms. By bringing those communications into the open, the bill targets one of jawboning’s central advantages: secrecy. 

You Can’t Sue Away Discretion

While the JAWBONE Act would represent a major step forward for accountability, transparency, and judicial review, lawsuits alone cannot fully protect the marketplace of ideas. As long as government agencies retain broad discretion over speech intermediaries, opportunities for jawboning will remain. 

The bill includes an exception for government action “taken pursuant to a lawful investigation under, or the enforcement of, Federal or State law” that “does not violate the First Amendment to the Constitution of the United States.” That carveout may be necessary to preserve agencies’ ability to perform legitimate regulatory functions. At the same time, it creates a potentially significant loophole whose boundaries courts would need to define. 

The FCC, for example, would likely argue that investigations of broadcast licensees fall squarely within its statutory authority to determine whether broadcasters serve the “public interest.” The FTC could similarly contend that investigations of speech platforms merely reflect its responsibility to police “unfair or deceptive” practices. 

The key legal question would be whether such actions cross the line into coercion under the standard articulated in Bantam Books v. Sullivan and reaffirmed by the Supreme Court in NRA v. Vullo. As I noted previously in evaluating the Jimmy Kimmel controversy: 

The problem for Jimmy Kimmel here is that the FCC has very broad authority over broadcasters because they can review the use (and transfer) of licenses under the “public interest” standard. Caselaw suggests the FCC can revoke licenses, deny their transfer, and issue fines for things like not giving political candidates the right to respond, news distortion, and broadcasting obscenity. As will be discussed below, the Supreme Court has allowed the FCC to be the cop on the beat, so to speak, in policing whether licensed broadcasters are acting “in the public interest” with little restriction on how the FCC defines those terms.

This makes the situation very different than the Rhode Island commission [in Bantam Books], which could, at best, merely recommend prosecution for obscenity. Here, the FCC is more like the cop who is giving a warning to stop breaking the law. The import of this difference would be the source of debate in a case brought by Kimmel against the FCC…

In this case, Kimmel could argue that Carr’s statements “went far beyond advising… [on] legal rights and liability” to a “scheme of state censorship.” For its part, the FCC could respond that Carr was merely “aiding” local broadcasters to comply with the FCC’s regulations, which the FCC has authority to enforce. A court would need to determine whether the comments crossed the line into coercion aimed at censorship, or whether they were merely advising broadcasters of their responsibilities. It isn’t clear which way this would go.

That uncertainty points to a broader problem. When an agency possesses licensing authority, investigative powers, or other forms of regulatory leverage, even a polite suggestion about how the law might apply can carry an implicit threat. Such conduct may influence speech without necessarily rising to the level of unconstitutional coercion under existing First Amendment doctrine. 

To get at the heart of the issue, the JAWBONE Act would need to be paired with broader structural reforms. One option would be to narrow the expansive discretionary authority granted to agencies like the FCC and FTC. Vague mandates—such as policing the “public interest” or “unfair or deceptive” practices—create opportunities for regulators to blur the line between legitimate oversight and pressure campaigns directed at speech. 

Congress, and ultimately the Supreme Court, might also reconsider legacy decisions such as Red Lion Broadcasting Co. v. FCC and FCC v. Pacifica Foundation. Those cases afforded broadcasters reduced First Amendment protection based on assumptions about media scarcity that are increasingly difficult to justify in today’s communications environment. Strengthening the constitutional protections available to speech intermediaries would make them less vulnerable to government pressure and reduce opportunities for future jawboning. 

The Cure Is Not the Prevention

The JAWBONE Act reflects a serious attempt to address a real constitutional problem. By allowing citizens to sue government actors directly for monetary damages, it would give those harmed by backdoor censorship a potential avenue for accountability, even when speech platforms decline to challenge government pressure themselves. 

The bill would also help bring government coercion into public view and deter the most serious abuses. But transparency and liability address only part of the problem. To safeguard the marketplace of ideas over the long term, policymakers should also confront the vague, expansive powers that allow federal agencies to pressure speech platforms in the first place. 

Jawboning thrives in the shadows of discretion. Shrink those shadows, and the First Amendment has room to breathe.

The post Government by Raised Eyebrow: The JAWBONE Act and the Problem of Censorship by Proxy appeared first on Truth on the Market.

The European Commission’s new European technological-sovereignty package deserves close attention in Washington, Brussels, and boardrooms on both sides of the Atlantic. Announced June 3, the package aims to strengthen Europe’s capacity in semiconductors, artificial intelligence, cloud computing, open source, and digitized energy infrastructure.

Those are worthy goals. Europe is right to care about resilience, cybersecurity, and whether its firms and public institutions can use advanced digital tools securely. But good objectives do not rescue bad instruments.

For present purposes, the centerpiece is the proposed Cloud and AI Development Act, or CADA. The Commission says CADA would support cloud and AI research, speed data-center deployment, and create a single EU framework to assess cloud and AI “sovereignty.” The ambition is no small thing: Europe wants to at least triple data-center capacity within five to seven years, expand cloud and AI adoption in public and strategic sectors, and reduce reliance on non-EU providers.

The package also includes Chips Act 2.0, an open-source strategy, and an energy-sector digitization roadmap.

These initiatives should be evaluated as proposals, not faits accomplis. CADA and Chips Act 2.0 still must proceed through the EU’s ordinary legislative procedure, with the European Parliament and Council acting as co-legislators. That means amendments, political bargaining, and, ultimately, either a more balanced transatlantic approach or a more deeply protectionist one.

The coming consultation and legislative process therefore matter. This is precisely the stage at which economically grounded criticism can do the most good.

From Security Standards to Nationality Tests

CADA’s most troubling feature is its sovereignty-assurance framework. The Commission’s public description identifies four levels. Level 1 turns on whether data processing and storage occur in the EU. Level 2 requires providers to demonstrate independence from third countries and transparency across the software supply chain. Level 3 requires EU ownership and control, along with additional criteria, such as personnel citizenship—though the Commission may recognize third-country providers. Level 4 requires full transparency and control over the software supply chain, with no third-country interference.

That language may sound administratively tidy. In economic terms, it looks more like a ladder whose upper rungs were built for European firms.

A U.S. cloud provider is not merely being asked to satisfy cybersecurity, reliability, or technical-performance standards. It is being asked to negate the legal consequences of being a U.S. firm. The CLOUD Act requires covered providers to comply with lawful U.S. process for data within their possession, custody, or control, regardless of whether the data are stored inside or outside the United States.

European officials may dislike that feature of U.S. law. They may reasonably treat it as a sovereignty concern. But using it to demote U.S. firms within a procurement or assurance framework is a market-access condition in all but name.

That distinction matters. A neutral security rule asks whether a provider can protect data, maintain resilience, prevent intrusions, document access controls, comply with lawful process, and protect customers from unauthorized disclosure. A nationality-inflected sovereignty rule asks who owns the provider, which legal system may reach it, where its personnel sit, and whether a third country can be said to exert legal influence.

The former measures performance. The latter sorts firms by political identity.

That shift is especially problematic because cloud is a scale market. Hyperscale cloud is not a local utility that lawmakers can duplicate by aspiration. It requires enormous capital investment, energy access, engineering talent, global service reliability, cybersecurity expertise, specialized hardware, developer ecosystems, and continuous innovation.

Steering public- and critical-sector demand away from the most capable providers because they are American would impose an invisible tax on European users. That tax may not appear on the face of the statute. It will show up in higher costs, weaker services, delayed AI deployment, and lower productivity.

The Protectionist Playbook, Updated for the Cloud Era

The Information Technology and Innovation Foundation’s (ITIF) Aegis Project offers a useful framework for understanding the problem. Aegis starts from a simple proposition: America’s leading technology companies are not incidental to U.S. economic power and national security. They are among the principal means by which the United States maintains technological leadership in a world where China pursues state-directed innovation mercantilism.

The project’s Non-Tariff Attack Tracker catalogs policies around the world that are framed as neutral domestic regulation but, in practice, weaken U.S. technology firms, extract value from them, or tilt markets toward national champions.

ITIF describes such measures as anticompetitive market distortions (ACMDs): “government-imposed policies or practices that distort from open trade, competition on the merits, and property rights protections.” Economic research on ACMDs—supported by econometric modeling and pioneered by trade-policy expert Shanker Singham—finds that such distortions misallocate capital, skew risk-taking, reduce productivity and innovation, and ultimately slow economic growth.

CADA’s sovereignty tiers fit this pattern. They are not tariffs. They do not explicitly say that U.S. cloud providers are unwelcome in Europe. In fact, the Commission stresses that much of the market will remain open to international partners.

But competition can be distorted without formally closing the border. If the most valuable public-sector and critical-infrastructure opportunities depend on independence from non-EU legal jurisdictions, EU ownership and control, EU personnel, or the absence of third-country influence, U.S. firms are not competing on the same terms as local firms. They are competing after regulators have imposed a handicap.

ITIF’s analysis of EU cloud-service restrictions identifies similar dynamics in related contexts. Cloud-certification schemes, digital-sovereignty requirements, data-localization mandates, and jurisdictional restrictions can function as de facto barriers to U.S. providers, even when participation remains formally voluntary.

This is a familiar behind-the-border trade problem. It rarely announces itself as protectionism. Instead, it arrives dressed as resilience, autonomy, privacy, procurement integrity, or cybersecurity. If the practical effect is to shift demand from globally competitive U.S. firms to less competitive domestic alternatives, the economic substance is protectionist.

Security Is Not a Passport

The strongest defense of CADA is that Europe has legitimate reasons to worry about legal and geopolitical dependence. That much is true. No serious observer should dismiss concerns about cyber risk, supply-chain fragility, or state access to sensitive data.

The question is not whether Europe may regulate for security. The question is whether the chosen regulatory proxy improves security at a reasonable cost.

On that score, nationality is a crude—and often misleading—proxy. A European-owned provider may have weaker cybersecurity practices than a U.S. provider. A U.S. provider may offer stronger encryption, more transparent compliance processes, better incident response, and more robust auditing than a local competitor.

Likewise, a provider organized under European ownership may still depend on non-European chips, open-source components, cloud tools, developer libraries, or security vendors. Digital infrastructure is not sovereign because a corporate charter is European. It is secure because its architecture, governance, controls, and incentives produce security.

The relevant policy question is one of comparative institutional performance: Which rule addresses the asserted problem at the lowest social cost?

A performance-based security standard targets risk directly. It can require encryption, logging, access controls, incident disclosure, independent audits, redundancy, supply-chain transparency, and enforceable commitments to customers. A nationality- or jurisdiction-based test targets risk indirectly, and often poorly. It excludes or downgrades firms that may be best able to satisfy the underlying security objective while privileging firms that satisfy a political criterion.

The error costs are substantial. In dynamic technology markets, regulations that divert demand from efficient suppliers can weaken investment incentives, slow technology diffusion, and shield incumbents or favored entrants from competitive pressure. That is not a recipe for European competitiveness. It is a recipe for a smaller, more expensive, and less innovative digital ecosystem.

There is also a rule-of-law concern. Public-procurement criteria should be administrable, transparent, and tied to the actual risks that justify them. A sovereignty ladder invites discretionary judgments about control, interference, legal exposure, and the recognition of third-country providers. Those judgments can easily become political bargaining chips.

A firm that satisfies demanding technical standards may still find itself in a lower tier because its parent company is organized under a disfavored jurisdiction. That is not competition policy; it is industrial administration.

Once procurement officials begin treating national origin as a signal of trustworthiness, moreover, that signal rarely remains confined to the most sensitive workloads. Risk categories tend to expand. Exceptional rules have a habit of becoming ordinary practice.

The predictable response from affected firms will be costly corporate engineering. Providers may create local subsidiaries, special governance structures, contractual firewalls, joint ventures, or bespoke sovereign-cloud offerings to satisfy formal requirements. Some of those arrangements may make sense for particular customers. When regulation mandates or privileges them, however, they consume resources that could otherwise go toward product improvements, stronger security, lower prices, or new services.

The cost does not stop with compliance departments. Customers bear it through higher prices, and the broader economy bears it through slower diffusion of better technology.

You Can’t Build an AI Continent on Anti-Cloud Policy

CADA also sits uneasily with Europe’s own AI ambitions. The Commission wants Europe to become an “AI continent.” That aspiration depends on cloud capacity, data-center deployment, advanced semiconductors, developer tools, model-training infrastructure, and a deep bench of complementary services.

For now, U.S. firms lead in many of those areas. One can regret that fact, explain it, or try to narrow the gap. Pretending it away is not an industrial strategy.

A policy that channels strategic demand away from U.S. hyperscalers may appear to create room for European challengers. In the short run, though, it risks raising input costs for the very firms and public agencies that need AI most.

Cloud services are general-purpose inputs. They do not merely affect cloud-provider profits. They affect hospitals, manufacturers, universities, logistics firms, software startups, financial institutions, energy companies, and public administrations. Making those inputs worse or more expensive reduces economy-wide productivity.

Nor is this just a European problem. The United States and Europe face a shared strategic challenge from China’s state-directed technology model. Fragmenting the transatlantic technology stack weakens the allied position. It reduces scale, complicates interoperability, undermines standards cooperation, and diverts political energy from the more important task of disciplining predatory or mercantilist conduct by strategic rivals.

If democratic allies build regulatory walls against each other, China need not build all the walls itself.

Chips Act 2.0 illustrates the same tension. The Commission says the new initiative will strengthen Europe’s semiconductor industry, support design and production, speed permitting, stimulate demand, and deepen strategic partnerships. Many of those goals are sensible.

But semiconductor capacity is not built in isolation from U.S. firms. American companies design, develop, supply, and invest across the European semiconductor ecosystem. If EU technology sovereignty becomes a preference for European ownership rather than an effort to build resilient allied capacity, it will undermine the investment and cooperation Europe needs.

Competition Policy or Competitor Policy?

There is a deeper concern. Current market shares and existing dependencies can tempt regulators to treat successful foreign firms as a problem to be managed rather than as engines of competition and innovation.

That temptation has surfaced repeatedly in European digital policy. The Digital Markets Act and related initiatives often begin with the premise that large U.S. technology firms are too powerful and then translate that judgment into regulatory obligations governing product design, data use, interoperability, contracting, and business models.

To be sure, there are signs that EU courts are placing greater emphasis on evidence, proportionality, dynamic competition, and rigorous legal reasoning when reviewing European Commission antitrust and digital-regulation decisions. CADA should be judged by those same standards.

Dynamic competition asks a different question from static market-share analysis: Which rules are most likely to maximize innovation, investment, entry, and productivity over time?

That framework does not assume that today’s largest suppliers deserve permanent market positions. But neither does it assume that regulators can improve outcomes by steering demand toward politically preferred suppliers. The goal should be competition on the merits.

If a European cloud provider offers better security, lower costs, stronger service, or superior compliance, it should win customers. If a U.S. provider offers those advantages, it should win customers. Public procurement should not become a vehicle for industrial favoritism masquerading as sovereignty.

The risk is not merely that U.S. firms lose business. The greater risk is that European users lose access to best-in-class tools and that European innovators must build on a less capable technological foundation.

The costs of that choice compound over time. Lower-quality cloud services slow AI adoption. Slower AI adoption weakens productivity growth. Weaker productivity growth reduces the fiscal and political capacity needed to invest in genuine resilience.

Protectionism marketed as sovereignty can leave Europe less sovereign in the ways that matter most.

Sovereignty Without Protectionism

There is a constructive alternative. Europe can pursue resilience without discrimination. It can define security requirements in technological and contractual terms rather than nationality-based ones.

For example, it can require providers to disclose supply-chain dependencies, submit to independent audits, maintain EU-based redundancy, support customer-controlled encryption, report government-access requests where lawful, and provide robust exit and portability tools. It can reserve the most sensitive national-security workloads for narrowly defined sovereign environments without turning broad public-sector procurement into a buy-European program.

For its part, the United States should take European concerns seriously while making clear that discriminatory market-access conditions are unacceptable. Diplomacy should come first. The goal is not to turn every regulatory disagreement into a trade war.

At the same time, as ITIF’s recent Section 301 report argues in the broader EU context, the United States should be prepared to identify, document, and respond to discriminatory policies that burden U.S. technology firms. Serious negotiations are more likely when both sides understand that behind-the-border discrimination carries consequences.

A transatlantic bargain should be possible. The United States and Europe could work toward mutual assurances on lawful access, transparency surrounding government requests, cybersecurity certification, procurement neutrality, cloud portability, and trusted-vendor criteria that distinguish allies from strategic adversaries.

They could also coordinate semiconductor incentives and AI-infrastructure investments so that public subsidies reinforce allied supply chains rather than fragment them. That approach would strengthen resilience without sacrificing competition.

The key is to avoid turning legitimate sovereignty concerns into a general license for industrial policy. Europe’s problem is not that American cloud firms are too capable. Europe’s problem is that too many of its own firms, agencies, and institutions have been slow to adopt digital technologies effectively.

Punishing the suppliers most capable of helping close that gap is an odd way to solve it.

Don’t Mistake Sovereignty for Strength

The Commission’s technology package remains only a proposal. That is good news. It means policymakers still have time to separate genuine resilience measures from protectionist market design.

CADA’s sovereignty-assurance framework, as currently described, raises serious concerns because it risks turning legal jurisdiction and corporate ownership into decisive competitive criteria. That would burden U.S. providers, reduce European access to leading cloud and AI infrastructure, and weaken the transatlantic technology ecosystem at precisely the wrong moment.

Europe should build. It should invest. It should streamline permitting, expand energy capacity, support research, and accelerate technology adoption. It should also protect sensitive data and critical infrastructure.

But it should pursue those goals through performance-based, evidence-based, and nondiscriminatory rules. Strategic autonomy should mean the ability to choose among excellent technologies from trusted allies, not an obligation to prefer a less capable domestic alternative because it carries the right passport.

If the European Union wants to become an AI continent, it should resist the temptation to become a regulatory island.

The path to technological sovereignty runs through competitiveness, not protectionism.

The post Europe’s Cloudy Judgment appeared first on Truth on the Market.

In March, the International Center for Law & Economics (ICLE) submitted comments to the European Commission on the General Data Protection Regulation (GDPR) and ePrivacy provisions of the Digital Omnibus. The GDPR is the European Union’s main privacy law. The ePrivacy rules govern confidentiality of communications and device-access rules, including the cookie-consent regime that has trained a continent to click “accept” while learning nothing substantial. 

We made four core arguments.

First, policymakers should adopt the proposed entity-relative clarification of the personal-data definition—that is, they should ask whether the particular organization holding the data can realistically identify someone, not whether someone, somewhere, with some imagined tool, might be able to do so.

Second, the package’s artificial-intelligence (AI) provisions represented a necessary legislative settlement of questions that the European Data Protection Board (EDPB), the EU body that coordinates national privacy regulators, had deliberately left unresolved.

Third, the proposed cookie-consent reforms pointed in the right direction, but did not go far enough.

Finally, the package’s greatest weakness was its silence on enforcement architecture. Without institutional reform, we argued, the same authorities that had interpreted the GDPR into a “law of everything” would read the new exemptions just as narrowly. 

Three months later, the two sister components of the Digital Omnibus have diverged sharply. The AI Omnibus (COM(2025) 836) reached a provisional trilogue agreement on May 7. A trilogue is the closed-door negotiation among the Commission, Parliament, and Council that often determines the final shape of EU legislation. The Data Omnibus (COM(2025) 837)—which contains the GDPR and ePrivacy reforms—remains before the Council, where the Cypriot Presidency has circulated successive compromise texts, most recently on May 21 (Council document 9547/26). 

That latest compromise deletes three of the Commission’s four principal GDPR reforms: the entity-relative personal-data test, the relocation of cookie consent into the GDPR, and the legitimate-interest basis for AI processing. It also removes the Commission’s proposed authority to define pseudonymization criteria. Pseudonymization means replacing direct identifiers, such as names, with substitutes, while keeping the possibility of re-identification under controlled conditions. 

What remains of those reforms has largely migrated into recitals and EDPB guidance. Recitals are the explanatory passages that accompany EU laws. They can influence interpretation, but they are not the same as binding operative text. At the same time, the compromise expands the EDPB’s own mandate. 

The concern at the heart of our March comments—that textual reform without enforcement reform would underperform—now looks almost understated. The Council appears poised to deliver less reform than the Commission proposed, while leaving the EDPB stronger than before. 

This post examines how the Council compromise affects each of the issues addressed in our comments, and what it means as the file moves to the European Parliament. 

The Referee Takes the Field

Our comments urged adoption of the amended Article 4(1), which would have codified the entity-relative approach to identifiability reflected in EDPS v. SRB and Breyer. We also supported Article 41a, which would have authorized the Commission—after consulting the EDPB—to specify, by implementing act, when pseudonymization places data outside the scope of the GDPR. An implementing act is a legal instrument through which the Commission sets technical or practical rules under authority granted by legislation. 

That allocation of authority mattered. We warned against leaving the scope of the law to be defined solely by the body that enforces it. 

The Council deleted both provisions. 

The entity-relative test disappeared from operative Article 4(1). Its substance survives only in Recital 27a, which states that identifiability “should be assessed by the controller or the processor, considering the actual technical, organisational and legal capabilities” of that controller or processor. In GDPR terms, a controller decides why and how personal data is processed; a processor handles data on the controller’s behalf. 

The reference to “actual … capabilities” is welcome. It addresses the recurring tendency to treat purely hypothetical means of identification as legally relevant. But the clarification now appears only in a recital, rather than in binding legal text. 

Article 41a fared even worse. The Council eliminated the Commission’s implementing-act authority entirely. Renumbered as Article 29a, the provision now directs the EDPB to issue an opinion on pseudonymization and anonymization within 12 months of the regulation’s entry into force. 

Peter Craddock greeted the new Article 29a with a pointed question: “Possibly the most useless legal provision ever proposed?” On our reading, it is worse than useless. A provision originally intended to check the EDPB’s steadily expanding jurisdiction—by giving the Commission authority to define the contours of pseudonymization—now does the opposite. It more explicitly empowers the Board to define the limits of its own jurisdiction. The revised Article 70(1) confirms the point by assigning the Article 29a opinion to the EDPB. 

That institutional shift matters because the Commission and the EDPB have not always agreed on how broadly data-protection rules should reach. In EDPS v. SRB, the Commission intervened to push back against a maximalist interpretation of identifiability. Under the Council text, by contrast, the EDPB will determine what qualifies as effective pseudonymization. 

This is the same body that interpreted Article 5(3) of the ePrivacy Directive so broadly that URL parameters and ordinary browser transmissions became consent-triggering forms of “access.” Expecting a notably pragmatic approach to pseudonymization would be optimistic. 

Of the two deletions, the loss of Article 41a matters more in practice. The Article 4(1) amendment was symbolically important, and we continue to believe it should be restored. Standing alone, though, it might not have changed an enforcement culture resistant to the idea that “personal data” has meaningful limits. 

Article 41a addressed the institutional problem directly by vesting interpretive authority in a body other than the EDPB. That is precisely why the EDPB and the European Data Protection Supervisor (EDPS), the EU’s independent data-protection authority for EU institutions, called for its deletion in their February joint opinion. It is unfortunate that the member states chose to follow that advice. 

The Deal That Unmade Itself

We recommended adopting both Article 88c, which would have created a legitimate-interest basis for AI development and operation, and Article 9(2)(k), which would have created a derogation for special-category data that appears incidentally in AI training. “Legitimate interest” is one of the GDPR’s lawful bases for processing personal data. Special-category data includes especially sensitive information, such as data revealing health, race, ethnicity, political opinions, religious beliefs, biometric identifiers, or sexual orientation. 

Article 88c has disappeared from the operative text, but much of its substance survives in Recital 33a. Most of the Commission’s language remains, albeit with important changes. Processing that the Commission said “may be pursued for” legitimate interests now merely “may be regarded as” carried out for a legitimate interest. The unconditional right to object—the central safeguard, which we described in March as stronger than the ordinary Article 21(1) standard—has been deleted, along with the enhanced-transparency requirement. The Council also added a caveat stating that the recital “does not affect the obligation … to choose the most appropriate lawful ground.” 

The result is a return to the ordinary Article 6(1)(f) balancing test, with EDPB Opinion 28/2024 serving as the de facto framework. That opinion was notable for what it did not resolve, leaving broad discretion to enforcement authorities. Both sides of the Commission’s original bargain have therefore disappeared: the clearer legal basis that AI developers were supposed to receive, and the unconditional objection right that data subjects—the people whose data is processed—were supposed to receive in return. 

To be fair, the distinction between operative text and a recital may matter less in practice than the attitude of the authorities enforcing it. A supervisory authority determined to reach a particular result can often interpret around either one. But that observation supports, rather than undermines, the point we made in our comments: institutional incentives matter at least as much as statutory language. 

Article 9(2)(k) survived, albeit in narrowed form.

The Council added the qualifier “incidental and residual,” and moved the accompanying conditions into a new Article 9(5). Controllers must take measures to avoid collecting special-category data; erase such data once identified; and, where erasure is impossible or manifestly disproportionate—for example, because the information has been memorized by a model—protect it against further processing, inference, or disclosure. They must also document the process. The provision expressly excludes data collected through prompts during deployment. 

Even in this narrower form, Article 9(2)(k) remains useful. Web-scale AI training makes the incidental collection of sensitive data effectively unavoidable. The provision answers a question that legislators can no longer sidestep. Indeed, it may be the only significant AI-related GDPR reform still standing in the package. 

The AI Omnibus, meanwhile, is essentially finished. 

The agreed text extends the bias-detection carveout—formerly Article 10(5) of the AI Act, and now a standalone Article 4a—beyond providers of high-risk systems to deployers and to non-high-risk AI systems and models. In AI Act terms, providers develop or place AI systems on the market, while deployers use them. The carveout preserves the “strictly necessary” threshold and the existing safeguards. It operates through Article 9(2)(g) GDPR; the AI Act supplies the required basis in EU law, while Article 4a supplies the accompanying safeguards. 

The agreement also postpones the compliance dates for high-risk systems, moving Annex III obligations to Dec. 2, 2027, and product-embedded AI obligations to Aug. 2, 2028. At the same time, it accelerates the Article 50 watermarking requirements to Dec. 2 of this year and adds a new Article 5 prohibition on AI-generated child sexual-abuse material and non-consensual intimate imagery. 

Taken together, the two files produce a striking asymmetry. The deliberate use of sensitive data for bias detection—the relatively uncommon case—now rests on firm legislative ground. The incidental presence of sensitive data in training datasets—the ubiquitous case—remains trapped in a contested file, within a provision the Council has already narrowed. 

Meet the New Cookie Rules, Same as the Old Cookie Rules

On ePrivacy, we were blunt in March. The Commission’s proposal was the weakest part of the package. It created a two-track regime, moving personal-data-related device access into a new Article 88a of the GDPR while leaving non-personal data under Article 5(3) of the ePrivacy Directive. Its analytics exemption was too narrow to cover standard third-party analytics services, and it omitted several low-risk exemptions that should have been obvious candidates, including fraud prevention, advertising measurement, frequency capping, and contextual advertising. 

The Council solved the two-regime problem by abandoning the reform altogether. 

Article 88a is gone. Device access remains governed by Article 5(3) of the ePrivacy Directive, albeit in revised form. That means seven years of expansive EDPB interpretations of Article 5(3)—covering everything from cookies and tracking pixels to URL parameters—remain the starting point. Whatever the revised exemptions ultimately say, the EDPB will continue to play a decisive role in determining how broadly or narrowly they are interpreted. 

The revised exemptions move toward some of the reforms we recommended, only to hedge them so heavily that they risk becoming ineffective in practice. 

The audience-measurement exemption now permits analytics conducted “by a third party acting on the provider’s behalf.” That directly addresses the analytics problem we identified and could help resolve the long-running Google Analytics disputes. But the exemption applies only if the data remains anonymous and aggregated, and is neither combined with other data nor shared. 

Fraud prevention now appears within the security exemption, but the drafting limits it to protecting the “security of the interface.” As Craddock observed, fraud prevention typically aims to protect the broader service from fraudulent users, rather than the interface itself. Under the current text, “in most cases, fraudsters will then need to give their consent before anything can be done against them.”  

The contextual-advertising exemption fared even worse. It appeared in the February compromise text but disappeared by May, even though the joint opinion of the EDPB and EDPS had itself recommended exempting contextual advertising. Contextual advertising targets ads based on the content a user is viewing, rather than by tracking that user across sites and services. 

The browser-signals provision survives. Renumbered as Article 88a, it now applies to operating-system providers as well as browsers. The provision also includes a one-click refusal mechanism, a six-month cooling-off period following refusal, and a 24-month transition period. National GDPR supervisory authorities would serve as ePrivacy enforcers. 

The comparison we drew with the United Kingdom has, if anything, become sharper. 

The cookie provisions of the Data (Use and Access) Act 2025 took effect Feb. 5. Rather than relying on a single “strictly necessary” exemption, the law creates five exemptions, including an analytics exemption that expressly permits third-party providers acting as processors under an opt-out model. It also authorizes the secretary of state to create additional exemptions through secondary legislation.  

Last month, the Information Commissioner’s Office (ICO), the United Kingdom’s data-protection regulator, went further, formally advising the government to remove low-risk advertising activities from consent requirements altogether. The ICO proposed a “first-party framework” covering ad delivery, measurement, frequency capping, brand safety, and ad-fraud prevention, while retaining consent requirements for cross-service tracking and profiling. The proposal was evidence-based, supported by cost analysis, and tested through citizen juries. 

The ICO now also operates under a statutory duty to balance privacy concerns against innovation, competition, and crime prevention. We do not endorse every conclusion the ICO reached. One could argue that the cleaner solution is simply to repeal the cookie-consent rule and allow the GDPR’s risk-based framework to govern device access. Even so, the contrast between the British and European approaches is stark. 

Our recommendation therefore remains unchanged. The best solution is to repeal Article 5(3) of the ePrivacy Directive and allow the GDPR to govern device access without a parallel consent regime. Short of that, the European Parliament should at least detach the fraud-prevention exemption from the concept of “interface” security, remove the restrictions that make the analytics exemption largely unusable, and restore the contextual-advertising exemption. 

The Changes You Might Have Missed

The compromise also reshaped three quieter corners of the file. Two concern provisions that our March comments did not address in detail. All three, however, follow a familiar pattern: legislative reforms survive, but key questions about their scope are increasingly delegated to the EDPB. 

The EDPB Defines Science

The compromise retains the new GDPR definition of “scientific research” in Article 4(38), but in a form largely rewritten to the EDPB’s specifications. Research must now be “conducted in an autonomous and independent manner,” follow recognized methodological standards, and produce “verifiable and transparent results.” 

The Commission’s original language acknowledging that research may support innovation and commercial objectives has been removed from the operative text and relocated to Recital 28. At the same time, the EDPB has published Guidelines 1/2026 on scientific research, currently open for public consultation, which seek to define the concept through guidance.

The result is familiar. The legislation preserves a broader category, but much of the practical work of deciding who qualifies for it will occur through EDPB interpretation.

Yes, But Only If You Don’t Use It

The Commission proposed, and the Council retained, a new Article 9(2)(l) allowing biometric verification when both the biometric data and the means of verification remain under the data subject’s sole control. Biometric verification confirms that a person is who they claim to be by comparing a live biometric sample—such as a face or fingerprint—to a stored template.

The Council added one further limitation: the verification must be “one-to-one,” meaning the system compares a user against a single stored biometric template rather than searching a larger database.

The more significant constraint may lie in Recital 34. As revised by the Council, the recital instructs controllers to prioritize authentication methods that do not use biometric data and to select the “less intrusive” option whenever two methods are equally effective.

That language comes directly from EDPB Opinion 11/2024 on facial recognition at airports. Notably, the opinion rejected biometric verification even where travelers had a genuine choice among authentication methods. The operative rule therefore creates a new permission, while the accompanying recital imports reasoning that could substantially limit its practical use. 

The Algorithm Still Says No

The Council retained the Commission’s broader rewrite of Article 22 on automated decision-making, but reverted to the familiar “right not to be subject” formulation. In plain English, Article 22 governs when consequential decisions—such as some credit, employment, or eligibility decisions—may be made solely by automated systems without meaningful human involvement.

By returning to the “right not to be subject” language, the Council preserved the prohibition-in-principle approach endorsed by the Court of Justice of the European Union (CJEU) in SCHUFA. 

The Commission’s most useful clarification did not survive in operative text. Its proposal would have made clear that a decision can be “necessary” for the performance of a contract even when a human could theoretically have made the same decision. That clarification now appears only in Recital 38. 

Meanwhile, the revised Article 70(1) assigns the EDPB responsibility for specifying the criteria governing Article 22 profiling decisions. 

Once again, the pattern is hard to miss. The legislation settles some questions, moves others into recitals, and leaves the EDPB with a larger role in determining how the final framework will operate in practice. 

The Board Wins

In our March comments, we argued that the Digital Omnibus was fundamentally incomplete. The same authorities whose interpretations helped create the current dysfunction would remain responsible for interpreting whatever reforms the legislature enacted. The package contained no mechanism to discipline that process.

The Council compromise has done more than validate that concern. It has expanded the EDPB’s formal authority in the very same text that removes many of the Commission’s proposed legislative settlements.

Recital 40a instructs national supervisory authorities to ensure that national guidance on matters covered by EDPB guidelines “does not contradict” the Board’s positions. Nothing comparable appeared in the Commission’s proposal. The result is a new form of vertical lock-in, giving EDPB interpretations greater influence over national enforcement.

The revised Article 70(1) likewise expands the Board’s role. The EDPB is now tasked with developing guidelines that specify the criteria for Article 22 profiling decisions and Article 32 security measures. It will issue the Article 29a opinion on pseudonymization and anonymization. It will also gain authority to establish data-protection-impact-assessment (DPIA) and breach-notification lists directly, rather than merely proposing them to the Commission. A DPIA is a risk assessment that organizations must sometimes conduct before engaging in higher-risk data processing.

Viewed together, these changes raise an obvious question: where do the substantive issues that the Commission originally sought to settle now reside? 

In other words, many of the questions that the Commission attempted to answer through legislation have been returned to the EDPB, either directly or indirectly.

Accountability remains limited.

In WhatsApp Ireland v. EDPB, the Grand Chamber confirmed in February that the Board’s binding decisions under Article 65 may be challenged before the EU courts. That is a genuine improvement. But it applies only to the relatively rare instances in which the EDPB issues formally binding decisions. 

The situation is different for the guidelines and opinions that do most of the practical work of shaping enforcement. In Meta v. EDPB, the General Court dismissed a challenge to the Board’s consent-or-pay opinion because an EDPB opinion “does not produce binding legal effects.” The appeal remains pending.

That distinction matters because the Omnibus’ substantive compromises have increasingly migrated into precisely those instruments—guidelines, opinions, and interpretive documents—that courts generally will not review.

That development makes every element of the structural reform agenda outlined in our March comments more urgent. Mandatory proportionality assessments for EDPB outputs, an independent multidisciplinary review mechanism, and a clearer standard for judicial review all matter more after the Council’s compromise than they did before.

Parliament’s Last Chance to Legislate

The Council may reach a general approach this month, before Cyprus hands the presidency to Ireland on July 1. In the European Parliament, responsibility is shared between the Committee on Industry, Research and Energy, led by rapporteur Aura Salla of the European People’s Party, and the Committee on Civil Liberties, Justice and Home Affairs, led by rapporteur Marina Kaljurand of the Socialists and Democrats. 

Kaljurand has been openly skeptical. At the committee’s January hearing, she warned that allowing AI systems to process sensitive data for bias detection would mean the “technological neutrality of the GDPR will no longer exist.” The Socialists and Democrats group has likewise rejected what it calls “unacceptable deregulation and weakening of the EU’s digital rules,” while groups to the right of the European People’s Party have pressed for the package to go further. As of this writing, the committees had yet to settle their working arrangements, and adoption realistically appears likely to slip into 2027. 

The open question is whether Parliament will restore any of the Commission’s reforms to binding text.

For Parliament—and for member states still negotiating the general approach—our recommendations follow directly from the analysis above:

• Restore the entity-relative test to operative Article 4(1). Recital 27a’s “actual technical, organisational and legal capabilities” language belongs in binding text. 
• Restore the Commission’s implementing-act authority over pseudonymization criteria. Article 29a, in its current form, should be deleted. It asks the body whose jurisdiction is at issue to define the limits of that jurisdiction. 
• Restore Article 88c as an operative legal basis for AI development and deployment, together with the unconditional right to object that the Council removed. Retain Article 9(2)(k), even in its narrowed “incidental and residual” form. 
• Keep Article 9(2)(l), but rewrite Recital 34. A permission in operative text should not be paired with a recital instructing authorities to disfavor the very processing the permission allows. 
• On ePrivacy, repeal Article 5(3) without replacement and let the GDPR govern device access. Short of that, detach fraud prevention from the “security of the interface,” remove the restrictions that make the analytics exemption impractical, and restore the contextual-advertising exemption. 
• Delete Recital 40a and reverse the Article 70(1) mandate expansion. A reform prompted by the costs of unaccountable interpretive power should not end by enlarging it. 
• Take up enforcement reform. The structural agenda of our March comments—separating investigation from adjudication, creating multidisciplinary review of consequential decisions, requiring mandatory balancing of all affected rights and interests, and enabling review of EDPB quasi-legislation—is still missing from the Digital Omnibus. 

The Digital Omnibus began as a promise to make EU data law more proportionate, predictable, and compatible with European competitiveness. Six months in, the Council’s answer is a text that keeps the procedural trimmings, moves the substance into recitals and EDPB opinions, and strengthens the very institution whose interpretive practice made reform necessary in the first place.

Parliament now has a chance to prove that EU data-protection law is still made by legislators, not outsourced to the referee.

The post EU Digital Omnibus Hands the Wheel to the Referee appeared first on Truth on the Market.

The underlying facts are not especially favorable to Perplexity. In granting a preliminary injunction, Judge Maxine Chesney of the U.S. District Court for the Northern District of California found “strong evidence” that Perplexity violated both the federal CFAA and California’s analogous statute. According to the court, Perplexity continued accessing Amazon’s systems after receiving a cease-and-desist letter and deliberately evaded the technical measures Amazon deployed to block that access. The 9th Circuit stayed the injunction pending appeal. 

The doctrinal question is the easy one. Amazon will probably win, and probably should. It is also the less interesting question. 

The harder and more consequential issue is whether the CFAA is the right body of law to govern this kind of dispute at all. More broadly, it raises a recurring problem in technology law: whether it is sustainable to keep asking statutes written for the technological realities of the mid- and late-20th century to govern technologies their authors could not have anticipated. 

We think the answer to both questions is no. As Greg Dickinson puts it in his masterful article, “Law Proofing the Future”:

Technological breakthroughs provoke wonder, then fear, then legislation. The resulting legal regimes entrench incumbents, suppress experimentation, and displace long-standing legal principles with bespoke but brittle rules. . . . [Meanwhile,] the most powerful tools for governing technological change—the general-purpose tools of the common law—are in fact already on the books, long predating the technologies they are now called upon to govern, and ready also for whatever the future holds in store.

The interests Amazon seeks to protect are real. But they are fundamentally interests in property and contract, and courts developed the core principles governing those interests long before Congress enacted the CFAA. When statutes track those common-law principles, they often work well. When they depart from them—or prevent the sort of incremental adaptation that characterizes the common law—they tend to generate exactly the kind of doctrinal strain the CFAA now exhibits. 

The lesson for agentic AI is not that Congress needs to enact a new statute. The legal system already possesses a framework capable of absorbing these new facts. Under current political and institutional conditions, any new legislation is more likely to depart from that framework than to reinforce it. 

When a Contract Dispute Wears a CFAA Costume

The factual record is unflattering enough to Perplexity that, under existing doctrine, Amazon has a strong CFAA case. According to the complaint, Perplexity’s Comet browser transmits the same user-agent string as Google Chrome, making its automated activity inside an authenticated Amazon session difficult to distinguish from that of a human shopper. When Amazon deployed a technical block targeting Comet in August 2025, Perplexity allegedly pushed a software update within 24 hours that changed Comet’s fingerprint and restored access. Amazon sent a cease-and-desist letter to Perplexity’s CEO on Oct. 31, 2025. Perplexity declined to comply and instead published a blog post titled “Bullying is not innovation.” Amazon sued four days later. 

The doctrinal question is whether an authenticated session—but one conducted contrary to terms of service that prohibit third-party automation and after Amazon revoked access by contractual and technical means—is the kind of environment in which an automated commercial agent acts “without authorization” under the CFAA. After the Supreme Court’s 2021 decision in Van Buren v. United States, authorization is, roughly, a function of whether the accessed system is “gates-up” or “gates-down.” Here, as Judge Chesney put it, Comet accessed Amazon “with the Amazon user’s permission, but without authorization by Amazon.” 

The Supreme Court in Van Buren took pains to read the statute narrowly, partly because a broader reading could turn routine terms-of-service violations into federal computer crimes. In particular, after Van Buren, it is hard to argue that a user’s authorized session becomes a federal computer crime simply because the user also does something that violates the terms of service. Amazon’s strongest theory is therefore not the bare terms-of-service violation, which, after Van Buren and the 9th Circuit’s decision in hiQ II, might not be enough on its own. 

Instead, Amazon’s case rests on three facts: Amazon deployed a technical countermeasure specifically targeting Comet at the point of access; Perplexity pushed an update within 24 hours to defeat it; and Comet continued to spoof a Chrome user-agent string to evade detection. On that record, Amazon’s claim is not merely that Perplexity breached a contract. It is that Amazon closed a gate, and Perplexity deliberately broke through it. That is much closer to the paradigmatic CFAA case preserved by Van Buren than to the terms-of-service cases the Supreme Court has narrowed. 

So Amazon probably wins.

The problem is that Amazon’s path to victory requires the CFAA to do work that exposes its awkward conceptual fit. The “gate” Amazon seeks to protect under the CFAA is one it erected because its underlying contractual prohibition on agentic access was, by itself, inadequate. Detection was costly, circumvention was fast, and ordinary contract remedies were too slow to matter. Pressing the CFAA into service this way is understandable, but it conflates the intrusion claim the statute was designed to police with the property-and-contract claim actually at issue. 

That raises the larger question: Why is this dispute being resolved under a federal anti-intrusion statute drafted at the dawn of personal computing, rather than through the bodies of law—property, contract, trespass, and duty of care—that courts normally use to allocate responsibility for such harms? And what should the answer tell us about how the law should govern agentic AI? 

The Matthew Broderick Theory of Legislation

The CFAA traces, somewhat embarrassingly, to a Hollywood blockbuster. 

WarGames premiered in June 1983. Within days, President Ronald Reagan reportedly raised the film’s hacking scenario with the Joint Chiefs of Staff. Congressional hearings followed within a year, and Congress enacted the statute that became the CFAA in 1986. The threat model was straightforward: an outsider breaking into a discrete government or defense computer system. The statute’s core concepts—“access without authorization,” “exceeds authorized access,” and “protected computer”—were drafted with that scenario in mind. 

Yet Congress was not entering a legal vacuum. By 1986, the common law had spent centuries developing doctrines for analogous disputes. Trespass to chattels and conversion addressed unauthorized use of another’s property. Nuisance governed conduct that interfered with the productive use of resources. Principal-agent doctrines distinguished between actions that bound a principal and actions that exceeded an agent’s authority. 

Indeed, before the CFAA became the dominant framework, courts often analyzed unauthorized automated access through the common-law doctrine of trespass to chattels. Cases such as CompuServe v. Cyber Promotions (1997) and eBay v. Bidder’s Edge (2000) reached results that look remarkably similar to what Amazon seeks here. The difference is that they did so through doctrines grounded in property and contract. The common law already had tools for addressing these disputes. In important respects, the CFAA displaced those tools rather than allowing them to continue evolving. 

The result is a textbook case of technological panic producing bad legislation. As Kevin Frazier recently observed (discussing Dickinson’s “Law Proofing the Future”): 

Enacted in 1986 to address hacking of government systems, the CFAA soon expanded into a sweeping prohibition on ‘unauthorized access’ to any computer connected to the internet—which, in practice, meant nearly everything. . . . What began as a targeted fix for a Cold War fear became a vague, overbroad, and ill-fitting statute that chilled ordinary activity. It is precisely the kind of misfire we should expect when lawmakers legislate in haste against technologies they barely understand.

The dispute in Amazon v. Perplexity bears little resemblance to the intrusion Congress had in mind in 1986. No one is breaking into a military computer or bypassing a security perimeter to obtain forbidden information. Instead, a commercial software agent is acting on behalf of a paying customer and carrying out transactions the customer is entitled to perform on a platform that invites the customer’s business. 

Amazon’s interests are real. It has legitimate concerns about advertising-impression integrity, fraud detection, authenticated-session security, and the economics of its affiliate program. But those are not the interests the CFAA was designed to protect. They are interests that fit far more naturally within contract, property, and tort law.

You Can’t Give Away Rights You Don’t Have

Absent the CFAA mismatch, the Amazon-Perplexity dispute seems relatively straightforward. 

Amazon owns the servers. Authenticated customer accounts run on those servers. The merchant, affiliate, and advertising relationships layered on top of the platform are contractual. The right to determine the terms of access—including whether and on what terms automated agents may operate within authenticated sessions—is part of the bundle of rights Amazon possesses and may choose to allocate. 

None of that requires the CFAA. Trespass to chattels, breach of contract, tortious interference with business relationships, and unfair-competition law all potentially reach the conduct Amazon challenges. Amazon’s reliance on the CFAA instead reflects the statute’s procedural and strategic advantages—federal jurisdiction, a criminal-law backstop, and enhanced remedies—not the nature of the underlying interests at stake. 

The strongest defense of agentic access rests on user consent. Perplexity’s argument, at bottom, is that its AI agent is simply doing what the end user has authorized it to do. If a user may browse Amazon, why can’t the user’s software agent browse Amazon on the user’s behalf? 

The common law has long had an answer. A guest’s permission to enter property does not automatically entitle the guest to bring along commercial agents whom the property owner has not authorized. Justice Clarence Thomas captured the principle succinctly in his dissent in Van Buren: 

As the Second Restatement of Torts explains, “[a] conditional or restricted consent to enter land creates a privilege to do so only in so far as the condition or restriction is complied with.” . . .

. . . What is true for land is also true in the computer context; if a company grants permission to an employee to use a computer for a specific purpose, the employee has no authority to use it for other purposes.

Use beyond the scope of consent renders the visitor a trespasser. 

The same logic appears in the common law of agency. An agent cannot possess greater authority than its principal. The principle traces to the ancient maxim nemo dat quod non habet: one cannot give what one does not have. Amazon’s contractual prohibition on third-party agentic access is therefore the relevant constraint. A user’s decision to employ Perplexity’s agent cannot expand the underlying property and contractual rights the user received from Amazon in the first place. 

The 9th Circuit has already confronted—and rejected—essentially the same user-consent argument. In Facebook v. Power Ventures, the court held that a third-party aggregator’s continued access to Facebook on behalf of users who voluntarily supplied their login credentials was nevertheless “without authorization” under the CFAA once Facebook issued a cease-and-desist letter and the aggregator circumvented Facebook’s technical restrictions: 

Similarly, for Power to continue its campaign using Facebook’s computers, it needed authorization both from individual Facebook users (who controlled their data and personal pages) and from Facebook (which stored this data on its physical servers). Permission from the users alone was not sufficient to constitute authorization after Facebook issued the cease and desist letter.

Once platform-level permission has been revoked, the court explained, neither technological workarounds nor user authorization can restore it. The implication here is straightforward: authorization from the user is not a substitute for authorization from the platform. A third party that acquires a user’s credentials acquires no greater rights against the platform than the user possesses. 

It’s Not ‘Bullying’ to Protect One’s Property

Perplexity’s public response illustrates a broader problem with how parts of the AI industry frame these disputes. After receiving Amazon’s cease-and-desist letter, Perplexity published its “Bullying is not innovation” blog post. The substance of the argument is that legal constraints on Perplexity’s preferred mode of operation are themselves illegitimate—that requiring an AI agent to identify itself, comply with contractual restrictions, or respect technical barriers constitutes “bullying.” 

That framing gets things backward. Describing the enforcement of property and contract rights as bullying is rhetorical sleight of hand. It asks the legal system to weaken the very institutions that make commercial cooperation and innovation possible. If anything, the stronger claim runs in the opposite direction: demanding unauthorized access to someone else’s property is not “innovation.” 

The concept of “permissionless innovation,” when applied accurately, denotes autonomy from government mandates, not an exemption from private contractual obligations. There is no contradiction in arguing that AI agents should operate without regulatory pre-approval while maintaining that they must obtain consent from the platforms they engage with. Far from being a substitute, a robust system of private ordering serves as the essential foundation for innovation that is truly permissionless.

Indeed, complying with basic conditions of access should be table stakes. Among other things, agentic browsers introduce attack surfaces that did not previously exist, including indirect prompt injection and session-hijacking risks of the sort that have come to be called “CometJacking”—literally named after Perplexity’s browser. Platforms have a legitimate interest in knowing which sessions are being driven by software agents, what those agents are doing, and how to respond when something goes wrong. None of this is rendered moot because it is new technology doing the accessing.

The Supreme Court’s decision in American Broadcasting Cos. v. Aereo offers a useful analogy. Aereo built a technically ingenious system consisting of thousands of tiny antennas, each assigned to an individual user. The architecture was designed to allow Aereo to retransmit broadcast television without paying the licensing fees that a conventional retransmitter would have owed to the broadcaster. 

The company’s argument was essentially that its novel technical design changed the legal character of the underlying conduct. The Supreme Court disagreed. 

The broader lesson of Aereo should apply here. A defendant cannot avoid legal obligations merely by inserting a clever technical architecture between itself and the conduct at issue. Agentic AI is, in this sense, the new Aereo: a novel technical layer placed on top of conduct that would otherwise require the platform’s consent. The novelty of the architecture should not alter the legal analysis. 

The Statute Trap

The CFAA is not unique in routing around the common law. Modern technology law has repeatedly taken problems the common law could have absorbed and recast them into statutes tailored to the anxieties, assumptions, and political pressures of a particular technological moment. The results vary. Some statutes ossify into broad immunity. Others stretch far beyond their original purpose. A few work tolerably well because they preserve enough flexibility for courts to adapt them over time. 

Agentic AI is now provoking the same legislative instinct. But before Congress reaches for the drafting pen, it is worth examining how similar efforts have fared. 

Section 230: The Immunity Machine

The CFAA’s drift away from the common law has a structural counterpart in another statute whose pathologies have come to dominate digital-policy debates: Section 230 of the Communications Decency Act. 

The common law of intermediary liability had already developed coherent answers to many of the problems Section 230 purported to solve. Courts had long grappled with analogous questions through doctrines governing innkeeper liability, premises liability, dram-shop liability, and the duty-to-control doctrine. Across these doctrines, a common principle emerged: liability generally turns on whether the intermediary is the least-cost avoider of the harm, whether the harm was foreseeable, and whether reasonable steps could have mitigated it. 

These are not abstract concepts. They are centuries-old institutions for allocating responsibility among parties whose conduct predictably affects one another. As we have noted elsewhere: 

While it was once in fashion to proclaim the Internet a wholly unique invention to which traditional laws could not readily be applied, a more sober analysis of the history of the common law demonstrates that new business models and new technologies are regularly and inevitably incorporated into the law.

Section 230 was enacted in 1996 to address the narrow doctrinal problem created by Cubby v. CompuServe and Stratton Oakmont v. Prodigy. Over time, however, it hardened into a near-categorical immunity that bears little resemblance to the common-law duty-of-care framework it displaced. Online intermediaries today enjoy a liability shield that no comparable offline business has ever possessed. 

Section 230’s failure lies in ossifying into an immunity that excludes conduct it arguably should reach. The CFAA’s failure lies in stretching to reach conduct it was never designed to govern. But the underlying mechanism is the same: A statute drafted for a particular technological moment imposes categories tailored to that moment, while the common-law principles it displaced lose the opportunity to evolve alongside the technology.

The Sherman Act: Blessedly Vague 

Not every statute touching a fast-moving industry produces this kind of strain. The clearest counterexample is the Sherman Antitrust Act, which, as one of us has argued recently, “is best viewed as a modest statutory extension of the common law.” 

Read literally, the Sherman Act is almost comically terse. But its brevity proved to be a virtue. Its vague language and standards-based structure lent themselves to a common law-like process of judicial interpretation. Over time, courts developed the rule of reason, the per se / rule-of-reason distinction, and eventually the consumer-welfare standard. 

The result is a 136-year-old statute that has survived repeated technological revolutions—from railroads and broadcasting to software platforms and digital markets—because courts have continually adapted its principles to new facts. 

It did not have to turn out that way, and the contrast with more prescriptive competition statutes is instructive. The Robinson-Patman Act in the United States and the European Union’s Digital Markets Act (DMA) both reflect a more top-down approach. The DMA, for example, “applies per se rules to broad swathes of conduct in so-called digital markets,” with “no scope for effects analysis or procompetitive justifications.” 

That is the opposite of the common-law method. It assumes legislators can identify in advance, across industries and technological contexts, which practices will prove anticompetitive. The resulting error costs are predictable and often substantial. 

And the political-economy costs may be worse. Antitrust, like any body of law governing significant commercial interests, attracts rent-seeking by firms hoping to hobble competitors through regulation rather than competition. 

The CFAA, even after Van Buren, sits closer to the DMA end of the spectrum than to the Sherman Act end. Its framework relies on categorical concepts rather than standards-based balancing. If agentic AI ultimately requires a statute at all, the better model would look less like the CFAA and more like the Sherman Act: principles-based, flexible, and designed to leave courts room to adapt the law incrementally. 

The political-economy point matters here as well. When Congress enacted the Sherman Act, organized interest groups had only limited ability to shape its implementation. The same cannot be said of agentic AI. The interested constituencies are already organized, already lobbying, and already shaping the proposals now circulating in Washington. 

The impulse to “just write a new statute” risks producing something closer to the DMA’s rigid framework than the Sherman Act’s relative durability. 

The Digital Millennium Copyright Act: A Useful Warning

The Digital Millennium Copyright Act (DMCA) is perhaps the most instructive example because it occupies the middle ground. 

Congress enacted the DMCA in 1998 to solve a problem that neither the common law nor private contracting could easily solve on their own: governing access to copyrighted works at internet scale. 

The transaction-cost problem is straightforward. Copyright is a property right ordinarily enforced through contracts. Rightsholders license works, users pay for access, and contracts define the terms. But where transactions become highly dispersed, low-value, and extraordinarily numerous, bilateral contracting becomes prohibitively expensive. 

Historically, markets responded by creating collective institutions. Performance-rights organizations such as ASCAP and BMI emerged to facilitate music licensing where individual negotiations would have been impractical. The DMCA represented Congress’s attempt to create a comparable framework by statute before market institutions had developed at sufficient scale. 

Viewed charitably, that logic maps closely onto the agentic-AI access problem. Amazon and Perplexity can, in theory, contract with one another. But bilateral negotiations between every platform and every developer of every agentic system are unlikely to scale. A statutory framework featuring identified agents, safe harbors, opt-outs, and knowledge-conditioned liability could, in principle, reduce those transaction costs. 

The problem is that the DMCA also illustrates how quickly statutory compromises can drift away from their original design. 

The provisions intended to protect rightsholders have gradually been narrowed through judicial interpretation (so judicial interpretation is not a panacea, either). Section 512’s notice-and-takedown framework was intended to provide a workable enforcement mechanism when contracting was impractical. Yet courts have often interpreted the statute to impose far less responsibility on intermediaries than Congress appears to have envisioned. 

Safe harbors intended as conditional protections for intermediaries that take meaningful steps against repeated infringement have, through cases such as Viacom v. YouTube and UMG v. Veoh, evolved into something approaching categorical immunity. Knowledge standards have narrowed. Repeat-infringer requirements have weakened. Decisions such as the 9th Circuit’s Lenz v. Universal have further limited enforcement by reading a fair-use precondition into the statute, despite the traditional rule that fair use operates as an affirmative defense. 

The result is a perpetual game of whack-a-mole. Rightsholders issue automated takedown notices against content that often reappears almost immediately. Meanwhile, the intermediaries whose conduct the statute was meant to regulate enjoy a liability regime increasingly disconnected from the common-law principles of intermediary responsibility that inspired it. 

The DMCA therefore serves both as a model and as a warning. Some aspects of its design closely track common-law principles. Notice-and-takedown resembles traditional abatement doctrines. The Section 512 safe harbors can be understood as a form of least-cost-avoider analysis. 

Its failures emerged when courts transformed those conditional protections into something resembling the broad immunity that Section 230 became. The underlying property interests became underprotected, and the duty-of-care framework gradually hollowed out. 

The implication for agentic AI is straightforward. Any statutory attempt to address the transaction-cost problem should preserve the conditional protections that make the framework work. Otherwise, the same interpretive forces that transformed Section 230 and weakened the DMCA may produce yet another technology statute whose practical operation bears little resemblance to its original design. 

The Market Is Already Solving the Problem

Meanwhile, a statute is necessary at all in this context only where private ordering is too costly. The most encouraging feature of the current moment, however, is that the market is not waiting for the litigation to end. 

Identified-agent regimes, negotiated-access arrangements, and commercial licensing agreements are emerging in real time. Amazon’s first-party agent, Buy for Me, identifies itself through a transparent user-agent token and honors opt-outs communicated through robots.txt and direct merchant requests. Reddit’s litigation against Perplexity has proceeded alongside Reddit’s own program of paid licensing agreements with major AI developers. Large publishers are negotiating direct deals with frontier-model companies. And robots.txt itself—the nearly 30-year-old industry standard through which web crawlers honor exclusion requests—demonstrates that workable private-ordering arrangements can emerge without statutory intervention. 

Whether the resulting equilibrium is optimal in every detail is a question better answered by the parties negotiating it than by legislators or courts. Disputes between platforms and developers of agentic systems involve exactly the kinds of interests that contract law routinely governs: server capacity, monetization, security, brand integrity, and obligations to merchants and advertisers. 

When private parties possess both the incentive and the technical capacity to negotiate access terms, the default presumption should be that they will do so. The legal system’s role is to enforce the resulting agreements, not to specify their contents in advance. 

None of this implies support for forced interoperability, mandatory disclosure, compulsory licensing, or treating platform “openness” as a regulatory objective. The absence of interoperability is rarely evidence of market failure. Forced interoperability often carries significant costs to security, product design, and firms’ ability to differentiate themselves—costs that its advocates routinely understate. 

Nor does the argument imply that Amazon must admit any agent that identifies itself. Quite the opposite. Treating this as a property-and-contract problem means preserving the platform’s right to say no. 

There is, however, a risk running in the opposite direction. The emerging private-ordering equilibrium is itself vulnerable to capture if Congress intervenes. Whatever statute Congress writes for agentic-AI access, incumbent AI developers and incumbent platforms will both seek to shape it to protect their existing positions.

The DMCA’s Section 512 notice-and-takedown regime, whatever its virtues, plainly reflects the interests and bargaining power of the incumbents of 1998. An AI-access statute enacted in 2026 would similarly reflect the interests and bargaining power of the incumbents of 2026, with predictably distortive consequences. 

The default position, unless and until Congress can devise a framework that supports rather than supplants common-law principles, should be to let parties negotiate, let markets adapt, and let courts resolve disputes using the property, contract, and tort doctrines we already possess. 

‘The Only Winning Move Is Not to Play’

For the 9th Circuit, the appropriate posture is restraint. If the panel concludes that Amazon has the stronger CFAA claim under existing doctrine—as it should—it ought to write narrowly. Any holding should turn on persistent circumvention of a deployed technical access control, not on a generalized theory of agentic access. Van Buren counsels exactly that approach. The court should resist the temptation to transform the CFAA into a general framework for governing AI agents’ interactions with commercial websites. 

The longer-run framework lies elsewhere: in property, contract, and trespass. 

For platforms and AI developers, the most productive path forward—private ordering—is, as noted, already emerging. Litigation will continue, and some of it will be necessary. But the equilibrium in this market will be shaped less by judicial opinions than by the agreements parties strike. The law’s modest contribution is to make property, contract, and trespass remedies predictable enough that those agreements become easier to negotiate. 

For Congress, the lesson is one of humility and timing. The CFAA and Section 230 are cautionary tales. The DMCA, the closest analog to a common-law-supportive statutory framework, has generated enough difficulties of its own to warrant skepticism. The Sherman Act is the rare success story, but its success depended in part on the political conditions of 1890—conditions that plainly do not exist in 2026. 

The default should be restraint. 

Amazon v. Perplexity is useful not because it presents a particularly difficult dispute, but because it exposes a recurring institutional temptation. Amazon’s interests are legitimate. Perplexity’s conduct, on the current record, is difficult to defend. The doctrinal stretch required to reach the right result under the CFAA is small enough that the court will likely make it. 

But that does not mean the statute is the right tool for the job. 

The broader lesson is that not every technological development requires a new legal framework, and not every legal problem requires a new statute. The common law has spent centuries adapting old principles to new facts. Agentic AI is new. The underlying questions of property, consent, agency, and trespass are not.

Let platforms exclude. Let agents identify themselves. Let parties negotiate. Let courts enforce the bargains they reach.

The law has seen this movie before.

The post WarGames, Shopping Bots, and the Statute Trap: The CFAA and Amazon v Perplexity appeared first on Truth on the Market.

That is the question Brussels would rather not answer after Apple announced that its new Siri AI features will not ship on iPhones and iPads in the European Union. The European Commission says Apple made a free choice. Apple’s actual choice was between opening the iPhone in ways that could break the privacy-and-security model its customers buy, shipping a product that would invite enforcement, or not shipping at all. 

Call it innovation by ultimatum. 

At Tuesday’s midday press briefing, European Commission spokesperson Thomas Regnier offered Brussels’ official explanation: 

The decision not to allow Siri AI in the EU is Apple’s and Apple’s only… Absolutely nothing in the DMA prohibits Apple from introducing new products in the EU… What Apple is, however, not allowed to do is to close the market. It’s not for them to choose which AI tools our EU citizens get to use or not… EU law is non-negotiable.

Regnier capped the intervention with an analogy: the Commission grants no exemptions, just as a police officer does not exempt a driver from the speed limit. 

The problem is that nearly every clause of his statement gets the economics backward. 

The Commission’s framing is that Apple faced a free choice and chose spite. The reality is closer to a catch-22. Under the Commission’s interpretation of the Digital Markets Act (DMA), Apple could open its operating system in ways that compromise the security architecture its customers pay for—and that European Union privacy law itself demands—or it could preserve that architecture and ship nothing. 

Apple chose the only option that was both lawful and commercially rational: it withheld the product. When a regulatory regime is structured so that the only safe harbor is nonparticipation, the claim that “nothing prohibits you from introducing new products” is technically true but substantively empty. 

To see why, it helps to recall what actually happened. It also requires asking a question the Commission has studiously avoided: If the open, deeply interoperable, agent-accessible operating system Brussels demands is both commercially viable and consistent with the rest of EU law, why has no one—on any platform, anywhere in the world—ever built one? 

The Architecture of a Catch-22

Apple unveiled Siri AI on June 8 at the Worldwide Developers Conference (WWDC): a rebuilt assistant, powered by Apple Intelligence, capable of carrying out multi-step conversations and acting autonomously across a user’s device. It can search messages and email, complete bookings, make payments, and edit files. In the same announcement, Apple said the feature would not ship on iOS 27 or iPadOS 27 in the European Union, and offered no timeline for its arrival. 

The same assistant will reach European users on Mac and Apple Vision—platforms the DMA does not designate as gatekeepers. That detail alone answers one question: the binding constraint is the regulation, not the technology. 

The sticking point is the DMA’s interoperability obligations. Under what Apple calls an “extreme interpretation” of those rules—but one the Commission appears to have embraced—shipping Siri AI in the EU would require giving rival AI agents equivalent access to the device. Those agents would be able to read messages, control apps, make purchases, and edit files autonomously, without Apple’s intermediating safeguards. 

Apple says it spent 18 months proposing alternatives. Those included a “Trusted System Agent” architecture that would have extended similar capabilities to third-party assistants through a mediated layer, as well as a phased rollout. The Commission rejected those proposals, insisting on openness without the accompanying safeguards. 

Now consider how Thomas Regnier described that history. Apple, he said, was “simply unable to develop interoperability solutions that meet essential EU privacy and security standards” and therefore requested an exemption instead. 

That sentence is worth reading twice, because it may be the most revealing thing any European Union official has said about this dispute. 

The Commission’s own spokesperson is effectively telling us that opening the iPhone to third-party AI agents in the manner the DMA requires could not be done—at least not by the company with arguably the greatest engineering resources and the strongest commercial incentive on Earth to do so—without violating the EU’s own privacy and security standards. 

Regnier presents that conclusion as an indictment of Apple’s diligence. It reads instead as a confession about the legal architecture. 

How to Ban a Product Without Banning It

The decision not to allow Siri AI in the EU is Apple’s and Apple’s only… Absolutely nothing in the DMA prohibits Apple from introducing new products in the EU…

Both halves of that statement are technically true. Together, they describe a menu of options that effectively left Apple with no meaningful choice and no viable path to introducing the product. 

Option 1: Violate the GDPR

Hand autonomous, system-level agents from any third party the keys to users’ messages, payments, photos, and files. 

Quite apart from the commercial damage—Apple’s entire competitive strategy rests on being the privacy-and-security platform, a distinction that even its courtroom adversaries have acknowledged as genuine—this option collides head-on with the European Union’s other flagship digital statute. 

The General Data Protection Regulation (GDPR) requires data protection by design and by default, security measures appropriate to the risk, and data minimization. A platform that grants external agents unmediated authority to read and act on the most sensitive personal data on a device is, at minimum, in serious tension with all three principles. 

Apple would almost certainly bear liability when an agent goes rogue, leaks data, or is compromised. Users, meanwhile, would bear the actual harm. 

Regnier’s briefing effectively concedes that this path is foreclosed. In his own words, the solutions that satisfy the DMA’s openness requirements do not meet “essential EU privacy and security standards.” 

Option 2: Violate the DMA

Launch Siri AI in the EU while denying rival agents equivalent access. 

This option runs directly into Article 6(7), the DMA’s interoperability mandate, which requires a gatekeeper to grant competing services free and effective access to the same operating-system and virtual-assistant features available to its own services. The Commission is currently pursuing a closely analogous case against Alphabet and its Android mobile ecosystem. 

A system-level Siri AI available to Apple’s users but unavailable to rival agents would be a textbook violation, at least under the Commission’s current interpretation of the provision.

The consequences are not trivial. The DMA authorizes fines of up to 10% of worldwide turnover, rising to 20% for repeat infringements. For Apple, that would likely mean penalties measured in the tens of billions of dollars per finding.

Option 3: Don’t Ship

This is the only square on the board where Apple violates neither the DMA, nor the GDPR, nor its customers’ trust. 

It is also exactly what happened on June 8. 

So when Regnier says the decision was “Apple’s and Apple’s only,” he is correct in much the same sense that a person offered a choice among a fine, a lawsuit, and walking away has “chosen” to walk away. The law did not prohibit the product by name. It constructed a set of options in which not introducing the product became the only rational—and indeed the only fully lawful—remaining move. 

Economists do not find this distinction mysterious. A tariff does not “prohibit” imports, either. It makes them uneconomic, and we correctly attribute the resulting absence of goods to the tariff.

The situation here is admittedly less self-referential than the catch-22 described above. It is more a regulatory pincer, with two bodies of law and basic commercial logic squeezing the feasible option set down to “exit.” The lived experience is the same. Every door that appears open closes the moment you walk toward it. 

This is also where Regnier’s speed-limit analogy collapses. A speed limit imposes a single, coherent constraint: drive slower than X. What the Commission has built is the equivalent of two officers standing at the same checkpoint, one ordering the driver to speed up and the other to slow down, while a spokesperson explains that nothing prevents the driver from continuing his journey. 

The DMA demands radical openness. The GDPR demands rigorous protection of personal data. The Commission rejected the intermediary architectures—Apple’s Trusted System Agent and a phased rollout—that might have reconciled the two. 

“EU law is non-negotiable” is an odd boast when EU law points in opposite directions. Non-negotiable in both directions at once simply means impossible. 

If It’s So Easy, Where Is It?

Suppose, though, that the Commission is right and Apple is bluffing. Suppose a commercially viable operating system that gives third-party AI agents deep, autonomous, system-wide access while remaining compliant with European privacy and security law is perfectly feasible, and Apple simply prefers not to build it.

That hypothesis generates a testable prediction. Somewhere in the world, on some platform, someone should have built it.

Operating systems are a fiercely contested, multi-trillion-dollar market spanning mobile devices, desktops, servers, automobiles, and wearables. If radically open agent access were a design consumers wanted at a price firms could profitably supply, its absence everywhere would be a striking anomaly.

Unfortunately for the Commission, that design is nowhere to be found. 

In a recent paper with my colleague Geoffrey Manne, we document what we call the “empty quadrant”: across operating systems, marketplaces, search engines, and technical standards, the radically open platforms with weak control over access and use that regulators often idealize have repeatedly failed to attract users. 

But set that broader pattern aside. The AI-agent version of the question is even more revealing. 

Survey the industry’s leading platforms—including the most open platforms and the AI labs themselves—and a striking regularity emerges. Everyone is converging on precisely the kind of mediated, gated, safeguard-wrapped architecture the Commission rejected when Apple proposed it. 

Google Keeps the Keys, Too

Start with Android, the open mobile operating system. 

Google allows users to designate a third-party assistant as the system default. You can replace Gemini with ChatGPT on an Android phone today. But look closely at what that designation actually provides. It gives a rival assistant a voice interface, not the keys to the device. 

A third-party assistant on Android cannot adjust system settings, control smart-home devices, or act across applications the way Gemini can. It cannot even claim the wake word. The deepest forms of agentic access—reading texts and WhatsApp messages, accessing call logs, analyzing on-screen content, and acting within other apps—remain reserved for Google’s own first-party assistant. 

Android, the platform regulators routinely invoke as the open alternative, ultimately draws much the same first-party/third-party distinction the Commission insists Apple cannot draw. 

Microsoft Built the Thing Apple Proposed

Microsoft offers another useful example. 

Among major operating-system vendors, Microsoft is arguably the firm most aggressively pursuing third-party AI agents. It is explicitly rebuilding Windows as what it calls an “agentic operating system.” Yet the architecture Microsoft is deploying looks remarkably familiar. 

Agents run under dedicated low-privilege accounts inside a contained “Agent Workspace” separated from the user’s primary session. Their access is limited to designated folders, and anything more requires explicit authorization. 

In other words, the company building perhaps the most open agent platform in the world built it around a trusted intermediary layer. Functionally, that is the same basic approach Apple proposed through its Trusted System Agent architecture. 

The punchline almost writes itself: Microsoft is rolling out this cautious, contained, safeguard-heavy system worldwide—except in the European Economic Area. 

Even the AI Labs Use Guardrails

The AI labs behave much the same way. 

OpenAI and Anthropic have more to gain than almost anyone from unfettered access to other companies’ operating systems. Yet their computer-use agents operate in sandboxed cloud browsers and virtual machines, not with unrestricted native privileges on users’ devices. 

Likewise, the interoperability standard the industry is increasingly converging on—the Model Context Protocol—is built around permissioned connections. Applications deliberately expose specific, declared, and revocable capabilities to AI agents. Agents do not simply help themselves to whatever system resources they want. 

The Market Chose Guardrails

In short, mediated access is not an Apple idiosyncrasy. It is the industry’s revealed answer to the question of how AI agents and operating systems should interact. 

Companies with radically different business models, incentives, and competitive positions have independently converged on the same basic architecture. 

This is exactly what our paper‘s evolutionary framework predicts. 

The empty quadrant is not empty because every platform developer in history independently arrived at the same anticompetitive scheme. It is empty because the alternative repeatedly fails market tests. 

On the supply side, a platform that cannot govern what runs on it cannot credibly sell safety and security, and therefore cannot monetize the trust those features create. On the demand side, consumers genuinely value curation. Gatekeeping—in the literal and often beneficial sense—helps exclude bad actors and contain the harms that one malicious agent can impose on everyone else. 

Just as importantly, today’s relatively closed platforms defeated more open rivals before they became dominant. Market power cannot explain the outcome. Market selection can. 

The AI-agent context sharpens these tradeoffs rather than softening them. 

An assistant that can read your email and move your money may be the highest-stakes software ever deployed on a consumer device. The access that makes such systems useful is exactly the access that makes them dangerous in the wrong hands. 

If there were ever a moment when a platform’s judgment about which agents to trust—and through which mediated channels—is doing its most valuable work, it is now.  

The DMA, as currently interpreted, does not engage that judgment. It overrides it. The GDPR then punishes whoever complies. 

Somebody Gets to Choose

Which brings us to the briefing’s most quotable line. According to Thomas Regnier, it is not for Apple to choose which AI tools Europeans get to use. 

Two observations.

First, choosing is what a platform does. Curation—deciding what runs, under what rules, and with what level of access—is the product Apple sells. It is also what hundreds of millions of consumers, including many Europeans, affirmatively choose when they buy an iPhone instead of the more open Android device sitting next to it on the shelf. 

The most important competitive choice consumers make is often between platforms. A curated platform is itself one of the options on that menu. The DMA is methodically narrowing that choice. 

Second, look at the actual outcome. After the Commission’s intervention, which AI tools do Europeans get to use on their iPhones?  

Not Siri AI. Not, in any deeply integrated form, the rival agents the interoperability mandate was supposed to usher in. They do not materialize simply because the incumbent’s product was blocked, any more than Windows XP N—of which just 1,787 copies were sold—produced a browser renaissance. 

In a market where OpenAI, Anthropic, and Google are racing ahead, the DMA’s concrete achievement this week was to remove a significant new competitor from the European field, albeit one powered in part by Google’s models. 

So somebody did end up choosing which AI tools Europeans get to use. It was not Apple. It was not Europeans. 

Fewer firms competing for European users is not more contestability. It is less. 

Nor is this an isolated incident. Google’s Gemini, Meta’s Threads, and Google’s AI Overviews all arrived late in Europe or launched in degraded form for similar reasons. 

The GDPR’s track record points in the same direction. Empirical research finds that it reduced app-market usage and consumer surplus by roughly one-third, while chilling investment in European data-driven ventures. 

Each of these laws imposes costs on its own. The Siri dispute shows what happens when they squeeze from both sides at once. 

When Compliance Means Exit

In our paper, Geoffrey Manne and I close with a deliberately modest set of heuristics for regulating fast-evolving platform ecosystems: first, do no harm; prefer case-specific scrutiny to one-size-fits-all design mandates; preserve firms’ ability to monetize and experiment; and build feedback loops—sunsets, sandboxes, and review clauses—that allow a regime to recognize when it is failing. 

The DMA inverts each of those principles. Its first self-review, published just six weeks before WWDC, nonetheless declared the regime “fit for purpose“—a feedback mechanism apparently incapable of registering the growing pile of products left sitting outside Europe’s door. 

“EU law is non-negotiable” is meant to evoke the rule of law. But the rule of law also requires legal obligations that can actually be satisfied. 

When one EU statute mandates a degree of openness that another EU statute—by the Commission’s own account—renders unlawful, and regulators reject every mediating architecture the regulated firm proposes, inflexibility ceases to be a virtue. It becomes the policy failure. 

The police officer in Regnier’s analogy enforces a limit any driver can obey by easing off the accelerator. The Commission is enforcing a limit that can only be obeyed by leaving the road altogether. 

On June 8, Apple left the road. Per the warnings in Mario Draghi’s report about the costs of Europe’s regulatory thicket, European users are the ones left walking. 

“The decision was Apple’s and Apple’s only,” the Commission says. In the narrowest possible sense, that is true. Apple chose from the options available to it. 

The more important question is who created those options in the first place. The option set was Brussels’ and Brussels’ only. That is the part of the story the midday briefing left out.

The post Brussels’ AI Catch-22: Siri, Define ‘Choice’ appeared first on Truth on the Market.

The pending rewrite of European Union merger-control guidance is the broadest review of the framework in roughly two decades. The Draft Merger Guidelines and accompanying technical-novelties summary seek to move beyond the compartmentalized structure of the 2004 Horizontal Merger Guidelines and 2008 Non-Horizontal Merger Guidelines. That is both a serious and welcome undertaking. Modern transactions rarely fit neatly into inherited doctrinal boxes. Firms compete through R&D pipelines, complementary assets, platforms, distribution networks, procurement relationships, data, manufacturing capabilities, and the ability to scale new products across borders. A unified framework can therefore offer a more coherent approach than a collection of analytical silos.

The draft also reflects a changed political economy. The Commission now speaks in the language of innovation, investment, resilience, sustainability, industrial scale, and global competitiveness. That vocabulary aligns with the European Union’s broader Competitiveness Compass and with concerns highlighted in the Draghi report about Europe’s growth and productivity challenges. It also reflects a recognition that merger control cannot intelligently assess competitive effects by looking only for short-run price increases in narrowly defined markets. Scale can be procompetitive. Integration can accelerate commercialization. Mergers can combine complementary capabilities that no firm could deploy as effectively on its own.

The draft’s promise, however, comes with a significant risk. The same document that expands theories of competitive benefit also expands theories of competitive harm. Innovation, investment, potential competition, entrenchment, ecosystem effects, portfolio effects, buyer power, and labor-market effects all enter a single analytical framework. Each may be relevant in a properly grounded case. Taken together, though, they risk making merger review less predictable unless the final Guidelines insist on concrete causal mechanisms, administrable limiting principles, and symmetrical treatment of harms and benefits.

From a law & economics perspective, the central question is not whether merger analysis should become more dynamic. It should. The real question is whether dynamic analysis can be disciplined enough to reduce error costs rather than simply expand agency discretion.

Dynamic Analysis, Not Dynamic Speculation

Merger enforcement is necessarily forward-looking. The European Commission must compare the world with the merger against a realistic counterfactual—the world likely to emerge without it—under conditions of uncertainty. That simple point has large implications. A false negative can permit durable market power, exclusionary conduct, or diminished innovation rivalry. A false positive can block an efficient transaction, chill investment, reduce exit opportunities for entrepreneurs, and prevent socially valuable assets from moving to better uses. Sound merger guidelines should therefore be written not only to catch bad mergers, but also to avoid condemning good ones.

Daniel Spulber and I make this point with special force in innovation cases. We caution against treating innovation harms as a presumption rather than an empirical question. Innovation competition is not ordinary price competition on a longer clock. It involves uncertain invention, uncertain commercialization, financing constraints, appropriability problems—the difficulty of capturing returns from one’s own innovation—and complementary assets. Some transactions reduce independent R&D rivalry. Others increase the odds that an invention actually reaches consumers. A dynamic approach must be able to recognize both possibilities.

The recent work of University of California, Berkeley, scholar David J. Teece is similarly relevant. Teece argues that antitrust analysis should focus on dynamic competition, innovation, entrepreneurship, and firm capabilities, rather than rely too heavily on static concentration measures. The Commission’s draft adopts much of this vocabulary. But vocabulary is not methodology. A genuinely dynamic framework must ask whether the merger changes the parties’ incentives and capabilities in ways that make stronger future competitive performance more or less likely. It should not simply paste innovation labels onto static structural analysis.

Harvard Law School’s Louis Kaplow underscores a related decision-theoretic problem in merger analysis. In his merger work, including “Out of Market, Out of Mind,” Kaplow criticizes rules that ignore relevant benefits because they arise outside a narrowly defined market. That point matters especially for the Commission’s draft, which contemplates balancing symmetric and asymmetric harms and benefits, including benefits across consumer groups and markets. Modern guidelines should not define the relevant market narrowly to establish harm, then invoke that same narrowness to disregard benefits.

A Theory of Harm for Every Occasion

The draft’s unified architecture is a sensible development. Modern transactions often combine horizontal overlaps, vertical relationships, conglomerate complementarities, and innovation effects. A single framework can help the European Commission assess the overall change in competitive constraints, rather than forcing complex transactions into artificial doctrinal boxes. It may also reduce the risk that benign vertical integration or complementary business relationships are treated as inherently suspect simply because they involve a large firm. The technical-novelties summary appropriately emphasizes that the draft relies on a more refined analytical toolbox for different merger effects, rather than rigid categorization.

Still, a unified toolbox can become an open-ended one. The draft recognizes theories of harm involving the loss of head-to-head competition, investment and expansion competition, innovation competition, potential competition, foreclosure, entrenchment, coordination, access to commercially sensitive information, portfolio effects, and other competitive dimensions. That breadth creates room for nuanced, case-specific economic analysis. It also creates a risk that almost any acquisition by a successful firm can be cast as problematic.

A firm with complementary assets can be accused of leveraging a portfolio advantage. A platform operator can be accused of ecosystem entrenchment. A company with a product roadmap can be accused of eliminating potential competition. A firm with scale can be accused of tipping a market. At some point, an analytical framework broad enough to capture every possible concern risks becoming one that can justify almost any outcome.

The final Guidelines should therefore require more than a plausible story. They should require the Commission to identify a merger-specific mechanism that changes incentives or capabilities in a way that is likely to reduce competition. That is not an artificial hurdle. It is the economic core of merger analysis under the European Union Merger Regulation.

If the concern is foreclosure, the Commission should show ability, incentive, and a likely competitive effect. If the concern is the loss of innovation competition, it should show that the relevant R&D paths are sufficiently substitutable and that the merged firm is likely to internalize or eliminate a meaningful competitive constraint. If the concern is entrenchment, it should show how the transaction makes future entry or expansion materially less likely—not merely that the merged firm will be better equipped to compete.

Dynamic Potential Needs Static Discipline

The draft is right to recognize that market shares are not always enough. In innovation-intensive sectors, market shares may be unavailable, stale, or misleading. A product may not yet be commercialized. An entrant may discipline incumbents through threatened innovation. A firm may possess valuable capabilities that do not show up in current sales. The draft’s new discussion of dynamic competitive potential is therefore useful. It acknowledges that future rivalry may arise from assets, capabilities, and trajectories, not just current output.

But the European Commission should avoid replacing old structural presumptions with a broader, less disciplined market-power inquiry. High margins can reflect innovation rents, temporary scarcity, product differentiation, or risky investment. Network effects—the tendency for a product or service to become more valuable as more people use it—can create consumer value and scale economies, as well as entry barriers. Data advantages may matter in some settings and be overstated in others. Ecosystems may entrench a firm, but they may also reduce transaction costs, improve interoperability, and expand output. David Teece’s dynamic-capabilities account is useful precisely because it asks whether a firm is shielded from innovation and entry, not simply whether it is large, profitable, or successful.

That distinction matters for legal certainty. An effects-based approach should not become a suspicion-based approach. The Commission should make clear that market power cannot be inferred from size alone, profitability alone, or technological sophistication alone. It should be inferred from the ability profitably to degrade competitive variables—price, output, quality, innovation, privacy, resilience, or choice—relative to a realistic counterfactual.

The European Court of Justice’s 2023 CK Telecoms judgment, which specified that the Commission need only show that a proposed merger is “more likely than not” to impose a significant impediment to effective competition based on “an overall assessment” of all relevant factors, illustrates the importance of proof and predictability in EU merger control. A modernized framework should strengthen those disciplines, not dilute them.

Not Every Acquisition Is a Killer Acquisition

The draft’s most important innovation may be its treatment of innovation itself. The Commission recognizes the loss of specific innovation competition, the loss of general innovation competition, and—importantly—an innovation shield under which certain acquisitions involving innovative firms or R&D projects are less likely to raise concerns. That is a welcome acknowledgment that not every acquisition of a small innovative firm by a larger firm is a killer acquisition. Indeed, most are not. Many serve as mechanisms for funding, testing, manufacturing, distributing, or commercializing technologies that might otherwise never reach consumers.

The Commission also deserves credit for recognizing that acquisitions can promote innovation. A small firm may possess a promising invention but lack regulatory expertise, distribution channels, complementary intellectual property, manufacturing scale, a sales force, or the capital needed to survive the long journey from prototype to market. A larger firm may provide precisely those assets.

Of course, the opposite scenario can occur. An incumbent may acquire a target to eliminate a competitive threat. Merger control must distinguish between those cases, not presume one from the other. Recent work on “Killer Acquisitions: Evidence from European Merger Cases,” along with the broader debate over acquisitions of nascent competitors, underscores why the question matters. It also illustrates why careful evidence is indispensable.

The draft’s innovation shield, however, appears too narrow. Conditions tied to market-share ceilings, the existence of alternative R&D projects, or the acquirer’s status as the largest firm or a gatekeeper may be administrable. They also risk misclassifying procompetitive transactions. The most socially valuable acquisitions may be precisely those in which a larger firm possesses the complementary assets needed to scale a breakthrough innovation. Requiring several comparable alternative R&D paths may likewise fit poorly in markets where innovation is lumpy, heterogeneous, and organized around differentiated technological bets. A safe harbor that systematically excludes leading firms may miss transactions that increase expected innovation output.

A better innovation shield would focus on the economics of commercialization. It would ask whether the target has a realistic independent path to market; whether the acquirer contributes complementary assets; whether the transaction accelerates innovation or increases the likelihood that it reaches consumers; whether alternative innovators remain credible; and whether the parties’ documents and market evidence demonstrate a merger-specific innovation benefit.

The shield should not immunize all acquisitions by incumbents. It should, however, provide meaningful protection for innovation-enhancing transactions, particularly where the Commission’s theory of harm is speculative and the claimed benefits are credible, even if difficult to quantify.

Potential Competition Is Not Potential Everything

The draft’s treatment of potential competition and entrenchment is equally important. Potential-competition doctrine has a legitimate role when a firm that likely would have entered a market and constrained incumbent firms is acquired by one of them. Herbert Hovenkamp’s 2024 article on potential competition argues for evidence-based analysis rather than speculation about future innovation and entry. It also emphasizes the need to weigh potential harms against the efficiencies that mergers may generate when they involve firms that are not yet direct competitors.

The European Commission’s decisional practice has long considered likely future rivalry. But potential competition is easy to overextend. Many firms are potential entrants into many adjacent markets in some abstract sense. That is not enough. The Commission should require evidence that entry was probable, timely, and competitively significant absent the merger.

Entrenchment theories require even greater caution. The draft’s focus on ecosystems, adjacent markets, and dominant positions may capture genuine exclusionary concerns. It also risks treating improved capabilities as anticompetitive simply because they make the merged firm harder to beat. Competition law protects the competitive process, not competitors from efficient rivalry. A merger that combines complementary assets, improves product quality, or lowers costs may harm rivals precisely because it benefits consumers. That is not a significant impediment to effective competition. It is competition working as intended.

The final Guidelines should therefore make clear that entrenchment theories require proof of a likely reduction in contestability—the ability of rivals and potential entrants to challenge the merged firm. The relevant question is whether the transaction makes entry, expansion, innovation, interoperability, switching, or multi-homing materially more difficult in a way that harms consumer welfare or the competitive process.

It should not be enough to point to a stronger ecosystem, a broader portfolio, or more efficient integration. Those features may be relevant evidence in some cases. They are not harms in themselves.

Benefits Should Not Need Better Evidence Than Harms

The draft’s expanded treatment of efficiencies is one of its strongest features. It recognizes direct efficiencies, including economies of scale and scope, elimination of double marginalization, complementary technology integration, resilience, sustainability, and effects on incentives and capabilities to invest and improve quality. Early practitioner commentary also observes that the draft creates a more visible channel for merger parties to present affirmative theories of benefit.

That is an improvement over a framework that treats efficiencies as an afterthought. Many valuable mergers do not primarily reduce marginal costs in ways that are easy to quantify. They combine capabilities. They redeploy assets. They enable scaled R&D. They integrate complements. They reduce supply-chain risk. They allow products to launch faster, with better quality, or at larger scale. These are real economic benefits, even when they resist precise measurement. The draft’s recognition of dynamic efficiencies is therefore a significant step toward the kind of dynamic merger analysis advocated by Daniel Spulber and myself, David Teece, and Louis Kaplow.

The problem is evidentiary asymmetry. The draft continues to require benefits to be verifiable, merger-specific, timely, and beneficial to consumers. Those requirements are sensible. But if they apply more stringently to benefits than to harms, they will bias the analysis against efficient mergers. Innovation harms are often probabilistic and qualitative. Innovation benefits are, too. The Commission should not accept a broad narrative of future innovation harm while demanding near-certainty from merging parties about future innovation benefits. Symmetry is essential.

The final Guidelines should expressly state that dynamic harms and dynamic benefits will be evaluated under comparable evidentiary standards. If internal documents, business plans, expert evidence, and market testimony can support an innovation-harm theory, the same evidence should be able to support an innovation-benefit theory. If uncertainty reduces the weight of claimed benefits, it should also reduce the weight of claimed harms.

A decision-theoretic framework should compare expected harms and expected benefits, discounted for probability and timing. It should not demand proof of benefits with a rigor it does not apply to harms.

The Map Is Not the Territory

One of the draft’s most important technical innovations is its willingness to consider balancing across different consumer groups and markets. This is where Kaplow’s “Out of Market, Out of Mind” is especially relevant. If a merger produces a small harm in one narrowly defined market and a large benefit in another, a rule that ignores the benefit simply because it falls outside the first market can be economically incoherent. The welfare consequences of a transaction do not depend on market boundaries drawn for analytical or litigation purposes.

That does not mean merger analysis should credit every broadly defined social benefit. Competition law should remain competition law. But out-of-market benefits should be cognizable when they are concrete, merger-specific, and connected to consumer welfare, output, quality, innovation, market contestability, resilience, or the competitive process. If a merger improves supply reliability for customers in adjacent markets, accelerates innovation in complementary products, or lowers costs for a broad class of users, those effects should not be disregarded merely because they arise outside the narrow market where a localized concern is alleged.

Resilience and sustainability require the same discipline. The draft properly recognizes them as non-price dimensions of competition. Customers may value secure supply, cyber resilience, decarbonization, reduced outage risk, or greater product durability. These can be quality dimensions on which firms compete. The Commission should not, however, convert resilience or sustainability into free-floating industrial-policy exceptions. The better approach is to ask whether the claimed benefit improves the competitive offering available to consumers or trading partners and whether it is sufficiently merger-specific.

This framework also helps avoid internal inconsistency. If the Commission worries that consolidation may reduce resilience by concentrating supply, it should also recognize that some consolidation may increase resilience by combining complementary capacity, improving redundancy, strengthening balance sheets, or enabling investment in critical infrastructure. The same principle applies to sustainability.

Resilience can be a harm or a benefit. Sustainability can be a harm or a benefit. The Guidelines should not presume the answer. They should require evidence.

Five Ways to Make the Guidelines Better

The final Guidelines can preserve the draft’s strongest features while improving administrability through a handful of targeted changes.

First, every theory of harm should require a concrete causal mechanism. Labels such as ecosystem, portfolio, gatekeeper, pipeline, potential competitor, or dominant firm should not substitute for evidence that the merger is likely to change incentives or capabilities in a way that harms competition.

Second, the innovation shield should be broadened. It should protect not only transactions that fall below structural thresholds, but also transactions in which the target lacks a realistic path to commercialization, the acquirer contributes complementary assets, and the expected effect is to accelerate innovation or increase innovation output. The Commission should also recognize that acquisition markets can create ex ante incentives to innovate. Excessively uncertain merger review may discourage venture investment by weakening the exit opportunities on which many start-ups depend.

Third, the Commission should evaluate dynamic harms and dynamic efficiencies symmetrically. In innovation cases, both sides of the ledger are uncertain. The Guidelines should not permit speculative harms to outweigh plausible, evidence-backed benefits simply because the benefits are harder to quantify. This is one area where the academic-research pieces cited above converge: merger analysis should be comparative, probabilistic, and dynamic.

Fourth, out-of-market benefits should remain part of the balancing framework. Kaplow’s warning remains highly relevant. Ignoring benefits because they arise outside a market boundary can turn market definition into an arbitrary welfare filter. The draft’s openness to cross-market balancing is a strength and should be retained, subject to clear requirements for evidentiary reliability and competition-relevant benefits.

Fifth, the final text should clearly distinguish harm to competitors from harm to competition. That distinction is familiar, but it becomes even more important in innovation-driven and platform markets. Integration, scale, interoperability, and complementary assets may disadvantage rivals because they produce better products. That is not a competition problem unless rivals are excluded in a way that reduces contestability, innovation, output, quality, or consumer welfare.

The Difference Between Dynamism and Discretion

The Commission’s draft Merger Guidelines deserve serious praise. They recognize that merger analysis must account for innovation, dynamic competition, resilience, sustainability, and scale. They provide a unified structure, articulate theories of benefit as well as theories of harm, and acknowledge that procompetitive mergers can strengthen the internal market. In these respects, the draft is more modern than the guidance it would replace.

The challenge is to match modern language with disciplined method. A dynamic framework should not merely add new reasons to intervene. It should improve the Commission’s ability to distinguish anticompetitive consolidation from procompetitive reorganization. It should reduce false negatives and false positives. It should give businesses clearer guidance while preserving enforcement against mergers likely to impede effective competition.

If the final Guidelines incorporate stronger limiting principles, broaden the innovation shield, apply symmetrical evidentiary standards, and preserve economically coherent balancing of benefits and harms, they could make an important contribution to global merger policy. If they do not, the draft’s breadth may undercut its promise by increasing uncertainty and chilling innovation-enhancing transactions.

Dynamic merger analysis is worth doing—but only if it is disciplined enough to tell better mergers from worse ones.

The post Brussels Reboots Merger Control. Now Debug the Discretion. appeared first on Truth on the Market.

Last week, I wrote about the Federal Trade Commission’s (FTC’s) appeal in FTC v. Meta Platforms. The appeal seems to me a bad idea.

I quarreled with several points raised by the commission in its opening brief. My main concerns, however, were these: The FTC failed to show that the conduct at issue—two long-ago consummated mergers—caused ongoing harm to competition or consumers. It was also hard to envision a remedy that would benefit competition or consumers if the FTC ultimately prevailed on liability (contingent, of course, on reversal and remand and, then, a new liability decision). 

More broadly, it seemed a serious waste of limited agency resources to appear before the U.S. Court of Appeals for the D.C. Circuit in late 2026 to argue for reversal and remand, so that the U.S. District Court for the District of Columbia could reconsider its finding against liability for acquisitions that the FTC investigated, reviewed, and allowed to close without complaint in 2012 and 2014. 

In brief, I argued that the FTC exercised its enforcement discretion poorly at several decision points, and perhaps in between them: in 2020, when it rushed to file a weak complaint in the waning days of the first Trump administration; in 2021, when it filed amended complaints following dismissal of the 2020 complaint; in 2025, when it took a flawed case to trial; and now, in 2026, with this appeal. 

This post is not about that case. It does, however, share some background concerns with my recent little (or exceedingly long) missive.

The First Rule of Holes

This continues the theme of rational quitting. It’s about the productive application of the sunk-cost fallacy—or, to be precise, the productive application of knowledge of the sunk-cost fallacy. It’s about … enough already. Dayenu!

If you are digging yourself deeper and deeper into a hole, stop digging. Climb out and find another hobby. If you are howling in pain, having shot yourself in the foot, don’t point the gun at your other foot. Head hurts? Stop beating it against the wall. Walk away from your “investment.” 

In times of yore—May 2025—I had a post about the Robinson-Patman Act (RPA) with the questioning, if not questionable, title, “RIP RPA?” Did I inquire hopefully? Rhetorically? Does it matter? 

The post was prompted by an eminently sensible exercise of enforcement discretion by the FTC under Chairman Andrew Ferguson’s leadership: the May 2025  vote to withdraw one of the two RPA cases the agency had launched, hurriedly, in the waning days of Lina M. Khan’s time at the helm. 

To recap, on Jan. 17, 2025, the FTC announced that it had:

sued PepsiCo, Inc. (Pepsi) alleging that the second-largest food company in the world has engaged in illegal price discrimination by providing one customer—a large, big box retailer—with unfair pricing advantages, while raising prices for competing retailers and customers. 

The vote took place three days before President Donald Trump’s second inauguration and, as it happened, three days before the new-again president would designate Andrew Ferguson chair of the FTC. 

Dissenting from the complaint—or, technically, from the vote to authorize FTC staff to file the complaint—then-Commissioner Melissa Holyoak, now U.S. Attorney for the District of Utah,  called it “the worst case” she had seen as a member of the commission: 

Today’s Complaint against Pepsi is wholly deficient, not only because the pleadings fail to state a claim, but because the Majority rushed the case out the door before it had evidence to support the allegations. I am astounded that the Majority has such little regard for our staff that it is willing to send them to court like a lamb to the slaughter. 

Ferguson was no kinder to his colleagues’ vote to proceed:

The gaping holes in the evidence that Commission staff collected in its limited investigation make it impossible to determine whether the defendant, PepsiCo, Inc. (“Pepsi”), has broken the law. The Commission majority sues Pepsi nonetheless. The paucity of evidence is not a problem for the majority, because the law is beside the point.

Other than that, Mr. Ferguson, how did you like the complaint? 

Still Digging

The FTC did not, however, vote to drop the other RPA case, FTC v. Southern Glazer’s Wine and Spirits. Why not? 

It’s not that Ferguson thought it was a good case. Like Pepsi, Southern Glazer’s was brought on a party-line vote, with both Ferguson and Holyoak dissenting. (By the way, I recommend both dissents, but especially Holyoak’s thoughtful and scholarly 83-page opinion, at least for readers who want a deeper dive into both the RPA and the FTC’s case. For more from her perspective, see the excellent article she co-authored with Christopher Mufarrige in this year’s Antitrust Law Journal: “From Protecting Competitors to Protecting Competition: The Past, Present, and Future of the Robinson-Patman Act.”) 

What was the case about, you ask, having been blessed with the power of forgetting? It’s right there at the top of the complaint: 

For years, Southern has violated the Robinson-Patman Act by selling wine and spirits to small, independent “mom and pop” businesses at prices that are drastically higher than the prices Southern charges large national and regional chains. Southern’s discriminatory pricing practices have victimized independent and family-owned neighborhood grocery stores, local convenience stores, and other independent retailers across the country.

Volume discounts? And tigers and bears? Oh my!

For the economists in the audience, “drastically higher” and “victimized” are spin, not legal terms of art. Here, “drastically higher” means higher (plus spin), and “victimized” means relatively disadvantaged (plus spin).

The allegation of higher prices was probably true. There were instances in which large chains received discounts unavailable to smaller buyers making smaller and/or less frequent purchases. The allegation that smaller competitors were thereby harmed? Debatable.

How about harm to competition and consumers? Nope. Not obviously, and not on the face of the complaint, even assuming the allegations to be true and construing them in the light most favorable to the FTC. The complaint included some conclusory mumbling about consumer benefits flowing from enhanced choice through the ad hoc protection of certain smaller retailers, but that gets it nowhere.

Neither Ferguson nor Holyoak was generally opposed to the idea that the FTC might bring an RPA case under the right facts and circumstances. To the contrary, when the FTC first brought the Southern Glazer’s case, Ferguson argued that “[t]he Executive Branch should not categorically and publicly refuse to enforce laws that Congress has passed and the President has signed.” Holyoak made a similar point in her own dissent. Fair enough, in the abstract.

Neither, however, suggested that the commission should bring bad RPA cases, and both were clear that this was one. Ferguson’s dissent identified both legal and factual infirmities. In short, he did “not believe the Commission is likely to prevail even under the traditional, protectionist understanding of the Robinson-Patman Act.”

In other words, the facts alleged seemed a poor fit for the complaint’s theory of liability for “second-line” price discrimination, and that theory itself might prove controversial. He also raised constitutional concerns.

More than that, while Ferguson argued strenuously that the FTC should, as a general matter, enforce the laws Congress has charged it with enforcing—RPA included—he also argued that “the Commission must soundly exercise discretion about when to enforce a law,” and that, in bringing the Southern Glazer’s case, the FTC “exercise[d] its discretion poorly.”

Ferguson argued that the case: 

may protect the disfavored retailers who allegedly paid higher input prices than their competitors, but it may do so by raising prices for millions of hardworking Americans.

As Holyoak explained:

Forcing Southern Glazer’s to discontinue supplier discounts—without evidence of harm to competition and consumers—for high-volume retailers may actually harm both intrabrand and interbrand competition. Without volume discounts, retailers supplied by Southern Glazer’s will have less incentive to compete against each other for sales of the same product (i.e., intrabrand competition). And at the same time, wine and spirit suppliers would cut back promotional efforts that drive competition with other wine and spirit suppliers (i.e., interbrand competition). Taken together, by condemning Southern Glazer’s pricing practices without evidence of harm to competition, the Commission ignores the Supreme Court’s most recent Robinson-Patman Act holding in Volvo, along with Supreme Court precedent that cautions against interpretations of the Act that “extend beyond the prohibitions of the Act and, in doing so, help give rise to a price uniformity and rigidity in open conflict with the purposes of other antitrust legislation.” (internal citations omitted)

So much for consumer welfare. And for competition.

As for whether meaningful head-to-head competition among retailers is really at stake … look. I’m sorry. I don’t know who posted this deposition clip, but it’s so cringe—Curb Your Enthusiasm-level cringe—that I’m not entirely comfortable even posting the link. It does not inspire confidence in the expert enforcement agency at which I long, and proudly, toiled.

The Bipartisan Case for Letting Go

Would Congress do well to repeal the RPA? Yes, I say, as do many, many others.

Geoffrey Manne discussed “misguided calls to reinvigorate the Robinson-Patman Act” in congressional testimony in 2022; and here’s Brian Albrecht at Truth on the Market last year. But this isn’t just about the International Center for Law & Economics’ (ICLE) or my own idiosyncratic views. Not remotely. 

See, for example, this paper by Bruce Kobayashi, former director of the FTC’s Bureau of Economics, and Tim Muris, former FTC chairman; two papers, among others, by Herbert Hovenkamp—this one, arguing that Congress “fell to interest group politics” when enacting “the Robinson-Patman Act, which subsidizes smaller businesses at consumers’ expense,” and this one, calling the RPA “irritating to almost anyone who is serious about antitrust”; this article by D. Daniel Sokol; this one by Judge Richard Posner; and this l’il old 320-page report issued by the U.S. Justice Department (DOJ) way back in 1977. 

And if I might quote the 2007 report of the bipartisan and congressionally mandated Antitrust Modernization Commission:

The Commission recommends that Congress finally repeal the Robinson-Patman Act (RPA). This law, enacted in 1936, appears antithetical to core antitrust principles. Its repeal or substantial overhaul has been recommended in three prior reports, in 1955, 1969, and 1977. That is because the RPA protects competitors over competition and punishes the very price discounting and innovation in distribution methods that the antitrust laws otherwise encourage. At the same time, it is not clear that the RPA actually effectively protects the small business constituents that it was meant to benefit.

Price discrimination—or “price differentiation,” “differential pricing,” etc.—has been studied at least since A.C. Pigou’s and Frank Ramsey’s landmark work in the 1920s, and more recently by Richard Schmalensee, Hal Varian, and Varian and Allesandro Acquisti, among others. 

On the one hand, it is well understood that, under certain circumstances, price discrimination can be anticompetitive. But that is not to say that it must be. It need not. Or that it tends to be. Nope.

It is also well understood that price discrimination can increase welfare and, moreover, that it can—and often does—reduce prices for price-sensitive or budget-constrained consumers. For a recent, accessible discussion of when price discrimination might or might not increase the gains from trade, see this piece by my ICLE colleague Brian Albrecht.

The case for repeal never turned on a claim that price discrimination could not possibly be anticompetitive. The questions, rather, concerned the mismatch between the RPA and the cases in which price differentiation might be anticompetitive, rather than procompetitive or benign.

That mismatch has been especially conspicuous in some lower courts, and especially in RPA cases brought by private plaintiffs. Consequently, we have been stuck with excessive incentives for plaintiffs—and the plaintiffs’ bar—to bring bad RPA cases and, at best, an excess of false positives.

Add to that a pointed question: Where are the demonstrable cases of anticompetitive price discrimination that could not be addressed through other provisions of the antitrust laws, including Section 1 or Section 2 of the Sherman Act? For most of us in antitrust law and economics, that is enough to recommend repeal. More than enough.

In brief, the RPA is unnecessary in principle, and it tends to be detrimental in practice.

Where’s the Good Case?

OK, the RPA has not been repealed, so there’s that.

As Holyoak and Ferguson noted in their dissents in Southern Glazer’s, there remains the question of where, exactly, we might find a good RPA case. That is partly a question of facts. It is also partly a question of law, including the Supreme Court’s RPA jurisprudence.

Ferguson and Holyoak found both the facts and the law wanting in the FTC’s Southern Glazer’s case. They were right about that.

Part of the problem is that the RPA’s “progressive” cheerleaders sometimes—or often—prefer to ignore the Supreme Court’s holding in Volvo, while selectively embracing dicta from the same opinion. They are similarly inclined to overlook Brooke Group, where the Court cautioned against interpretations and enforcement of the RPA that are inconsistent “with broader policies of the antitrust laws,” to the detriment of competition and consumers.  

Step Away from the Shovel

The Southern Glazer’s case is a case, not a statute or binding precedent. It involves particular allegations under particular facts and circumstances. If it is a bad case—and it is, and they know it—the FTC can drop it without any general repudiation of the RPA or the agency’s statutory mission. 

That leaves the proper exercise of the FTC’s enforcement discretion. The puzzle here is that Chairman Ferguson has been right about this case from the beginning. He was right that the FTC is unlikely to “prevail even under the traditional, protectionist understanding of the Robinson-Patman Act,” and he was right that the FTC “exercises its discretion poorly” in bringing the Southern Glazer’s case. 

He has also been right—in Pepsi and elsewhere—that the proper exercise of the commission’s enforcement discretion really matters. It matters to avoiding harmful results. It matters to maintaining the agency’s integrity and reputation. It matters to the agency’s ability to marshal its limited resources in service of competition and consumers. Not incidentally, it matters to the district courts, whose dockets are crowded enough without expert enforcement agencies bringing obviously bad cases. 

So why didn’t the commission vote to drop Southern Glazer’s when it voted to drop Pepsi? The vote to drop the Pepsi case was 3-0: Ferguson, Holyoak, and Meador against nobody. Presumably, a motion to drop Southern Glazer’s would have prevailed 3-0 or, at worst, 2-1.

The answer: Beats me. I could guess at this or that, but I really don’t know.

One more thing: An old saw says hard cases make bad law. This is not a hard case. Or it shouldn’t be. And there’s the rub: Easy cases can make bad law, too.

But there’s still time. You, FTC, can drop it. You’ve shown that you can drop a bad case. It is not too late to save a perfectly good foot.

Better to limp away than fall on your face. Don’t ask me how I know.

The post The FTC’s Robinson-Patman Hangover appeared first on Truth on the Market.

The United Kingdom’s Competition and Markets Authority has opened a formal review of the deal, giving itself until Aug. 7 to decide whether to clear the transaction or launch a deeper investigation. California, New York, and possibly other states are preparing a lawsuit to block it. The European Commission is conducting its own review, while leaks suggest the U.S. Justice Department (DOJ) may ultimately approve the merger. 

That much is familiar. Large mergers often attract scrutiny from multiple regulators at once. This particular battle has been brewing for more than a year and has taken several unexpected turns, including Netflix’s failed attempt to acquire WBD. 

At first, the focus was streaming. Critics warned that combining Paramount+ and HBO Max would create a video-streaming giant. That theory has quietly faded. Even after the merger, the combined company would rank only fourth among streaming services, behind Netflix, Disney+, and Amazon Prime Video, which together account for roughly 65% of subscription viewers. Kristian Stout and Ben Sperry analyzed the viewing data and found the merged firm would still trail YouTube in total TV viewing time. A company struggling to achieve scale is a difficult monopolist to imagine. 

The debate has since moved upstream, from streaming platforms to the studios that produce movies and television shows, and to the people who work in them. That is where the Writers Guild, the Teamsters, and state attorneys general have concentrated their fire. It is also where the stronger antitrust argument may lie—maybe. 

To see why, it helps to remember why WBD is for sale in the first place. As I wrote when the company announced plans to break itself apart, it is carrying nearly $38 billion in debt from two previous mergers while its cable networks generate shrinking cash flows. This is a company searching for scale and a cleaner balance sheet, not one so dominant that it can afford to starve Hollywood of work. 

The Five-to-Four Mirage

Hollywood’s “Big Five” studios—Disney, Universal, Warner Bros., Paramount, and Sony—accounted for roughly three-quarters of the domestic box office in 2024. A Paramount-WBD merger would reduce that group from five major studios to four. The combined firm would control less than one-quarter of theatrical revenue, making it the second-largest distributor behind Disney. 

For critics, that five-to-four consolidation is the entire case. It shouldn’t be.

I ran the same concentration calculations antitrust agencies would run. Using 2024 box-office shares, a combined Paramount-WBD would produce a post-merger Herfindahl-Hirschman Index (HHI)—a standard measure of market concentration—of roughly 1,790 in theatrical distribution. Under the DOJ and Federal Trade Commission’s (FTC) 2023 Merger Guidelines, markets generally become “highly concentrated” only when the HHI exceeds 1,800, the threshold at which the agencies’ structural presumption against a merger kicks in. By the agencies’ own yardstick, this deal falls just short of that line, although it comes close. 

The result is even less dramatic when viewed over a longer period. Using 2014 box-office shares, the HHI would be only about 1,540. 

That points to a deeper problem with treating box-office shares as a reliable proxy for market power: they do not stay put for long. A studio’s annual share often rises or falls on the strength of a few tentpole releases. Warner Bros., for example, jumped from roughly 12% of the box office in 2024 to about 18% in 2025, driven by the performance of its film slate rather than any meaningful change in market structure. Numbers that swing that sharply from year to year provide a shaky foundation for judging durable market power. 

The Best Case Against the Deal

The labor case is more serious, and it deserves a serious answer rather than a wave-off.

The Writers Guild of America (WGA) came out against the deal in October, calling it a “disaster for writers” and pledging to help regulators block it. The International Brotherhood of Teamsters, which represents thousands of drivers, location scouts, and other below-the-line workers, filed a white paper urging the DOJ to sue unless the parties commit to enforceable production and jobs guarantees. 

Their theory is monopsony, or buyer power. Most merger analysis asks whether a combined firm can raise prices for consumers. Monopsony asks the mirror-image question: whether the merged firm can push down what it pays suppliers. Here, those suppliers are the writers, directors, actors, and other creative workers who sell their labor to studios. 

The 2023 Merger Guidelines put monopsony squarely on the table, and the logic is intuitive. If a screenwriter can shop a script to five major buyers, those buyers bid against one another. Take one buyer away, and the writer’s next-best option gets worse. 

This is not a fringe theory. When Penguin Random House tried to buy Simon & Schuster, the DOJ blocked the deal in 2022, and Judge Florence Pan agreed that combining two of the “Big Five” publishers would harm competition to acquire “anticipated top-selling books.” The relevant market was not all books. It was the narrower segment where a few deep-pocketed buyers compete for marquee work. Apply that lens to Hollywood, and one can see a plausible claim about the market for A-list talent. 

But a caution about monopsony: Buyer-power claims, especially in labor markets, are among the hardest in antitrust to win. A plaintiff must define a market for a particular kind of worker, show that the merger meaningfully shrinks that worker’s set of buyers, and trace lower pay to the deal rather than to the many other forces moving wages at the same time. 

Penguin Random House cleared that bar, but it stands out because so few cases do. That is why heavy reliance on a labor theory is often a tell. When the fight over a merger like this settles on monopsony rather than prices or output in the studio and streaming markets, it usually signals that the conventional output-market case has already come up short. 

So does the labor claim hold here? Three things cut against it. 

First, a Paramount-WBD combination presents a milder buyer-power problem than the Netflix-WBD deal everyone was modeling last winter. Paramount runs a smaller content operation than Netflix, so the combined firm’s pull in the talent market would be more modest than the scarier scenarios assumed. 

Second, writer and director pay has been under pressure for a decade, and mergers are only a small part of that story. The shift from network television to streaming moved creators away from backend profit participation and toward upfront fees, in part because streaming services do not disclose the viewership data that backend deals depend on. Residual formulas come from industry-wide bargaining. Production has drifted offshore in pursuit of tax credits. Moving from five studios to four touches none of those forces. Blaming consolidation for trends that long predate this deal confuses the diagnosis. 

Third, the closest natural experiment points the other way. The unions cite Disney’s 2019 purchase of 20th Century Fox as the cautionary tale. But the story is messier than that. Disney trimmed its film slate before the Fox deal closed. Then the 2020 pandemic gutted theatrical releases across the entire industry. Meanwhile, Disney’s content spending has swung significantly over the past few years—starting around $24 billion in 2019, ballooning past $30 billion in 2022 as Disney built its streaming platforms, and settling back into the low-to-mid $20 billions for 2025. . 

A streaming-hungry company also has reason to make more films, not fewer. Survey work by Roku and National Research Group found that the heaviest moviegoers are also the heaviest streamers. A strong theatrical run builds the awareness that drives later streaming demand. Movies fill the catalog and pull in subscribers. A firm trying to catch Netflix has every reason to keep its studios busy 

When Rivals Discover Antitrust

It also pays to ask who is pushing hardest against the deal.

Netflix bid for WBD, lost to Paramount in February, and has not gone quietly. In a June 5 letter to the DOJ, Paramount accused its former rival of running a “scorched-earth campaign” to derail the transaction. 

A losing bidder amplifying unions’ genuine anxieties is a familiar play. The workers’ concerns are real. Runaway production and shrinking incomes are not imagined problems. But when a rival tries to sink a competitor’s merger, self-interest often plays at least as large a role as the public interest. 

For its part, Paramount has committed to releasing at least 30 films a year in theaters and has told analysts that most projected merger synergies would come from technology and back-office consolidation, not from cuts to production crews 

Those promises deserve scrutiny. Matthew Belloni, who covers Hollywood’s business side at Puck and hosts The Town podcast, has been among the deal’s most persistent skeptics. He has repeatedly questioned how many jobs the merger will ultimately eliminate and whether Paramount’s synergy projections can really spare production workers.

That skepticism is warranted, and Paramount’s commitments deserve verification. At the same time, they cannot be dismissed as mere cheap talk. If the combined company hopes to catch the streaming leaders, it needs more content, not less. Expanding output is not just a public-relations talking point; it is the core of the business strategy. 

Show Me the Harm

None of this means regulators should rubber-stamp the deal. If they can show, with evidence rather than a studio headcount, that the merger would suppress pay in a defined market for top creative talent, the right response would be a tailored condition—such as a fixed-term, monitored commitment on production volume or licensing. 

That approach tracks what a group of former federal antitrust enforcers urged Attorney General Pam Bondi to do this winter: judge the deal by its effects on consumers and workers, favor remedies over outright prohibition, and resist structural presumptions untethered from real-world harm. 

A five-to-four count in a business where one or two blockbusters can move a studio’s market share is not, by itself, a reason to stop this merger. The case against it must rest on demonstrated harm to consumers or to the people who make the content. 

So far, the loudest voices have supplied a theory and a grievance. They still owe regulators the proof.

The post Paramount’s Mission: Impossible Antitrust Case appeared first on Truth on the Market.

The European Commission’s newly adopted proposal to reallocate the 2 GHz mobile-satellite-service (MSS) band—a slice of radio spectrum used for satellite communications—would reserve most of that spectrum for European operators. The same proposal would limit non-EU militaries and startups, including both EU and non-EU firms, to just one-third of the available and highly desirable band. 

The move reflects a broader push for “tech sovereignty,” the idea that Europe should reduce dependence on foreign technology providers in strategically important sectors. That concern now spans data centers, payments processing, telecommunications, and, increasingly, space. But the EU’s 2 GHz plan is industrial policy dressed up as tech sovereignty. It assumes Europe can create a globally competitive satellite champion by reserving critical inputs for favored firms and denying them to more efficient rivals. 

That is bad economic policy for what is inherently a global communications system. For anyone who remembers the first wave of digital-sovereignty fights two decades ago, it is also eerily familiar. 

Spectrum by Passport

The European Commission has adopted a draft regulation to create an EU-level licensing process for the 2 GHz MSS band, which U.S. operators Viasat and EchoStar have held since 2009 under licenses set to expire in May 2027. Under the  proposal, one-third of the band would be reserved for government, security, and defense uses, supplied by an EU operator and folded into the bloc’s IRIS² constellation. Of the remaining two-thirds reserved for commercial use, half would be limited to EU operators, leaving only one-third genuinely open to both EU and non-EU firms. The incumbents’ licenses would be extended by two years, but their rights would be frozen and nontransferable in the interim. 

Although framed as a security measure, the proposal effectively allocates market share by nationality. The defense set-aside may be defensible on its own terms. Reserving a third of the commercial band for “EU operators entering the market” is not.

In the short run, scarce spectrum would sit idle by administrative fiat. In the long run, firms that do not yet exist—and may never prove commercially viable—would receive a subsidy in the form of excluding firms that already do. 

When Licenses Become Suggestions

Much of the commentary on this proposal has focused on its implications for Starlink, which needs access to the 2 GHz band. But the larger problem is the blow it strikes against regulatory predictability—the basic expectation that governments will not rewrite the rules after companies have invested under them. 

EchoStar and Viasat have held these 2 GHz rights for more than 15 years under the shared understanding that an EU-harmonized allocation meant what it said. SpaceX, meanwhile, agreed to pay roughly $17 billion for EchoStar’s 2 GHz MSS holdings—the AWS-4 and H-block licenses—on the assumption that those rights would remain usable through the current license term, which expires in 2027. Reallocating the band by nationality at renewal, while freezing transfers in the meantime, pulls the rug out from under exactly the sort of long-term, capital-intensive investment that spectrum policy is supposed to encourage. 

The irony is that Europe says it wants more investment in space while signaling that spectrum rights can be effectively renationalized whenever political priorities favor a domestic champion—or a hoped-for domestic champion. Capital is not so trusting. A regime that treats vested, purchased rights as contingent on political discretion will be priced accordingly by investors, including those backing European firms. 

When political priorities displace legitimate investment expectations, capital goes elsewhere. Firms focus on more predictable markets, and European consumers bear the opportunity costs. As the International Center for Law & Economics (ICLE) has argued in its comments on the EU Space Act, discriminatory measures of this sort function as nontariff barriers: They shield incumbents—or, here, prospective entrants—from competition while raising the cost of capital across the market. The European Union has run this play before. Its proposed “fair share” levy on content providers reflected the same instinct in a different guise.

You Can’t Nationalize Orbit

The deeper problem is conceptual. A low-Earth-orbit (LEO) satellite constellation is not a national asset that happens to cross borders. It is a global network whose economic logic depends on scale and scope. In plain English: These systems work because the same costly infrastructure can serve many users in many places at once. 

The same satellites can serve users across multiple continents simultaneously. The marginal cost of extending service to one more country is low because the fixed costs are spread across a global customer base. Direct-to-device service, which would allow ordinary mobile phones to connect directly to satellites, is valuable for precisely this reason. A handset that works seamlessly across borders is far more useful than one that does not. 

Fragmenting that system along national lines sacrifices those scale economies. If every bloc reserves “its” spectrum for “its” champions, the likely result is not a thriving ecosystem of national constellations. It is higher capacity costs, slower deployment, and balkanized coverage that leaves consumers worse off—including European consumers. 

The 2 GHz proposal also undermines one of satellite broadband’s greatest promises: making rural and remote coverage economical for the first time. That promise depends on spreading fixed costs across the widest feasible market. Divide the market into national or regional preserves, and the economics get worse quickly. 

Beneath the sovereignty rhetoric lies a serious concern, and it deserves a serious response. Europe worries about dependence. The war in Ukraine showed how reliance on a single foreign-owned, privately controlled constellation can become a strategic vulnerability if the operator’s interests—or whims—diverge from those of the user. That concern is legitimate. No continent should want critical communications infrastructure to depend entirely on one company beyond its jurisdiction. 

The relevant policy goal, however, is resilience, not exclusion. Resilience comes from multisourcing, interoperability, and clear rules governing who can suspend service and under what conditions. Regulators can impose those requirements on operators regardless of nationality. 

Reserving a third of the commercial band for European-only firms does little to improve resilience. One European champion remains a single point of failure. The policy does, however, make higher prices and slower innovation more likely. The Commission reached for a protectionist tool when more narrowly tailored alternatives were available. The language of “sovereignty” may obscure that choice for casual observers, but it does not justify it. 

The Ghost of Dubai

None of this is new. The impulse to bring a global communications network under national or intergovernmental control has been a recurring feature of governance debates since the internet became commercial. 

One of the clearest examples came in 2012 at the International Telecommunication Union’s World Conference on International Telecommunications in Dubai. There, Russia and a coalition of states sought to expand the International Telecommunication Regulations to place internet naming, addressing, and traffic management under greater intergovernmental control. Put less technically, they wanted more internet governance to move from open technical bodies to governments negotiating with one another. 

The United States, the European Union, and dozens of other countries refused to sign on. The open, multistakeholder model—embodied by institutions such as the Internet Corporation for Assigned Names and Numbers and the Internet Engineering Task Force—prevailed, to the enormous benefit of the global internet. That model is not anarchic. It is governance by technical standards, broad participation, and interoperable rules rather than by national gatekeepers. 

The throughline from Dubai in 2012 to Brussels in 2026 is the conviction that borderless networks must ultimately answer to borders. The European Union’s 2 GHz proposal reflects that impulse in the form of economic protectionism. A related debate is now unfolding ahead of the 2027 World Radiocommunication Conference, where Russia and Iran are advancing Agenda Item 1.5, a proposal that would allow governments to require satellite operators to shut off service within their territory. 

The objectives differ. The EU proposal aims to favor domestic industry; the World Radiocommunication Conference proposal is closer to traditional information control. Both, however, occupy different points on the same sovereignty spectrum. Each rejects the premise that some networks work best under neutral, harmonized rules rather than political boundaries. 

Two Wrongs, One Spectrum

The right response is not to answer European protectionism with an American version of the same mistake. Candor requires turning the lens around. The Federal Communications Commission (FCC) has opened a proceeding to reconsider the longstanding presumption that World Trade Organization-licensed satellites may access the U.S. market, and it has signaled that the United States may “mirror” whatever restrictions the European Union ultimately adopts. 

The reciprocity impulse is understandable as a bargaining tool. A principled commitment to neutral, merit-based spectrum policy, however, points in the opposite direction. As the ICLE argued in its comments to the FCC on satellite market-access reciprocity, the goal should be to expand access to global markets, not accelerate a race to the bottom in which every jurisdiction walls off “its” spectrum. Mirroring Brussels’ mistake does not correct it. It merely doubles down on it. Ideally, trade negotiators can resolve the dispute before regulators do. 

What would neutral rules look like instead? Operators would compete on technical capability, interference management, and demonstrated ability to coexist with other systems—not on the nationality of their shareholders. Regulators would honor vested, paid-for rights and alter them only prospectively and predictably. Governments would pursue reciprocal openness, rather than reciprocal exclusion, by employing mutual-recognition agreements, harmonized interference standards, and transparent licensing processes that any qualified operator can navigate.  

That is how spectrum becomes a platform for competition rather than a trophy for favored firms. It is also the standard against which both Brussels’ and Washington’s policies should be judged. 

In the end, treating an inherently global resource as a national trophy does not create strong domestic champions. It creates protected firms that face less competition, higher costs for consumers, and a slower closing of the digital divide that satellite broadband is uniquely positioned to bridge. 

Europe can have a thriving satellite industry, or it can have spectrum sovereignty. The more it chases the latter, the harder it becomes to achieve the former.

The post The EU’s Bid to Nationalize Space appeared first on Truth on the Market.

That is what is at stake in Linney’s Pizza, LLC v. Board of Governors of the Federal Reserve System, now before the 6th U.S. Circuit Court of Appeals. 

In an amicus brief, the International Center for Law & Economics (ICLE) argues that forcing the Federal Reserve to push debit-card interchange fees below what the Durbin Amendment’s text requires would ultimately backfire on everyday consumers. 

Someone Has to Pay for the Airline Miles

To understand interchange fees, it helps to understand that payment-card networks operate as “two-sided platforms.” They must attract both consumers and merchants, and the value they provide to each group depends on participation by the other. As the Supreme Court explained in Ohio v. American Express: 

Sometimes indirect network effects require two-sided platforms to charge one side much more than the other… The optimal price might require charging the side with more elastic demand a below-cost (or even negative) price. With credit cards, for example, networks often charge cardholders a lower fee than merchants because cardholders are more price sensitive. In fact, the network might well lose money on the cardholder side by offering rewards such as cash back, airline miles, or gift cards. The network can do this because increasing the number of cardholders increases the value of accepting the card to merchants and, thus, increases the number of merchants who accept it. Networks can then charge those merchants a fee for every transaction (typically a percentage of the purchase price). Striking the optimal balance of the prices charged on each side of the platform is essential for two-sided platforms to maximize the value of their services and to compete with their rivals. [citations omitted]. 

In other words, payment-card networks must balance the prices charged to consumers and merchants in order to maximize value and compete effectively. Historically, interchange fees charged to merchants helped issuing banks subsidize consumer benefits, including debit-card rewards, free checking accounts, and lower minimum-balance requirements. 

When Congress Tilted the Scales

The Durbin Amendment and the Federal Reserve’s implementing rule, Regulation II, disrupted that balance by capping debit-card interchange fees. Merchants benefited from lower transaction costs, but a growing body of research—including studies by Federal Reserve economists—found significant unintended consequences. 

Covered banks responded to the lost revenue in predictable ways. They raised account fees, increased minimum-balance requirements, and reduced access to free checking. Those changes fell hardest on lower-income households, contributing to increases in the number of unbanked and underbanked consumers. Meanwhile, there is little evidence that merchants passed their savings on to consumers through lower retail prices. 

The appellants in Linney’s Pizza now seek to push those fee caps even lower. They argue that the Durbin Amendment permits the Fed to consider only the incremental costs directly associated with authorizing, clearing, and settling a debit-card transaction when setting interchange-fee caps. 

Chevron Is Gone. Agency Discretion Isn’t.

The timing of Linney’s Pizza matters because it arrives amid a major shift in administrative law. In Loper Bright Enterprises v. Raimondo, the Supreme Court overturned Chevron deference, holding that courts must exercise independent judgment when interpreting statutes rather than deferring to an agency’s reading of the law. 

That does not mean agencies have lost all discretion. As ICLE’s amicus brief notes, Loper Bright expressly recognized that Congress sometimes delegates policymaking authority to agencies through broadly worded statutes. The Durbin Amendment is a textbook example. 

The statute directs the Federal Reserve to “establish standards” for determining whether an interchange fee is “reasonable and proportional to the cost incurred by the issuer with respect to the transaction.” The central dispute in Linney’s Pizza is what Congress meant by “cost.” 

The Durbin Amendment requires the Fed to distinguish between two categories of costs. It says the Fed “shall” consider “the incremental cost incurred by an issuer for the role of the issuer in the authorization, clearance, or settlement of a particular electronic debit transaction.” It also says the Fed “shall not” consider costs that are not specific to a particular transaction. 

In Regulation II, the Fed concluded that the statute leaves room for a third category: costs that are specific to individual transactions, but are not merely incremental authorization, clearance, and settlement (ACS) costs. The agency therefore included fixed ACS costs, transaction-monitoring costs, issuer fraud losses, and network-processing fees when setting the interchange-fee cap. 

Linney’s Pizza argues that the Fed lacked authority to recognize this third category and that an earlier challenge to Regulation II succeeded only because courts then applied Chevron deference. 

But even after Loper Bright, the better reading of the statute supports the Fed. Linney’s Pizza’s interpretation would run afoul of the canon against surplusage, which instructs courts to avoid reading statutory language in a way that renders portions of a law unnecessary. If Congress wanted interchange fees to equal only incremental ACS costs, it could simply have said so. Instead, it directed the Fed to establish a fee standard that is “reasonable and proportional” to an issuer’s costs. 

As the district court explained: 

If the statute were as restrictive and mechanical as Linney’s Pizza suggests, then there would be no need to require the fee standard to be “reasonable and proportional” as it “would merely equal the incremental ACS costs…”

In short, the Durbin Amendment gives the Fed substantial latitude to determine which transaction-specific costs may be reflected in interchange fees. The agency exercised that authority by including fixed ACS costs, transaction-monitoring costs, network-processing fees, and fraud losses—not just bare incremental ACS costs. 

The Argument the Court Hasn’t Heard  

The most distinctive aspect of ICLE’s brief is that it argues the Federal Reserve could have gone much further under the Durbin Amendment’s text, consistent with the economics of two-sided markets: 

In recognition of the benefits to consumers enabled by interchange fees, the Board could have determined that a higher interchange fee is justified because it is “reasonable and proportional” to the cost incurred by the issuer in a specific transaction. The Board could have even published standards, enforceable ex post, for what a reasonable percentage for an interchange fee could be, considering consumer benefits that they enable, rather than setting any percentage themselves. Cf. Linney’s Pizza, 804 F. Supp. 3d at 731 (“[T]he Board’s role in implementing [the statute] involves more than mere calculation and cost-sorting drudgery.”). 

The key statutory phrase is “reasonable and proportional.” Those words suggest that the Fed’s job involves more than mechanically tallying costs. After all, determining whether a fee is “reasonable” makes little economic sense without considering what that fee accomplishes. 

In a two-sided market, interchange fees help fund benefits for consumers on the other side of the platform. If those fees enable banks to offer rewards, free checking, or other services that benefit consumers, the Fed could reasonably conclude that a higher fee remains “reasonable and proportional” to the costs associated with the transaction. The statute gives the agency room to make that judgment. 

Notably, the Supreme Court’s analysis of two-sided markets in Ohio v. American Express has remained largely confined to antitrust law. There, the Court held that “[e]valuating both sides of a two-sided transaction platform is… necessary to accurately assess competition.” 

Linney’s Pizza presents an opportunity to apply that same economic insight in a different context. If evaluating both sides of a payment-card network is necessary to understand competition, it is also necessary to understand consumer welfare when regulating interchange fees. The 6th U.S. Circuit Court of Appeals should therefore recognize that the Fed has discretion to consider the consumer benefits that interchange fees make possible when determining whether those fees are “reasonable and proportional.” 

Otherwise, a statute intended to help consumers risks doing the opposite. As the evidence discussed above suggests, Regulation II’s interchange-fee cap is already far too low—not too high, as the appellants contend. 

Further choking off debit-card interchange fees will not make your next pizza cheaper. It will more likely make your checking account more expensive. 

There is no such thing as a free lunch—even at Linney’s Pizza.

The post No Free Lunch at Linney’s Pizza appeared first on Truth on the Market.

The court’s partial annulment of the European Commission’s Digital Markets Act designation of Meta—limited to Facebook Marketplace—is not a revolution. It did not invalidate the DMA. It did not reject the Commission’s authority to designate large digital platforms as gatekeepers. It did not save Meta’s Messenger service—a standalone instant-messaging platform that lets users send texts, share high-definition photos and large files, and make voice or video calls without needing a phone number—from the DMA’s reach.

Formally, the decision is narrow. The court upheld the Messenger designation, while annulling the designation of Facebook Marketplace—a digital classifieds platform embedded in the Facebook app and website—because the Commission had not adequately justified its analysis.

Precisely because the judgment is narrow, it may prove significant. It suggests how European Union courts can discipline regulatory overreach without openly repudiating the political choices embodied in modern EU digital regulation.

Labels Are Not Analysis

The decision matters because it insists that even ex ante regulation—rules imposed before specific competitive harm has been proved—must receive close legal scrutiny.

The Commission may be tempted to treat gatekeeper designation as an administrative shortcut: identify a large firm, invoke the statutory thresholds, and impose behavioral obligations meant to reshape digital markets in the name of fairness and contestability. But the General Court’s Marketplace holding suggests that labels matter, statutory categories matter, evidence matters, and reasons matter.

If a service is to be treated as a core platform service, the Commission must explain why the legal category fits the economic reality. It cannot infer legal consequences from the size, notoriety, or political salience of the undertaking.

That is a modest but valuable judicial signal. The greatest danger in contemporary competition policy is not merely that regulators sometimes err. Error is inevitable. The more serious danger is that regulators will institutionalize a framework in which mistakes are cheap for the state and expensive for markets.

The DMA, like other ex ante regimes, lowers the Commission’s burden relative to traditional antitrust. It replaces case-by-case proof of competitive harm with categorical obligations triggered by designation. That makes procedural discipline especially important. When a legal regime operates before harm has been demonstrated, courts should be more—not less—demanding about whether the triggering conditions have been satisfied.

The Meta judgment therefore has implications beyond Marketplace. It may become part of an emerging judicial vocabulary for reining in excessive interventionism—not by announcing a deregulatory manifesto, but by requiring the Commission to match regulatory ambition with legal precision and economic evidence.

The court’s message is not that large platforms are immune from regulation. It is that even politically disfavored platforms deserve intelligible reasoning. In a legal order committed to the rule of law, this should be uncontroversial. In practice, it may be transformative.

When Competition Policy Becomes Product Management

The Commission’s modern competition and digital-policy agenda has increasingly moved from policing exclusionary conduct to engineering market structure and product design.

The DMA does not merely prohibit certain anticompetitive acts. It often dictates how integrated services must be unbundled, how data may be combined, how app stores must be opened, how default settings must be designed, and how interoperability must be supplied. This represents a shift from antitrust as law enforcement to competition regulation as ongoing industrial administration.

That shift should concern anyone who takes dynamic competition seriously. Digital platforms compete not only through price, but also through integration, security, convenience, network effects, data-driven improvements, and rapid product iteration. Treating seamless integration as presumptively suspect risks sacrificing real consumer benefits to a static vision of market structure.

The point is not that integration is always benign. It is that integration is often how innovation reaches consumers. Search linked to maps, messaging linked to commerce, operating systems linked to app distribution, and social networks linked to marketplaces may all create risks. They may also reduce transaction costs, improve quality, and make products easier to use.

The DMA’s core danger is that it can convert this ambiguity into a regulatory presumption against scale and integration. If a large platform improves a service by connecting it to another service, regulators may characterize the improvement as leveraging. If it designs a seamless ecosystem, they may call that self-preferencing. If it maintains a closed architecture for privacy or security reasons, they may attack that as foreclosure. If it opens the architecture, they may blame it for new security risks.

That is a difficult environment in which to innovate.

The General Court’s ruling does not solve those problems, but it offers a procedural foothold. If the Commission must justify the classification of each service, then it must conduct a matter-specific economic evaluation, rather than rely on administrative convenience.

Facebook Marketplace, for example, could not simply be swept into DMA coverage because it sat inside the Meta ecosystem. The Commission had to show why Marketplace fit the relevant statutory concept and why Meta’s arguments did not undermine that conclusion. Requiring that kind of analysis can force regulators to confront tradeoffs that broad political narratives obscure.

Five Ways Courts Can Keep the DMA Honest

Future General Court and Court of Justice cases could develop this discipline in several ways.

First, EU courts should insist on genuine service-by-service analysis. The DMA’s structure invites a dangerous conflation between the undertaking and the service. A company may be large, but not every service it offers necessarily functions as an important gateway between business users and end users.

Treating every service offered by a large digital firm as presumptively gatekeeping would collapse the statutory inquiry into a company-level judgment. Courts should resist that move. The question should not be whether Meta, Apple, Google, Amazon, or Microsoft is important. The question should be whether the specific service at issue satisfies the legal and economic conditions for designation.

Second, courts should require the Commission to address countervailing evidence meaningfully. If a platform argues that a service is declining, that users multi-home, that business users have alternative routes to customers, that the service is not monetized as the Commission assumes, or that the relevant functionality is ancillary rather than independent, the Commission should not be able to wave those points away with conclusory language.

The duty to state reasons should become a duty to confront the economic record. That would not turn DMA designation into a full Article 102 abuse case, but it would prevent designation from becoming a rubber stamp.

Third, courts should scrutinize remedy specification. The most intrusive DMA interventions may arise not at designation, but during compliance and specification proceedings. Once a firm is designated, the Commission can exert substantial pressure over product architecture.

Courts should therefore remain attentive to proportionality. A compliance measure that degrades privacy, cybersecurity, product quality, or innovation incentives should not be accepted merely because it increases the formal number of choices available to users or rivals. Choice is valuable when it improves welfare. It is not valuable when regulators manufacture it by degrading the default product or forcing users through confusing consent screens and fragmented experiences.

Fourth, courts should revive administrable limits on open-ended concepts like fairness and contestability. These terms are politically attractive but economically slippery. A market can be contestable because entrants can challenge incumbents. It can also become less efficient if regulation props up less capable rivals.

A practice can be called unfair because it disadvantages competitors, but competition law should not confuse harm to competitors with harm to competition. EU courts need not import the American consumer-welfare standard wholesale to recognize that competition policy requires limiting principles. Without them, fairness becomes a license to redistribute rents from successful firms to less successful rivals.

Fifth, the Court of Justice should use appeals to clarify that ex ante regulation remains subject to proportionality, legal certainty, and institutional competence. The Commission is not a legislature with roving authority to redesign the digital economy. Nor is it a venture capitalist charged with choosing the optimal architecture of future markets.

The Commission’s powers derive from the treaties and from legislation adopted under them. When it uses those powers to impose obligations with major economic consequences, courts should require a clear legal basis, a reasoned explanation, and a plausible relationship between the measure and the statutory objective.

Europe Does Not Need More Regulatory Swagger

This kind of judicial reining-in would not be anti-European. On the contrary, it may be essential to Europe’s economic renewal.

The Draghi report and related commentary on Europe’s competitiveness problem have highlighted a sobering reality: Europe has strong institutions, educated workers, sophisticated consumers, and deep scientific capabilities, but it has struggled to generate and scale world-leading technology firms.

The problem is not a shortage of regulatory ambition. Europe has produced the General Data Protection Regulation, the Digital Services Act, the DMA, the Artificial Intelligence Act, and numerous national initiatives. The problem is that regulatory ambition has not translated into entrepreneurial dynamism.

Innovation requires more than subsidies, research programs, and public strategies. It requires permission to experiment. It requires the possibility of scale. It requires exit opportunities for startups, including acquisition by larger firms. It requires predictable rules that allow entrepreneurs and investors to estimate risks. It requires tolerance for business models that policymakers may not fully understand at inception.

An economy that regulates first and learns later will tend to get less of the experimentation that produces transformative growth.

The Costs You Cannot Count

A market-oriented approach to EU competition law should therefore emphasize error costs. False positives—mistakenly condemning beneficial conduct—are especially damaging in innovation markets.

If regulators mistakenly prohibit or burden a practice that would have benefited consumers, the lost innovation may never be observed. A delayed product launch, an abandoned integration, a startup that cannot find a buyer, or a platform feature never introduced in Europe will not always appear in enforcement statistics.

The Commission can count investigations, workshops, designations, and fines. It is much harder to count the innovations that regulation prevented.

Those unseen costs matter. When compliance burdens fall on large platforms, some observers assume the costs fall only on wealthy foreign companies. That is a mistake. Large platforms are infrastructure for smaller firms, developers, advertisers, creators, retailers, and consumers.

If regulation makes platforms less efficient, less integrated, or slower to deploy new tools, the costs propagate through the ecosystem. European startups may find fewer distribution channels, fewer acquisition opportunities, fewer integrated services, and less access to frontier technologies. Consumers may receive products later than users in the United States or Asia. Small businesses may face more fragmented marketing and transaction systems.

A judicially enforced discipline of evidence and proportionality could help reverse this pattern. If the Commission knows courts will demand careful reasoning, it may become more selective. If it must explain how a designation or remedy improves competition rather than merely handicaps a large firm, it may focus on clearer cases of exclusion. If it must account for innovation and product-quality tradeoffs, it may avoid interventions that make digital services less useful.

Such discipline would not eliminate EU regulation. It would improve it.

Competition Is a Process, Not a Diorama

The implications extend to traditional EU competition law. The Commission’s Article 101, Article 102, and merger-control enforcement has often been more interventionist than the American approach, particularly in its suspicion of dominance, vertical integration, rebates, tying, self-preferencing, and mergers involving potential competition.

But recent decades have also seen EU courts require more rigorous effects analysis in important cases. That trend should be strengthened. The future of EU competition law should not lie in abandoning economics for administrability. It should lie in using economics to make administrability honest: clear rules where experience justifies them, but serious evidence where intervention threatens dynamic rivalry.

A more market-oriented EU competition policy would begin with several presumptions. It would presume that competition is a process, not a market-design endpoint. It would presume that scale can be efficient, especially in markets characterized by network effects, high fixed costs, data complementarities, and global rivalry.

It would presume that consumer welfare includes quality, privacy, security, convenience, and innovation—not merely the number of competitors or the formal availability of alternatives. It would presume that intervention should be justified by a theory of harm and disciplined by a theory of error costs. And it would presume that protecting competitors from hard competition is not the same as protecting competition.

This approach could enhance innovation in the EU in concrete ways.

First, it would make Europe more attractive for product launches. Firms are more likely to introduce new technologies in jurisdictions where legal risk is predictable and proportionate.

Second, it would help startups by preserving efficient integration with larger ecosystems. Many startups do not become standalone giants. They succeed by being acquired, by partnering, or by building on platform infrastructure.

Third, it would encourage European firms to scale. A policy culture that treats size as a problem will not produce many large technology firms.

Fourth, it would reduce fragmentation. One promise of EU law is the creation of a single market, but overlapping EU and national interventions can recreate fragmentation through compliance complexity.

Fifth, it would shift attention from performative enforcement to welfare-enhancing enforcement.

A More Interesting Brussels Effect

The Meta decision may also have transatlantic implications. American competition law has recently flirted with more interventionist theories, particularly through neo-Brandeisian critiques of bigness, platform integration, and merger policy. Some American commentators have cited the DMA as a model for regulating large technology platforms.

But if EU courts begin to cabin the Commission’s discretion, the lesson for the United States may change. Europe may not show that aggressive ex ante regulation is the future. It may show that judicial review remains an essential corrective to administrative enthusiasm.

American antitrust law already contains doctrinal resources that support a market-oriented approach: burdens of proof, effects analysis, skepticism toward protecting competitors rather than competition, and concern for administrable rules. A European judicial turn toward evidence, proportionality, and dynamic competition could reinforce those tendencies.

It would make it harder for American interventionists to claim that the global consensus favors structural regulation of digital markets. It would also give American courts and agencies comparative support for a humbler proposition: competition policy should be especially cautious in fast-moving markets.

There is an irony here. For years, the so-called “Brussels effect” was invoked to suggest that EU regulation would become the global default because large firms would conform worldwide to Europe’s rules. But the Meta judgment hints at a different possibility: a rule-of-law Brussels effect. If European courts insist that the Commission justify intervention with rigor, the EU could export not regulatory maximalism, but regulatory discipline.

That would be a healthier model for both sides of the Atlantic.

Small Rulings Can Cast Long Shadows

One should not overstate the likelihood of a dramatic shift. The General Court upheld Messenger’s designation, and the DMA remains politically popular in Brussels. The Commission is unlikely to abandon its view that large platforms require close supervision. The Court of Justice may also proceed cautiously on appeal, especially where the legislature has expressly chosen an ex ante regulatory model.

The path toward a more market-oriented European competition law will therefore be incremental. But incremental judicial discipline can matter. Competition policy is shaped not only by grand doctrines, but also by burdens of explanation.

If the Commission must explain more, it may assume less. If it must address tradeoffs, it may intervene more carefully. If it must classify services according to legal criteria rather than political narratives, it may narrow its targets. If it must defend product-design mandates in terms of proportionality and welfare, it may hesitate before turning competition law into engineering supervision.

The Meta Marketplace ruling is best understood in that light. It is a small legal defeat for the Commission, but potentially a larger institutional warning.

Europe’s competitiveness problem will not be solved by giving regulators more discretion to rearrange markets. It will be solved, if at all, by allowing markets to discover better arrangements, while reserving intervention for cases where evidence shows genuine competitive harm. Courts cannot create European dynamism by themselves. But they can remove some obstacles by insisting that economic regulation remain tethered to law, evidence, and proportionality.

A market-oriented reading of the decision therefore sees hope in its restraint. The court did not announce that the Commission is wrong to care about digital competition. It announced that caring is not enough. The Commission must reason. It must justify. It must respect categories. It must answer arguments.

In a legal order increasingly tempted by technocratic management of markets, that is a message worth taking seriously.

If future EU courts build on this foundation, they could help restore a better balance between competition enforcement and economic liberty. They could remind regulators that innovation often comes from the very practices—scale, integration, experimentation, and ecosystem design—that interventionist policy tends to distrust. They could help Europe move from a culture of precautionary control to one of competitive discovery.

And, in doing so, they might influence American antitrust at a crucial moment by strengthening those who argue that the goal of competition law is not to punish success, but to preserve the market process that makes success contestable.

That would be a Brussels effect worth welcoming.

The post The DMA Meets the Rule of Law appeared first on Truth on the Market.

That debate is already too narrow.

President Donald Trump’s recent executive order on “Promoting Advanced Artificial Intelligence Innovation and Security” reflects the tension. The order emphasizes maintaining American leadership in artificial intelligence while directing federal agencies to develop voluntary frameworks, reporting standards, and oversight mechanisms for advanced models. Its stated objective is laudable: encourage innovation without imposing unnecessary burdens on one of the most important technologies of the 21st century.

Yet the executive order also highlights a deeper problem in the current debate. Both sides assume the central question is who should regulate AI. Should authority reside in Washington or in the states? Should Congress act, or should state legislatures lead?

That is the wrong question.

The more fundamental question is whether either level of government possesses the knowledge, incentives, or institutional capacity to regulate a technology evolving at extraordinary speed. Before deciding who should regulate AI, we should first ask whether government can regulate it effectively at all.

The answer is no.

Critics of AI regulation often focus on innovation, competitiveness, or economic growth. Those concerns matter. But they are secondary to a more fundamental insight developed by economists associated with the Austrian and Virginia schools of political economy. The problem is not simply that regulation may slow innovation. It is that neither federal nor state regulators possess the knowledge necessary to determine what AI should become. And even if they did, the political process would steadily transform limited oversight into expansive control.

The result is a regulatory project almost certain to fail on its own terms.

The Federalism Distraction

The executive order has reignited a debate that has been simmering for months: Should AI be regulated by the federal government or by the states?

Both sides raise legitimate concerns. Advocates of federal regulation worry that a patchwork of state laws will fragment national markets, raise compliance costs, and weaken America’s ability to compete with China. Supporters of state authority respond that federal regulation could entrench incumbent firms, suppress experimentation, and impose a one-size-fits-all framework on a technology still taking shape.

Both positions have merit. But both share the same mistaken premise: that the central question is which government should regulate AI.

From an Austrian perspective, the more important question is whether any political institution has the knowledge needed to regulate a technology whose most important applications, risks, and governance mechanisms are still being discovered. Put differently, the choice between federal and state regulation assumes the relevant knowledge already exists and merely needs to be assigned to the right regulator.

But much of that knowledge does not yet exist at all.

That is why the debate is ultimately not about federalism. It is about knowledge. Whether authority sits in Washington or 50 state capitals, regulators face the same basic problem: They must make decisions today about technologies whose future uses, risks, and opportunities remain largely unknown.

Federal regulation may reduce jurisdictional variation, but it cannot solve the knowledge problem. State regulation may allow experimentation among jurisdictions, but it remains constrained by the same informational limits. Either way, policymakers are trying to govern a discovery process before the relevant discoveries have been made.

A National Center of Ignorance

More than 80 years ago, F.A. Hayek articulated one of the most important insights in economics: the knowledge needed to coordinate a complex economy is dispersed among millions of individuals. It exists in fragments. It is local, often tacit, and constantly changing. No central authority possesses all of it. Markets coordinate this dispersed knowledge through the price system.

The same insight applies with even greater force to artificial intelligence. The challenge facing policymakers is not simply that AI is evolving rapidly. It is that nobody knows what AI ultimately will become. The technology’s most valuable uses may not yet have been discovered. The most important safety mechanisms may not yet exist. The most effective governance structures may emerge years from now through experimentation and competition.

In other words, AI is not merely a product. It is a discovery process.

That distinction matters. Unlike a traditional product, whose characteristics are largely known before it reaches consumers, AI’s capabilities, limitations, and most valuable applications are being discovered through use. Every interaction between users and AI systems generates information about what works, what fails, what creates value, and what risks warrant attention. That information emerges through experimentation and experience, not centralized planning.

Governments routinely regulate products. Automobiles, pharmaceuticals, and household appliances have relatively stable characteristics. Regulators can inspect them, test them, and develop standards around known risks.

AI is different. Every day, millions of people interact with large language models, coding assistants, image generators, scientific research tools, and enterprise software. Through those interactions, users discover new applications, identify flaws, develop workarounds, and reveal preferences. The technology evolves in response.

The relevant knowledge therefore does not reside in Washington, Sacramento, Austin, or Albany. It resides with the millions of users, developers, entrepreneurs, and businesses experimenting with AI in real time.

Centralizing regulatory authority does not solve this problem. It merely moves decision-making farther away from the people who possess the relevant information.

The current debate often assumes that federal regulation is preferable because it avoids a patchwork of state rules. But a single national standard cannot overcome the knowledge problem. It simply creates a single national center of ignorance.

Replacing 50 imperfect regulators with one imperfect regulator does not eliminate the underlying difficulty.

It magnifies it.

Freezing the Future

Israel Kirzner extended Hayek’s insight by emphasizing entrepreneurial discovery. Markets are not static systems moving neatly toward equilibrium. They are dynamic processes through which entrepreneurs discover opportunities others have missed. Competition matters not because it produces a predetermined outcome, but because it reveals information nobody previously recognized.

AI development exemplifies this process. No regulator predicted the explosive growth of prompt engineering—the practice of shaping inputs to get better outputs from AI systems. Few anticipated the rapid rise of AI coding assistants. Fewer still foresaw how quickly businesses would integrate generative AI into legal services, drug discovery, customer support, software development, and scientific research. Those discoveries emerged through experimentation.

That poses a serious problem for regulators. Any reporting requirement, disclosure standard, certification process, or safety framework necessarily reflects current knowledge. It embodies policymakers’ best understanding of responsible AI development at a particular moment.

But what if that understanding is wrong? More realistically, what if it is incomplete? The danger is not merely that regulators may make mistakes. It is that regulation freezes today’s assumptions into tomorrow’s rules.

A reporting framework developed in 2026 reflects what policymakers believe AI risks and opportunities look like in 2026. Yet the market’s understanding of those risks and opportunities may look entirely different in 2028 or 2030.

The more detailed the framework becomes, the greater the risk that it will block the discovery of better alternatives. Innovation frequently comes from directions experts fail to anticipate. The history of technology is filled with entrepreneurs discovering opportunities that established firms, government agencies, and academic specialists overlooked.

That is precisely why regulatory efforts to direct innovation so often disappoint. They attempt to manage a process whose most important outcomes have not yet been discovered.

The Machinery Matters More Than the Mission

Suppose regulators somehow overcome the knowledge problem. Suppose they develop a genuinely modest framework, act with the best intentions, and remain committed to encouraging innovation.

The problem still remains.

The Virginia School of political economy teaches us to abandon what James Buchanan famously called the “romantic” view of politics. Government officials are not omniscient guardians of the public interest. They are human beings responding to incentives, just like everyone else.

That insight has profound implications for the executive order. The administration deserves credit for recognizing AI’s importance to American prosperity, innovation, and national security. The order repeatedly emphasizes the need to maintain U.S. leadership in AI and says America should not “stifle this innovation with overly burdensome regulation.” Taken at face value, those aspirations are entirely sensible.

The difficulty is that institutions often matter more than intentions. The order directs federal agencies to develop a voluntary framework for advanced AI models, establishes reporting and disclosure mechanisms, and contemplates a federal role in evaluating frontier systems—the most advanced AI models at the cutting edge of development.

Many observers view these structures as limited and reasonable. Perhaps they are. The more important question is whether they will remain limited and reasonable.

Public-choice theory suggests otherwise.

Bureaucracies have incentives to expand their authority. Agencies benefit from larger budgets, larger staffs, and broader mandates. Politicians benefit from claiming credit for solving perceived problems. Interest groups benefit from influencing rules that affect their competitors.

Reporting and disclosure requirements often represent the first stage of a broader regulatory architecture. Before government can certify, license, restrict, or otherwise supervise an activity, it must first gather information about it. Supporters understandably view these provisions as modest transparency measures designed to improve visibility into frontier AI development. Yet from a public-choice perspective, information collection is rarely the endpoint. Once reporting mechanisms exist, they create both the capacity and the temptation for future policymakers to convert information gathering into more active oversight.

The executive order should therefore be evaluated not only by what it does today, but by the machinery it creates for tomorrow.

The history of regulation is instructive. Most regulatory regimes begin modestly. They aim to address specific concerns, often with assurances that they will remain narrowly tailored. Yet American regulation is full of agencies and programs that expanded far beyond their original scope.

New reporting requirements lead to new oversight responsibilities. New oversight responsibilities justify new staff and budgets. New bureaucratic capacities create pressure to identify additional problems requiring additional intervention.

That result does not require corruption or bad faith. It is simply what happens when ordinary incentives operate inside political institutions.

Why Big AI Loves Regulation

The greatest threat posed by AI regulation may not come from government officials themselves. It may come from the firms invited to help write the rules.

Gordon Tullock’s theory of rent-seeking explains that whenever government acquires the power to distribute benefits or impose burdens, individuals and firms devote resources to influencing those decisions. Rather than competing solely in markets, they compete for political advantage.

AI regulation creates precisely these incentives. Large AI companies have compliance departments, legal teams, lobbying operations, and deep financial resources. Small firms and open-source developers often do not. As a result, incumbent firms may welcome regulatory frameworks that appear neutral while imposing costs their smaller rivals struggle to bear.

The language of safety, transparency, accountability, and security can therefore become a mechanism for raising rivals’ costs.

A federal reporting standard may sound modest. Yet every reporting requirement requires personnel, documentation, legal review, and administrative infrastructure. For a trillion-dollar technology company, those costs are manageable. For a startup operating out of a garage, they may be prohibitive.

The result is a familiar pattern: regulation that ostensibly protects the public often ends up protecting incumbents.

Milton Friedman observed this dynamic repeatedly throughout the 20th century. Regulatory agencies established to protect consumers, he argued, frequently evolved into institutions that protected established firms from competition.

There is little reason to believe AI regulation will be different. If anything, the incentives may be even stronger. The stakes are enormous, the potential rewards are vast, and the firms involved have every reason to shape the rules governing the market they already dominate.

The Interventionist Ratchet

Proponents of federal AI regulation often emphasize that current proposals are limited. That misses the point.

The central lesson of Ludwig von Mises’ analysis of interventionism is that government interventions frequently produce consequences policymakers neither anticipate nor desire. Those unintended consequences then become the justification for additional interventions. The process feeds on itself, generating ever-larger and more complex regulatory regimes.

Consider AI safety. No regulatory framework will eliminate all risks associated with AI systems. Hallucinations will occur. Security incidents are inevitable. Misuse is unavoidable.

When those events happen, policymakers will face pressure to respond. Will they conclude that regulation has failed and should be scaled back?

History suggests otherwise.

The more likely response is that the existing framework did not go far enough. Additional reporting requirements will be proposed. Expanded disclosure obligations will be considered. New certification standards will be developed. Additional oversight bodies will be created.

Each intervention generates new perceived shortcomings. Each shortcoming becomes the rationale for another intervention.

This dynamic does not require bad intentions. It emerges from the incentives embedded in political institutions. Once a regulatory system is in place, the pressure almost always runs in one direction: toward expansion.

What begins as a voluntary framework may become an expected industry practice. What begins as a disclosure requirement may evolve into a certification requirement. What begins as a reporting standard may become a licensing regime.

None of this is inevitable. But all of it becomes possible once the institutional foundation has been laid.

That is the interventionist ratchet. It turns temporary solutions into permanent structures and modest frameworks into increasingly comprehensive systems of control.

Too Big to Discipline?

Advocates of federal regulation often dismiss market discipline as inadequate because the largest AI companies are supposedly too powerful. On this view, firms such as Microsoft, Amazon, Google, Meta, OpenAI, and Anthropic have resources so vast that meaningful accountability requires government intervention.

But that argument overlooks an important reality: Even the largest firms in the AI ecosystem face intense competitive pressure.

The current AI race illustrates the point. Hyperscalers—large cloud-computing providers such as Amazon, Microsoft, and Google—and frontier-model developers are investing hundreds of billions of dollars in new infrastructure. They are not doing so because regulators require it. They are doing so because customers demand better performance, lower latency, greater reliability, and more sophisticated capabilities.

Data-center construction has accelerated because users want more computing power. Companies are pursuing nuclear-power agreements, natural gas generation, and other energy solutions because reliable, always-on electricity has become a competitive necessity. These firms are not dictating market outcomes. They are responding to them.

The same pattern appears in AI safety. Long before governments developed comprehensive regulatory frameworks, leading firms were already investing heavily in alignment research, red-teaming, model evaluations, and safety testing. Alignment research aims to make AI systems behave consistently with human goals. Red-teaming means stress-testing systems to find vulnerabilities before bad actors do.

Cynics may dismiss these efforts as public relations exercises. Some undoubtedly are. But that misses the larger point: Firms devote resources to these activities because customers, enterprise clients, investors, and the public increasingly demand them.

Enterprise customers do not want unreliable systems generating inaccurate outputs. Businesses integrating AI into critical operations care deeply about security, transparency, and predictability. Investors worry about reputational risk. Users abandon products that consistently fail to meet expectations. When allowed to function, these market pressures create powerful incentives for self-correction.

Thomas Sowell’s distinction between market participants and surrogate decision-makers is useful here. Firms and customers bear the consequences of their decisions. When an AI model fails, developers lose revenue, customers lose productivity, and investors lose capital. The costs are immediate and tangible.

Regulators, by contrast, typically bear little personal cost when their decisions prove mistaken. Their feedback mechanisms are weaker, slower, and more indirect.

This does not mean markets are perfect. Market signals can be noisy. Firms make mistakes. Consumers misjudge risks. But those imperfections are not an argument for replacing markets with political control. They are precisely why competition matters. As Hayek observed, competition serves as a discovery procedure: Different firms pursue different approaches, and the market reveals which ones work.

Government standards tend to homogenize behavior. Markets encourage variation. In a technology as uncertain as AI, that variation is essential.

The hyperscalers’ enormous investments in infrastructure, energy, and safety are not evidence that markets have failed.

They are evidence that markets are working.

The Real Regulators

If neither the states nor the federal government can effectively regulate AI, who can?

The answer is surprisingly simple: the people who use it.

Mises described this idea as consumer sovereignty. His central insight was that markets are governed not by bureaucrats, but by consumers. Producers succeed only by satisfying the preferences of the people they serve. In “Human Action,” Mises likened the consumer to the captain of a ship:

The captain is the consumer. Neither the entrepreneurs nor the farmers nor the capitalists determine what has to be produced. The consumers do that.

AI is no exception.

Every day, users evaluate AI systems based on accuracy, reliability, bias, speed, privacy, security, and usefulness. When systems perform well, users reward them with greater adoption. When they do not, users leave—or worse, switch to a competitor.

When companies fail to address legitimate concerns, they lose customers, revenue, and reputation. These feedback mechanisms operate continuously and at extraordinary scale.

Unlike government regulators, users possess direct knowledge of their own needs. Unlike bureaucratic oversight, consumer feedback generates immediate consequences. Unlike regulatory mandates, consumer preferences evolve as circumstances change.

Most importantly, consumer choice encourages experimentation. Different users value different things. Some prioritize safety. Others prioritize creativity. Others care most about speed, privacy, transparency, or cost.

Markets accommodate those differences. Regulatory frameworks often suppress them.

Put differently, consumer sovereignty allows millions of individuals to make decisions for themselves. Government regulation substitutes the judgment of a relatively small number of officials for the judgment of everyone else.

Who Do You Trust?

The debate over AI regulation ultimately reflects a deeper disagreement about how knowledge is generated and how social order emerges. The prevailing view assumes wise policymakers can identify the right balance between innovation and safety, codify that balance into rules, and steer society toward the desired outcome.

The Austrian tradition suggests something very different. Knowledge emerges through discovery. Order emerges through interaction. The future cannot be designed in advance because the most important information about that future does not yet exist.

Hayek explained why no central authority can possess the knowledge needed to direct a complex economy. Kirzner showed that entrepreneurial discovery is how new knowledge gets generated. Buchanan and Tullock reminded us that political institutions respond to incentives no less than markets do. Mises showed how interventions tend to expand over time and why consumers—not bureaucrats—ultimately determine economic outcomes.

That final point is especially important. In “Human Action,” Mises argued that the apparent captains of industry are not the true sovereigns of the market. Business leaders hold their positions only so long as they satisfy consumers. The ultimate authority rests not with producers, but with the individuals who choose what to buy, what to use, and what to reject.

The same principle applies to artificial intelligence. The future of AI will not ultimately be determined by task forces, reporting standards, advisory committees, or federal frameworks. It will be shaped by millions of users deciding which systems they trust, which capabilities they value, and which firms deserve their business. Those choices generate the feedback that guides investment, rewards innovation, punishes failure, and encourages continuous improvement.

Put simply, the debate is not really about whether AI will be regulated. It already is. Every day, users regulate AI through their choices in the marketplace. They reward firms that provide value and abandon those that do not. They encourage useful applications and reject those that fail to meet their needs. They do so continuously, dynamically, and at a scale no government institution could replicate.

The question, then, is not whether we should have regulation. It is whether we trust the decentralized judgments of millions of users or the centralized judgments of a comparatively small number of policymakers. The executive order reflects faith in the latter. The Austrian tradition counsels faith in the former.

If the goal is to promote both innovation and security, policymakers should remember a lesson economists from Hayek to Mises emphasized again and again: The knowledge needed to govern society is not concentrated in Washington. It is dispersed throughout society itself.

The challenge is not to replace that discovery process with regulation.

It is to get out of its way.

The post Artificial Intelligence, Natural Ignorance appeared first on Truth on the Market.

We—people—often have a hard time letting a bad thing go. That’s true even for those who are well acquainted with the sunk-cost fallacy and should know better. I’ve been there. I’ve felt that misguided determination, and I’ve felt the consequences. 

The same is true of the people who run federal agencies, including the two with which I’m most familiar: the Federal Trade Commission (FTC) and the Antitrust Division of the U.S. Department of Justice (DOJ). 

Which brings me to . . .

The Appeal of a Sunk Cost

On May 22, the FTC filed an appeal in FTC v. Meta Platforms, Inc.; that is, the FTC filed its opening brief appealing the agency’s loss in November 2025 before the U.S. District Court for the District of Columbia (readers might do well to consult the revised December 2025 opinion, which is largely the same as the November opinion, but with fewer redactions). As I noted back in April 2025, the FTC’s case seemed to me (among many others) an uphill battle. And as I noted in December 2025, Judge James Boasberg’s decision:

Doesn’t exactly provide the law & economics analysis I would have produced, had anyone asked for it. But it is, nonetheless, a good decision by a thoughtful, generalist trial-court judge wrestling with both the evidence before him and relevant precedent. He got key things right, not least of which was the decision for the defendant. 

For additional critiques of the FTC’s case, see Brian Albrecht, Mark Jamison, Joe Coniglio, Alden Abbott, and—for a banger Tech Policy Podcast—Geoff Manne. For additional background, see Daniel Crane and Herbert Hovenkamp’s critique of the “monopoly broth” theory applied to what was originally the Facebook case, albeit in an amicus brief filed in another matter.

I won’t recapitulate most of the case against the FTC’s position. There’s plenty to be found in the opinion itself (linked above), and I discuss many of these points in the Truth on the Market posts I identified (and linked) above.

Here, I want to focus on the FTC’s appeal, which gets a bit into the weeds of the FTC Act and the Sherman Act, both in terms of statutory construction and jurisprudence. I’m sorry about that, but that’s the law part of law & economics, and law & economics is “the business we have chosen.”

The brief advances two arguments that, on the FTC’s account, identify “fundamental and independent ways” in which the district court erred, “each of which warrants reversal.” The second seems to me to hint at the FTC’s basic challenge on appeal. There, the commission argues that the district court “misapplied fundamental antitrust law principles by failing to recognize the validity of the PSN [personal social networking] market, whether in 2020 or 2025.”

I think that’s wrong, but I’m not going to spend this blog post rehashing the better part of Judge Boasberg’s 89-page decision identifying the central reason the FTC lost at trial: the burden of proof belonged to the plaintiff, and the plaintiff did not meet it. The challenge for the FTC is that the U.S. Court of Appeals for the District of Columbia Circuit is not about to review the case de novo, nor is it likely to repudiate the trial court’s findings of fact.

I think that’s why the FTC leads with an argument about the proper reading of Section 13(b) of the FTC Act. The goal is to identify a reversible error by the trial court. In practical terms, that would have to be an error of law, and a reversible one at that. It’s all about timing, and while the issue is presented as a simple matter of statutory construction, it’s not so easy.

The FTC’s argument about Section 13(b) was nowhere in my mind when I first wrote about the case in 2023, long after the FTC filed its complaint, but well before the matter went to trial. Nor is it what interests or concerns me most about the appeal. But it is the agency’s principal argument at this stage, so here goes. 

The FTC’s Section 13(b) Detour

The heart of the commission’s appeal, as it presents the matter, can be found in this passage from the brief’s introduction:

Start with the major premise—that the FTC was obligated to prove a 2025 monopoly. The court’s cursory analysis focused on Section 13(b) of the FTC Act, which authorizes the Commission to “bring suit” in certain cases “[w]henever” it “has reason to believe” that the defendant “is violating, or is about to violate” any law the Commission enforces. 15 U.S.C. § 53(b), (b)(1). By its plain terms, this is a pleading requirement that applies at the time of the complaint. Nothing in the statutory text requires the Commission to make any substantive showing as of the time of decision. . . .

That simple observation resolves the appeal. As a result, this Court need not consider the broader question of whether Section 13(b)(1) applies here at all. The answer is no. This case was brought under the second proviso of Section 13(b), which empowers the Commission to seek permanent injunctive relief in federal court. Section 13(b)(1) is located in a different part of Section 13(b) and does not apply to second-proviso cases.

I think I see where they’re going with this, but it’s a bit confusing, not least because the latter paragraph is wrong. And the FTC repeats the mistake multiple times, in multiple sections of the brief. On page 16, it asserts that “fundamentally, Section 13(b)(1) does not apply to cases like this one, where the Commission seeks a permanent injunction . . .” directly in federal court. We find more of the same on page 20, and again on pages 30-32.

But it’s hard to see any way around the opposite conclusion. That “different part of Section 13(b)” is Section 13(b)(2), and the plain language of the statute does not permit 13(b)(1) and 13(b)(2) to be separated.

The relevant section of the FTC Act, 15 U.S.C. § 53, authorizes the commission to bring suit in federal court. Section 13(a) addresses suits involving advertisements, where the commission has “reason to believe” the ads violate Section 12 of the FTC Act.

Then comes Section 13(b), which authorizes suit under two conjunctive conditions. First is Section 13(b)(1)—the forward-looking provision. Second is Section 13(b)(2), which authorizes suit for a temporary restraining order when the commission has reason to believe there is, or is about to be, a violation of “any provision of law enforced by the Federal Trade Commission.” That same provision also states that “in proper cases the Commission may seek, and after proper proof, the court may issue, a permanent injunction.”

But the statute bundles Sections 13(b)(1) and 13(b)(2) together. It is Section 13(b)(1) and Section 13(b)(2)—both satisfied—that authorize the commission to go to court. It’s all Section 13(b), all the time. The subheading for Section 13(b) is somewhat misleading because it references “[t]emporary restraining orders; preliminary injunctions.” But there is no other relevant part of Section 13(b). You cannot get to Section 13(b)(2) without first passing through Section 13(b)(1), and both preliminary and permanent injunctions appear in Section 13(b)(2).

What about Section 13(b)(3)? Well, there “ain’t no sanity clause,” and there’s no such thing as Section 13(b)(3), either.

The whole business about “a different part of Section 13(b)” is a distraction. The claim that Section 13(b)(1) does not apply to that “different part” is wrong—or, in the language of Section 5, “deceptive” (that is, material and false or misleading).

The commission also tries to support its position with precedent, including the Supreme Court’s opinion in Alaska v. United States. The citation is strained, at best, and provides no support for the FTC’s conclusion that “[t]he permanent injunction proviso in Section 13(b) grants an additional power not contingent on anything that precedes it.” For that matter, neither Alaska nor the FTC’s accompanying citation to Beaty concerns the meaning or construction of any provision of the FTC Act, much less Section 13(b).

Enough of that aside. I don’t think the argument works. But I also think it’s something of a red herring.

The opinion does state that “[t]o win the permanent injunction that it seeks here, the FTC must prove a current or imminent legal violation.” That’s right there, in black and white, on page 23 of the 89-page memorandum opinion.

What difference that makes is a separate question. According to the FTC, Judge Boasberg’s reading “could improperly deprive courts of the power to remedy past unlawful action even if it threatens to recur or continues to cause grievous harm.”

I don’t think that’s right. And I’m not entirely sure the commission believes it, either.

The brief is clear enough—at least later in the game—that its authority to go directly to district court seeking a permanent injunction “in proper cases” comes from Section 13(b)(2). The two sitting commissioners, along with the staff in the FTC’s Bureau of Competition, are experienced attorneys and more than capable of reading the conjunction that joins Sections 13(b)(1) and 13(b)(2).

There may well be difficult questions of statutory interpretation lurking in this portion of the opinion. But Section 2 of the Sherman Act (allegedly violated by the Instagram and WhatsApp acquisitions) and Section 7 of the Clayton Act (typically at issue in merger challenges, although the FTC did not plead a Section 7 violation here) are both forward-looking. Yet neither bars challenges to consummated mergers.

As the FTC’s opening brief itself puts it, the relevant question is whether the conduct at issue is harmful and whether it “threatens to recur or continues to cause grievous harm.”

I don’t think there is any serious argument that Judge Boasberg was confused about that. Which suggests that the FTC’s appeal is really about a different question of timing.

The Timing Problem 

To get at that question, I’ll construct an abridged—if not brief (sorry)—timeline.

February 2004: Facebook launches (as “the Facebook”).

2006: Facebook continues expanding its reach, adding functionality, and developing its business model.

By the end of 2007, Facebook has surpassed MySpace in social-media traffic (a metric the FTC would later distinguish from its proposed PSN market).

2012: Facebook acquires Instagram. The proposed transaction is reported to the antitrust agencies and cleared to the FTC for review. Following an investigation, the FTC votes 5-0 to close the matter and issues a closing letter.

Yes, there were the supposed “hot docs.” But they weren’t that hot, and they didn’t—and shouldn’t—have made a case.

At the time of the acquisition, Instagram had just 13 employees, and its app was one of many photo-sharing apps on the market, including Google’s predecessor to Google Photos.

The staff—and the Commission—rightly concluded that this was a defensible buy-or-build decision to add functionality, not a “killer acquisition,” not a violation of Section 7 of the Clayton Act. I don’t recall whether any time was wasted considering a Sherman Act Section 2 complaint.

The FTC was hardly alone in that view. The UK Office of Fair Trading also reviewed the transaction, concluding that it “does not believe that it is or may be the case that the merger may be expected to result in a substantial lessening of competition within a market or markets in the United Kingdom.”

Had the commission possessed a crystal ball, it would have seen what we know now: there were still many photo-sharing apps in 2020, and there are still many today, all with far greater functionality than Instagram offered in 2012.

2014: Facebook acquires WhatsApp—roughly the same story, albeit with less drama. 

WhatsApp operated a paid messaging app. It was tiny and barely present in the U.S. market. There were many other messaging apps. 

The transaction was reviewed by the FTC, which again voted 5-0 to close the investigation. 

The European Commission likewise reviewed the deal, concluding that Facebook and WhatsApp were not close competitors and that the transaction posed no threat to competition in online advertising.

Had the commission possessed a crystal ball here, it would have found much the same result.

2020: On Dec. 8, the FTC filed a monopolization complaint alleging that the Instagram and WhatsApp acquisitions—transactions the commission itself had previously reviewed and declined to challenge—were key elements of a broader scheme of unlawful monopoly maintenance in violation of Section 2 of the Sherman Act.

FWIW, I don’t recall any economists in the bureau of economics—or anyone in my office, the Office of Policy Planning—who thought bringing the case was a good idea.

2021: Facebook moves to dismiss. On June 28, Judge James Boasberg dismisses the complaint, albeit with leave to amend. 

Reading the complaint now, it strikes me as rushed and, I’m sorry to say, not very good. Judge Boasberg was right to dismiss it.

Later in 2021, the FTC files an amended complaint and then a substitute amended complaint.

2022: Facebook again moves to dismiss. This time, Judge Boasberg concludes that the FTC has cleared the pleading bar, while cautioning that “Ultimately, whether the FTC will be able to prove its case and prevail at summary judgment and trial is anyone’s guess.” 

2024: The parties file cross-motions for summary judgment. On Nov. 13, Judge Boasberg largely denies both motions, and the case proceeds. 

2025: A six-week bench trial takes place in April and May. In November, Judge Boasberg rules for Meta. An unredacted version of the opinion follows on Dec. 2. 

AND THAT SHOULD HAVE BEEN THE END OF IT. 

But it brings us back to the timing issue, and to pages 22-24 of the opinion.

The language there, too, might be a bit confusing. The FTC is hanging its first—and primary—hat on this hook: “Throughout this case, the [district] Court has held that the agency must prove that Meta is violating the law now.” And that, the opinion says, should be distinguished from a finding that there is “lingering harm” from a past violation of the FTC Act.

The FTC and the opinion do not disagree about the high-level requirements for establishing a Section 2 violation. As the Supreme Court held in United States v. Grinnell Corp., “[t]o prove monopolization under Section 2, a plaintiff must show ‘(1) the possession of monopoly power in the relevant market and (2) the willful acquisition or maintenance of that power as distinguished from growth or development as a consequence of a superior product, business acumen, or historic accident.'”

The opinion maintains that the FTC must show monopoly power and willful (impermissible) maintenance of that monopoly as, for example, through exclusionary conduct. The FTC agrees. But while the opinion says the FTC had to make that showing in 2025—presumably at the time of trial, although the FTC conjectures that “now” means the time at which Judge Boasberg drafted or published his opinion—the FTC says no. According to the commission, the immediate implications of Section 13(b)(1) are merely pleading requirements and, in any event, do not bear on the FTC’s authority to seek a permanent injunction under Section 13(b)(2).

Rather, citing Grinnell, the FTC argues that “[t]he relevant market in a Section 2 case is the market at the time of the alleged anticompetitive conduct.”

The brief doesn’t put a date on that. Given that the Instagram and WhatsApp acquisitions were the conduct at issue, we might suppose 2012. Or 2014. Or the period between the two acquisitions. Or thereabouts. Again, the FTC doesn’t quite nail it down.

There is a line early in the brief stating that “as far as Section 13(b) is concerned, what matters is the . . . time of the complaint.” I think what the FTC means is that the time of the complaint governs pleading. If that’s right, there’s no tension between that statement and the brief’s citation to Grinnell. If that’s wrong, the appeal has a more serious problem.

That brings us to . . . 

The Minor Premise That Ate the Brief

Up top, I guessed that the FTC structured its brief as it did—and led with the argument about the proper construction of Section 13(b) of the FTC Act—because it knew that an appeal challenging findings of fact would be tough sledding. At best.

It really is a guess. I have no special insight into the thinking of the two sitting commissioners when the case first went to trial—Chairman Andrew Ferguson and former Commissioner Melissa Holyoak, now U.S. Attorney for the District of Utah. Commissioner Mark Meador was confirmed two days into the trial. And I have no special insight into the thinking of the Bureau of Competition personnel taking the laboring oar on appeal. 

But the guess might also explain why, if the “simple observation” about Section 13(b) “resolves the appeal,” the FTC spends so much time arguing against what it calls the opinion’s “minor premise”: “that Meta’s personal social networking monopoly had somehow dissipated between 2020 and 2025.” It also may explain why the FTC devotes pages 36-69 of its 69-page brief to the claim that “THE COMMISSION DEMONSTRATED THAT META HELD MONOPOLY POWER IN THE MARKET FOR PSN SERVICES.” 

That “minor premise” occupies a good deal more space than what the FTC casts as its main argument—the one that supposedly “resolves the appeal.” There is nothing improper about arguing in the alternative. But while the FTC does not quite present it as such, this second argument reads like an attempt to relitigate the merits—and the findings of fact—below.

The stated focus may be monopoly power. But the brief also addresses findings about the FTC’s personal-social-networking-services market definition, the FTC’s allegation that Facebook had monopoly power in that market, and the FTC’s allegations of competitive harm caused by the exploitation of that alleged monopoly power.

I will not rehash Judge Boasberg’s findings of fact. There were many. They are covered in some detail in the opinion, and I have sketched them already—both in anticipation, based on the pleadings, and in discussing the 2025 opinion itself. 

I do not argue that every point raised in this 30-plus-page section of the brief is wrong. But I do think key parts are wrong, that some of it misrepresents Judge Boasberg’s opinion, and that much of it is conclusory. It sketches evidence the FTC presented in its pleadings, pokes here and there at the court’s treatment of that evidence, and does little to refute or rehabilitate most of the shortcomings in the FTC’s own case. 

But to return to the timing issue so central to the first part of the FTC’s appeal: rather little in the FTC’s affirmative case seemed to address monopolization at the time the FTC says matters. As noted above, citing United States v. Grinnell Corp., the FTC argues that “[t]he relevant market in a Section 2 case is the market at the time of the alleged anticompetitive conduct.” And while the agency does not quite put a date on it, the mergers alleged to be anticompetitive under Section 2 were the chief courses of “conduct” in the FTC’s allegations, and they were  noticed, screened by the FTC, and consummated in 2012 (Facebook/Instagram) and 2014 (Facebook/WhatsApp). 

The second part of the FTC’s brief argues that the 2025 decision against the agency turns on the notion “that Meta’s personal social networking monopoly had somehow dissipated between 2020 and 2025.” But neither 2012 nor 2014 falls within that interval. Having a monopoly does not violate Section 2. And most of the allegations in the FTC’s complaint—and most of the evidence the FTC presented at trial—addressed or purported to address conditions many years after the acquisitions.

Of course, evidence about competitive conditions, including competitive harm, might bear on “lingering effects” or the long-term consequences of the acquisitions. But beyond conclusory allegations and selected “hot docs,” there is not much that goes to whether the FTC’s personal-social-networking-services market definition, first introduced in 2020, was viable—much less correct—in 2012. Nor is there much to show that Facebook had monopoly power in such a market in 2012, or that any such monopoly power was achieved by dint of the Instagram acquisition. With respect to Instagram, 2012 is “the time of the alleged anticompetitive conduct.” 

The FTC could, of course, argue that Facebook later engaged in unlawful exclusionary conduct, given the 2012 and 2014 acquisitions. And it is not as if the agency has said nothing on that score. But it did not say—never mind prove—all that much.

As I have covered before, the FTC’s economic witness, C. Scott Hemphill, raised some good points about certain issues in Meta’s defense, but he offered a mysteriously weak—indeed, dubious—account of the FTC’s case in chief. The PSN market definition seemed, and still seems, gerrymandered: the product of a hash of the subjective and long-discredited “Brown Shoe factors.” As Herbert Hovenkamp has observed, “[t]he so-called Brown Shoe factors are wrong in most cases” and, in any case, even as described by the Supreme Court in its poorly aging opinion, “there was no reason to think that it was doing anything more than summarizing fact findings for that particular case.” 

And the “indirect evidence” of monopoly power that turned on that hash was no better. Neither was the “direct evidence” of monopoly power presented in the alternative, which tried to hang a story about competitive harm on a weak claim about declining product quality, based solely on ad load—and no other nonprice factors—rather than price or output effects. 

Judge Boasberg rightly identified some key flaws in that evidence, which, not incidentally, was not about monopoly power in 2012 or 2014 either. 

As for the selective hot docs: feh. They are mostly about alleged anticompetitive intent and easily misread. For a substantial discussion of the background issue, I will point you to Geoff Manne and E. Marcellus Williamson’s article, “Hot Docs vs. Cold Economics: The Use and Misuse of Business Documents in Antitrust Enforcement and Adjudication.” For now, I will simply quote part of my earlier assessment:

The problem with email generally is that C-suite executives, like kids, can say the darndest things. Indeed, they should say and consider many things. Selected excerpts from a multitude of documents (and discovery yields multitudes) might suggest any number of things about intent. What dated excerpts show about current intent is anyone’s guess.

More than that, the excerpts in question seem suggestive of complex concerns and interests, just as one might expect. It’s not at all obvious that they signal an intent to acquire instead of competing. The deals were not “killer acquisitions”; Instagram and WhatsApp were nurtured, not killed.

And, of course, while email “hot docs” evidence “is not necessarily irrelevant, to the extent it might provide a circumstantial bolster for allegations of effects,” it is also true that:

effects matter directly, and intent is not an element of a Section 2 claim: there’s no mens rea issue here. Moreover, a decade or more post-merger, that sort of evidence can cut both ways, as it underscores the question why there isn’t better direct evidence of competitive effects. 

For a bit more, see Will Rinehart’s 2023 discussion and, while we’re at it, David Balto’s 2014 post, “Spicy Documents Serve up a Paltry Antitrust Meal.”

Knowing When to Fold

No doubt the commission can answer the question “why appeal?” by saying that it is right on both the law and the facts and, in particular, in alleging ongoing harm to competition and consumers flowing from the allegedly anticompetitive acquisitions. It can say that reversal is necessary to reach the correct liability determination and, in turn, a remedy that will set things right. 

Right. 

Then again, the puzzle of persistence is not merely that people stick with a losing hand, sunk costs notwithstanding. It is that we sometimes convince ourselves that we should stick with a losing hand, even when we should know better. 

The FTC might also argue that it seeks to correct a dangerous precedent regarding its authority under Section 13(b) of the FTC Act. But the district court’s holding in FTC v. Meta is not binding precedent on any other federal court or, for that matter, even on the U.S. District Court for the District of Columbia in future cases. It is, at most, persuasive authority. That’s not nothing, but it is less. 

Still, I’m left wondering why the commission thinks this is a good case when it really wasn’t. The case was weak on the pleadings and weaker still at trial. And it was weak quite apart from the correct construction of Section 13(b). 

I’m also left with concerns I’ve had all along.

One is that a bad liability decision—a false positive—is likely to produce remedies that do more harm than good. That means harm not merely to Meta as a competitor, but to competition itself and, critically, to consumers. 

Moreover, while the district court’s decision establishes no new legal precedent, a successful appeal could. I don’t think that’s very likely, but I don’t know enough to put a meaningful subjective “probability” on my own assessment. 

And ultimately, I think this is a substantial waste of scarce resources.

Current FTC leadership has rightly emphasized that the agency must exercise discretion in choosing cases. The FTC’s jurisdiction is astonishingly broad, and its resources are limited. That reality applies not only to bringing cases in the first instance, but also to deciding which losses are worth appealing. 

The work already devoted to this appeal could have been put to more productive use. And the appeal is not over. More time. More resources. 

And if the FTC succeeds, then what?

A reversal and remand would send the case back to the district court. More process. More resources. And all of it would carry us further down the timeline, into 2027 and perhaps beyond, in a market that has remained dynamic and differentiated, and in a case that, as Judge Boasberg rightly recognized, was always about dynamic and differentiated competition. 

What are the chances of a procompetitive remedy at such a remove from the acquisitions in question—even assuming, for the sake of argument, that the FTC was right to file the complaint and wrong to clear the mergers in 2012 and 2014? 

I think the agency should drop it.

It won’t.

But it should.

The post The FTC’s Sunk-Cost Social Network appeared first on Truth on the Market.

That is where antitrust standing does its work. In ordinary litigation, standing asks whether a plaintiff has enough of a stake in the dispute to be in court at all. Antitrust standing asks a more pointed question: Is this the kind of plaintiff, and the kind of injury, the antitrust laws allow to recover damages?

That inquiry often gets treated as a threshold pleading hurdle—a box plaintiffs satisfy almost automatically once they allege anticompetitive conduct and economic harm. But modern antitrust doctrine demands more. The question is not merely whether a plaintiff lost money. It is whether that loss reflects diminished competition in the allegedly restrained market.

Section 4 of the Clayton Act provides that any person injured in their business or property “by reason of anything forbidden in the antitrust laws” may sue and recover treble damages—three times the damages proved. Read literally, that language sounds expansive. Courts have never read it that way. 

Instead, plaintiffs must show, as the Supreme Court put it in Brunswick Corp. v. Pueblo Bowl-O-Mat (1977), that their injury is “of the type the antitrust laws were intended to prevent and that flows from that which makes defendants’ acts unlawful.” 

The reason is straightforward: antitrust law protects competition, not competitors. A business harmed by rivals acting badly has not necessarily suffered an antitrust injury. The harm must reflect damage to the competitive process in a relevant market. Antitrust standing also serves a practical purpose: limiting exposure to treble damages in private enforcement actions before liability sprawls beyond any administrable boundary. 

Two recent proceedings show how much work this doctrine still does. The first is X Corp. v. World Federation of Advertisers, in which Judge Jane Boyle dismissed Elon Musk’s lawsuit against an advertising-industry coalition, holding that X Corp. failed to plead antitrust injury despite extensive allegations of coordinated advertiser boycotts. The second is the Live Nation-Ticketmaster litigation, where a federal jury found Live Nation and Ticketmaster liable for antitrust violations in a venue-facing ticketing market. That verdict raises a harder follow-on question: Can downstream consumers—and states suing on their behalf—recover federal damages for harm that began upstream? 

The cases arise in very different settings. But they expose the same doctrinal tension: antitrust liability and antitrust recovery are not the same thing. Conduct may violate the Sherman Act while a particular plaintiff’s injury remains too remote to support damages under Section 4 of the Clayton Act. In both X Corp. and Live Nation, the central question became whether the plaintiffs’ losses flowed from harm to competition, or merely from conduct alleged to be anticompetitive. 

Not Every Wound Is an Antitrust Injury

In X Corp., X—formerly Twitter—alleged that a coalition of major advertisers, coordinated through the Global Alliance for Responsible Media (GARM), an initiative of the World Federation of Advertisers (WFA), orchestrated a group boycott of the platform after Elon Musk acquired Twitter in 2022. X claimed that GARM members, including Mars, CVS, Colgate, Nestlé, Abbott, and others, agreed collectively to withhold advertising unless X complied with brand-safety standards developed by GARM. According to X, the resulting advertising withdrawal cost the company hundreds of millions of dollars in revenue.

The district court nevertheless dismissed the case with prejudice, meaning X could not simply refile the same claims. The court did not dispute that X had suffered economic loss sufficient to establish injury in fact—the basic injury needed for constitutional standing. Instead, it held that X had not alleged the type of injury the antitrust laws are designed to remedy.

According to the court, X alleged lost advertising revenue and reduced fees for advertising placements. Those are real business harms. But citing the 5th U.S. Circuit Court of Appeals’ decision in Rx Solutions v. Caremark, the court noted that classic antitrust injuries typically involve higher prices, reduced output, diminished quality, or some other harm to the competitive process.

X argued that it need not satisfy the antitrust-injury requirement because the alleged boycott was a per se antitrust violation. A per se violation is conduct deemed so likely to harm competition that courts condemn it without the usual detailed market analysis. The court disagreed. Even per se violations require plaintiffs to show antitrust injury before they may recover damages.

Illegal Doesn’t Mean You Can Sue

Assume, for the sake of argument, that the WFA-GARM boycott satisfied the threshold conditions for per se condemnation. That is far from certain. Under Northwest Wholesale Stationers (1985), not all group boycotts are per se Sherman Act violations. But even if X cleared that hurdle, its claims faced a separate and fatal defect: X failed to allege antitrust injury.

That distinction matters. Per se treatment is a rule of liability. It shortcuts the market analysis ordinarily needed to prove that the defendant’s conduct harmed competition. It does not decide whether a particular plaintiff has standing to seek private damages.

As the Supreme Court held in Atlantic Richfield v. USA Petroleum (1990): 

A loss flowing from a per se violation of § 1 does not automatically satisfy the antitrust injury requirement, which is a distinct matter that must be shown independently. The purpose of per se analysis is to determine whether a particular restraint is unreasonable. Actions per se unlawful may nonetheless have some procompetitive effects, and private parties might suffer losses therefrom. The antitrust injury requirement, however, ensures that a plaintiff can recover only if the loss stems from a competition-reducing aspect or effect of the defendant’s behavior.

In other words, conduct may be per se unlawful while a particular plaintiff’s losses still fall outside antitrust injury. The plaintiff’s harm must flow from the anticompetitive mechanism that makes the conduct illegal.

Consider a simple example. Suppose several major steel manufacturers enter a per se unlawful price-fixing conspiracy, agreeing to raise the price of industrial steel sold to automakers. The cartel is plainly illegal under Section 1 of the Sherman Act because horizontal price fixing—competitors agreeing on price—is presumed to harm competition in the steel market.

Now suppose higher steel prices lead automakers to reduce production and buy fewer trucking services to transport finished vehicles. A trucking company loses substantial business and suffers serious economic harm.

The trucking company’s losses are real, and they are causally connected to the unlawful cartel. But the company has not suffered antitrust injury. Its harm does not flow from the anticompetitive mechanism that makes the cartel unlawful: the suppression of price competition in the steel market. 

As the Supreme Court observed in 1982’s Blue Shield of Virginia v. McCready, an antitrust violation may send ripples of harm through the economy. But liability cannot extend forever. At some point, the chain becomes too attenuated.

The Court permitted standing in McCready because the plaintiff’s injury—Blue Shield’s refusal to reimburse her for psychotherapy sessions with a psychologist rather than a psychiatrist—was not merely a side effect of the alleged conspiracy. It was the very instrument through which Blue Shield allegedly sought to achieve its anticompetitive ends.

The trucking company stands in a materially different position. Its injury is remote and consequential. That is not enough to establish antitrust standing under Associated General Contractors v. Carpenters (1983). 

Other factors point the same way. The trucking company is neither a consumer nor a competitor in the restrained market—the two categories antitrust-standing doctrine most readily recognizes. Allowing recovery also would risk duplicative damages or force courts to apportion damages across multiple layers of claimants. More direct victims exist: the automakers who bought steel at cartel-inflated prices. Their claims would more cleanly serve the deterrent goals of private antitrust enforcement.

The steel cartel may be per se unlawful. The trucking company still lacks antitrust injury.

Standing Takes Center Stage

Antitrust injury also played a central role in the Live Nation-Ticketmaster litigation. In 2024, the U.S. Justice Department (DOJ) and 33 state attorneys general sued Live Nation and Ticketmaster under Sections 1 and 2 of the Sherman Act, alleging unlawful agreements and monopolization. The DOJ proceeded under Section 4 of the Sherman Act, while the states also invoked Section 4C of the Clayton Act and various state antitrust statutes. 

By summary judgment—the stage at which a court decides whether claims can proceed to trial—the district court had substantially narrowed the case. Among other claims, it dismissed allegations involving fan-facing ticketing markets. Two theories survived: claims involving the venue-facing primary-ticketing market, in which venues purchase ticketing services directly from Ticketmaster; and claims that Live Nation unlawfully tied concert-promotion services to access to its amphitheaters. 

After the DOJ settled midway through trial, several states continued the case and secured a jury verdict finding Live Nation liable on all remaining claims. The jury awarded damages based on an estimated average overcharge of $1.72 per ticket across 22 states.

Live Nation then renewed its motion for judgment as a matter of law, asking the court to set aside the verdict. Among other arguments, Live Nation contended that the states lacked a cognizable antitrust injury and were not proper—or at least efficient—enforcers under federal antitrust-standing doctrine.

Standing in the Cheap Seats

Section 4 of the Clayton Act raises two distinct questions the Supreme Court highlighted in  Illinois Brick v. Illinois (1977). First, which plaintiffs have suffered injuries too remote to recover damages? Second, which plaintiffs paid the unlawful overcharge in the legally relevant transaction? 

To answer the first question, the district court relied heavily on McCready. There, the Supreme Court held that a plaintiff may suffer antitrust injury even without directly participating in the restrained market, so long as she falls within “that area of the economy endangered by the breakdown of competitive conditions.” The district court concluded that Live Nation was on all fours with McCready. That conclusion is difficult to sustain.

In McCready, the alleged conspiracy involved a concerted refusal to reimburse patients who sought psychotherapy from psychologists rather than psychiatrists. The Court found that the refusal was the mechanism through which Blue Shield allegedly implemented the conspiracy. McCready’s injury was therefore inextricably linked to the harm the conspirators sought to inflict on psychologists and the psychotherapy market. Her injury was not merely foreseeable. It was “integral” to the scheme.

The fans’ position is materially different. Their alleged injury is not the instrument through which Live Nation purportedly advanced its anticompetitive objectives. Nor is it an integral feature of the alleged scheme. It is downstream harm.

The distinction becomes clearer when viewed through the structure of the relevant market. Unlike McCready, ticket-buying fans are neither consumers nor competitors in the market allegedly restrained by the conspiracy. The direct purchasers of venue-ticketing services are venues. The relevant competitors are firms such as SeatGeek and AXS. Fans sit one step further away. They are end users of a product whose upstream supply conditions are the subject of the alleged restraint.

At most, fans can show consequential harm flowing from conduct directed at others. That is a much weaker basis for antitrust standing.

The Supreme Court emphasized this point in Associated General Contractors. The case for extending standing to a more remote plaintiff weakens substantially when there exists “an identifiable class of persons whose self-interest would normally motivate them to vindicate the public interest in antitrust enforcement.” Such a class is readily identifiable here. Venues, along with Live Nation’s and Ticketmaster’s direct competitors, suffered the alleged harm more directly and have both the injury and the incentive to litigate. Fans occupy a derivative position.

The existence of more direct victims also highlights the practical dangers of extending standing further down the chain. When plaintiffs closer to the alleged violation can enforce the antitrust laws, allowing more remote parties to recover creates a substantial risk of duplicative damages and the complex apportionment problems that Illinois Brick and Associated General Contractors sought to avoid.

Taken together, the remoteness of the fans’ alleged injury, the availability of more suitable plaintiffs, and the risk of duplicative recovery all weigh against recognizing standing for the states’ federal claims.

The Illinois Brick Wall

A plaintiff may satisfy McCready and still be barred by Illinois Brick because it was not the direct purchaser in the market where the anticompetitive conduct occurred. Thus, the district court’s reliance on McCready resolves only the first question: whether concertgoers suffered an injury sufficiently connected to the alleged violation to satisfy antitrust-injury principles. The second question—who was injured by the alleged overcharge for purposes of Section 4 standing—is governed by Illinois Brick and Kansas v. UtiliCorp United (1990). Live Nation and Ticketmaster arguably should have pressed that question more aggressively.

In Illinois Brick, the Supreme Court held that only direct purchasers may recover federal antitrust damages. Indirect purchasers—those who bear overcharges passed through an intermediary—cannot recover under federal law. 

The Court reaffirmed that principle in UtiliCorp, holding that Section 4C did not alter the Illinois Brick rule for parens patriae actions. A parens patriae action allows a state to sue on behalf of its residents. Section 4C, the Court explained, “did not establish any new substantive liability, but simply created a new procedural device to enforce existing rights of recovery under Section 4.” States therefore could not recover on behalf of residential utility customers who merely absorbed overcharges passed through from upstream transactions.

The Live Nation ticketing structure bears a strong resemblance to the arrangement at issue in UtiliCorp. The face value of a ticket is set by the event organizer. The facility charge is set by the venue, and Ticketmaster retains none of it. The service fee is negotiated between Ticketmaster and the venue and then divided between them. 

In other words, the all-in ticket price reflects several upstream pricing decisions. Any supracompetitive overcharge attributable to Ticketmaster would therefore appear to pass through the venue before reaching the fan. 

Milton Handler and Michael Blechman warned of precisely this problem in their influential critique of parens patriae actions, which the Supreme Court cited approvingly in Illinois Brick. They observed that many antitrust violations occur at levels of the distribution chain “so remote from consumers as to make it virtually impossible . . . to prove that they suffered any compensable injury at all.” 

Once multiple intermediaries and independent pricing decisions separate the alleged restraint from the consumer transaction, determining whether—and to what extent—an overcharge reached consumers becomes less a matter of proof than of speculation.

As Handler and Blechman explained, establishing consumer injury in such settings requires tracing how an alleged overcharge moved “from one level to another in the chain of distribution” through a series of independent economic decisions that courts are poorly equipped to reconstruct after the fact.

The Live Nation ticketing system presents many of the same difficulties. Even assuming anticompetitive conduct in the venue-facing ticketing market, the states’ damages theory requires a court to determine how much of any allegedly supracompetitive fee negotiated between Ticketmaster and venues ultimately reached consumers through a pricing structure shaped by promoters, venues, facility charges, dynamic-pricing practices, and revenue-sharing arrangements.

That is exactly the sort of speculative pass-through inquiry that Illinois Brick and UtiliCorp sought to avoid by limiting federal antitrust-damages recovery to direct purchasers.

Direct Purchasers of What, Exactly?

A possible response is that Apple v. Pepper (2018) points in the opposite direction. There, the Supreme Court held that consumers who purchased apps through Apple’s App Store were direct purchasers for purposes of Illinois Brick. 

The Court emphasized the direct transactional relationship between Apple and iPhone users. Consumers bought apps directly from Apple, paid Apple directly, and Apple collected the allegedly monopolistic commission directly through those transactions. The Court expressly rejected the argument that Illinois Brick turns on who sets the retail price. Although app developers set app prices, consumers were still direct purchasers.

Instead, the Court treated Illinois Brick as a bright-line rule that asks a simple question: from whom did the plaintiff purchase the product or service? Because iPhone users purchased apps directly from Apple, with no intermediary between buyer and seller, they had standing to sue.

That reasoning does not necessarily rescue the states’ claims against Live Nation and Ticketmaster. The question is not whether fans transact with Ticketmaster at checkout. They do. The question is whether fans purchased the product sold in the market allegedly restrained by the anticompetitive conduct.

The alleged restraint occurred in the venue-facing primary-ticketing market. Fans do not participate in that market. Venues purchase ticketing services from Ticketmaster. Fans purchase admission to live events. Those are distinct products sold in distinct markets.

Viewed this way, the Illinois Brick problem is not that venues stand between Ticketmaster and fans as traditional reseller intermediaries. It is that fans never purchased the allegedly restrained product in the first place. Their injury, while potentially sufficient under an expansive reading of McCready, flows downstream from a market in which they were never buyers.

Accordingly, fans fall outside the class of direct purchasers that Illinois Brick and UtiliCorp require for federal damages recovery under Section 4 of the Clayton Act. If that analysis is correct, states likewise cannot maintain a parens patriae action on their behalf under Section 4C.

The practical answer is that the states did not rely solely on federal law. They also asserted claims under state antitrust statutes, several of which contain Illinois Brick repealer provisions that permit indirect purchasers to recover damages under state law. The jury’s $1.72-per-ticket damages award therefore likely rests on those state-law theories rather than on federal Section 4C claims, which UtiliCorp appears to foreclose.

Liability Is Not Standing

Both cases turn on the same question: Can the plaintiff’s loss be tied directly to a competitive mechanism the antitrust laws protect?

In X Corp., that connection was missing from the pleadings. WFA and GARM allegedly coordinated buyers, but they did not control a supply-side chokepoint, and X’s lost revenue did not flow from a restraint on the competitive structure of any market. Antitrust-injury doctrine therefore ended the case at the pleading stage.

In Live Nation, the competitive mechanism exists in the venue-facing ticketing market. But the path from that upstream restraint to downstream consumer damages runs through antitrust-standing doctrine—specifically, whether fans who did not purchase in the restrained market may recover federal damages through the states’ parens patriae claims.

The lesson is the same in both cases: antitrust liability and antitrust recovery ask different questions. A plaintiff can identify real harm, describe anticompetitive conduct, and still fail. Harm to a competitor is not necessarily harm to competition. Downstream injury is not direct-purchaser standing.

Antitrust law does not compensate every casualty of bad conduct. It compensates injuries to competition—and only the plaintiffs close enough to prove them.

The post Antitrust Standing Room Only appeared first on Truth on the Market.

The ruling is narrow. It concerns only Facebook Marketplace, and it turns largely on procedural rather than substantive grounds. But narrow rulings can still cast long shadows, and this one does. The court’s reasoning reaches beyond Marketplace to questions that will shape future DMA enforcement. 

This post argues that the decision’s greatest importance lies not in what it says about Facebook Marketplace, but in what it signals for the DMA’s future application. After outlining the case’s background, it explores three implications: for qualitative gatekeeper designations, for the DMA’s “gateway” logic, and for the wisdom of applying the DMA to challengers rather than incumbents. 

When Process Becomes Substance

The DMA applies only to “core platform services” designated as “gatekeepers.” A gatekeeper is a platform that exerts a significant impact on the European Union internal market, serves as an important gateway between businesses and end users, and enjoys an entrenched market position. The statute presumes these criteria are satisfied when a platform meets certain quantitative thresholds, such as turnover, market capitalization, and the number of business and end users. If those thresholds are not met, the European Commission must establish gatekeeper status through a qualitative assessment. 

The General Court’s judgment caps three years of DMA proceedings concerning Facebook Marketplace. In 2023, shortly after the DMA took effect, the Commission designated several Meta services—including Marketplace—as gatekeepers. 

A gatekeeper designation is not necessarily permanent, however. In 2024, Meta asked the Commission to reconsider Marketplace’s designation. It pointed to three changes designed to alter Marketplace’s role as a gateway between businesses and end users. Most importantly, Meta limited Marketplace to transactions between end users, excluding businesses from the platform. It also removed indications that businesses could use the service and stepped up enforcement of its terms. As a result, the number of users operating as businesses remained minimal and fell below the DMA’s quantitative thresholds. 

In light of those changes, the Commission repealed Marketplace’s gatekeeper designation in 2025. Meta nevertheless continued to seek annulment of the original designation because repeal alone did not restore it to the position it occupied before the Commission’s decision. In particular, Meta argued that the Commission should have considered the Marketplace changes during the original designation process, which might have led it not to designate the service at all. 

The General Court agreed. By annulling the decision, the court confirmed—with ex tunc effect—that the designation was unlawful because of the temporal scope of the evidence the Commission considered. Put simply, the judgment clarifies that a qualitative gatekeeper designation must be based on the facts and circumstances that exist at the time of designation. That differs from designations based on the DMA’s quantitative presumptions, which rely on data from the preceding three years. 

To borrow a phrase from James Venit, the judgment comes close to providing a “procedural answer to a substantive question.” Formally, the court annulled the Commission’s decision because it failed to provide adequate reasons and thereby prevented effective judicial review of the evidence presented by the parties. Even so, the ruling has potentially significant implications for the future direction of DMA enforcement. The next section highlights three of them. 

When Gatekeepers Aren’t Gateways

First, the judgment will shape the ongoing DMA investigations into cloud services. The European Commission is currently investigating whether certain cloud providers—especially Amazon Web Services and Microsoft Azure—should be designated as gatekeepers under the DMA. If those services fall short of the DMA’s quantitative thresholds, the Commission will have to rely on a qualitative assessment. 

That is where the General Court’s judgment matters. The court made clear that qualitative designations are not tethered to data from the preceding three years. They must reflect the facts as they stand at the moment of designation. In any future cloud designation, the Commission’s treatment of ongoing market developments will therefore face close judicial scrutiny. 

Second, the judgment supports the view that cloud services sit uneasily with the DMA’s regulatory logic. The DMA was designed for platforms that operate as “gateways” between businesses and end users. Marketplace’s gatekeeper designation was reversed precisely because Meta eliminated the possibility that businesses could operate there as business users. The Commission therefore concluded that “Marketplace should no longer be considered as an important gateway for business users to reach end users.” 

It is similarly difficult to describe cloud services as gateways in the DMA’s sense. They intermediate much like cable providers do: they provide infrastructure, but they do not typically connect business users to end users in the way the DMA contemplates. Read in this light, the General Court’s insistence that a gatekeeper must intermediate between businesses and end users sets a standard the Commission may struggle to satisfy for cloud. 

Third, and finally, comes the policy question: What did this judgment—and the Commission decisions that preceded it—actually achieve? 

In practice, Facebook Marketplace’s gatekeeper designation was reversed only after Meta removed the feature that made Marketplace a competitive presence on the business-user side. Once businesses could no longer operate on the platform, whatever competitive pressure Marketplace exerted on stronger marketplaces disappeared. The DMA proceeding thus likely made online marketplaces less contestable, not more. 

Unfortunately, this is not an isolated episode. The DMA tends to keep titans in quarantine. But the most effective competitive constraint on a large firm often comes from another firm of comparable scale. Yet DMA enforcement has once again taken aim at challengers rather than incumbents. TikTok was the first casualty in this campaign. Marketplace appears to be the latest. 

Similar outcomes can be avoided. The DMA is the law of the land, but the Commission retains meaningful administrative discretion in how it enforces the statute. It should wield that discretion with a clearer account of enforcement’s intended and unintended consequences. 

That does not require giving challengers to larger firms carte blanche. When the DMA is a poor fit as a matter of enforcement policy, competition law can fill the gap more surgically—just as the Commission did against Facebook Marketplace in 2024. 

The Cost of Keeping Titans in Quarantine

The real significance of Meta Platforms v. Commission may lie less in Facebook Marketplace than in what it signals for future DMA enforcement. By insisting that qualitative designations rest on the facts as they stand at the moment of designation, and that gatekeepers genuinely function as gateways between businesses and end users, the General Court has raised the bar for the cloud designations now on the horizon. 

The broader lesson concerns enforcement strategy. The DMA’s tendency to keep titans in quarantine risks blunting the very competitive pressures it was designed to protect, particularly when enforcement falls on challengers rather than incumbents. The law may be settled, but its application remains a matter of choice. As the Marketplace saga illustrates, today’s enforcement decision can become tomorrow’s lost competitive constraint.

The post The EU’s Facebook Marketplace Decision: The Gatekeeper That Wasn’t appeared first on Truth on the Market.

The White House’s new AI executive order is framed around innovation and security. Its clearest signal, however, is concern about frontier AI capabilities themselves: how they are benchmarked, who gets early access to them, and how their release should be coordinated with the federal government. That focus lands squarely in the middle of a growing debate over whether AI policy should target chips, models, or what Anthropic recently described as the “capability layer”—the environment where models are deployed, monitored, secured, and used.

Recent arguments within parts of the AI-policy community—most notably from Anthropic—have emphasized export controls, compute restrictions, and centralized governance as the primary tools for preserving U.S. leadership in frontier artificial intelligence. The concern is understandable. AI systems are likely to accelerate innovation across strategically important sectors, including semiconductors, cybersecurity, biotechnology, advanced manufacturing, and military systems. Maintaining American leadership in those areas is a legitimate national objective.

Much of the current debate, however, still views the strategic landscape primarily through the lens of chip denial and hardware restrictions. That perspective risks overstating the long-term importance of compute controls, while understating the significance of where AI systems actually meet the world: through cloud services, application programming interfaces (APIs), identity systems, monitoring tools, and security controls.

As models proliferate, open-weight ecosystems mature, and agentic orchestration techniques improve—that is, as systems become better at coordinating multiple tools and models to complete tasks—the core strategic challenge increasingly shifts. The question is less whether adversaries can ever obtain compute and more whether deployed systems can be hardened against misuse, monitored effectively, and integrated into resilient defensive architectures.

In practice, the emerging equilibrium looks far more like cybersecurity or fraud prevention than a traditional nonproliferation regime. Overindexing on the latter risks weakening both American national security and the ability of U.S. firms to remain at the technological frontier.

Put differently, current AI policy discussions too often blur the line between national security and incumbent protection. A sustainable strategy should preserve U.S. ecosystem leadership, focus governance on the places where models are actually accessed and abused, and encourage defensive open-source proliferation. Treating blunt hardware denial as the master key to AI security is unlikely to achieve any of those goals.

A Very Expensive Speed Bump

In a recent essay on AI competition with China, Anthropic argued that U.S. export controls on advanced AI chips have been highly effective. It suggests that Chinese firms have narrowed the gap primarily through talent, loopholes in export-control enforcement, and large-scale distillation attacks on frontier U.S. models. Distillation, in this context, means copying some of a model’s capabilities by systematically querying it and training another model on the results.

There is at least some evidence that export controls have imposed real friction on Chinese AI development, particularly by limiting access to leading-edge semiconductor manufacturing equipment and hyperscale computing resources. The controls have increased costs, reduced efficiency, and slowed progress at the frontier.

But slowing progress is not the same as stopping it. Substantial leakage, smuggling, stockpiling, and substitution remain. The restrictions have also encouraged “good enough” domestic alternatives, such as Huawei chips, that can leverage China’s abundant energy supplies to scale training workloads. In plain English: better chips do more work with less power, but weaker chips paired with enough electricity can still produce highly capable systems. 

A recent Federal Reserve Bank of New York staff report adds an important domestic-cost dimension. Affected U.S. suppliers complied with the controls by reducing sales to Chinese customers, but many struggled to replace those relationships through reshoring or friendshoring. Meanwhile, targeted Chinese firms developed alternative supplier networks. The report estimates that affected U.S. suppliers lost roughly $130 billion in market capitalization, underscoring that export controls can impose substantial costs on the same firms they are supposed to protect. 

The farther one moves from training frontier models from scratch, the less persuasive the hardware-denial framework becomes. At this point, enough advanced compute already exists around the world—and much of it is accessible remotely—that comprehensive denial is difficult to imagine. Even maintaining enough friction to preserve a meaningful gap across every relevant frontier appears increasingly implausible.

Perhaps export controls would have fundamentally altered the strategic landscape had they been implemented a decade earlier. But that counterfactual would also have changed the trajectory of American industry itself, making the ultimate outcome impossible to know. In any event, the world now contains a vast installed base of capable hardware, mature open-weight ecosystems, sophisticated distributed-training techniques, and increasingly commoditized model architectures. Those realities change the security problem.

Jeffrey Ding makes a related point. Anthropic’s short-horizon scenario depends on a contestable assumption about technological diffusion: general-purpose technologies often take years, if not decades, to translate into organizational and military advantage. That uncertainty matters because the longer the relevant timeline, the more significant the downsides of export controls become, including indigenous substitution, supply-chain reorientation, and permanent parallel technology stacks.

The more important question is no longer whether adversaries can access compute. It is whether the systems built on that compute can be hardened, monitored, defended, and instrumented effectively.

The Fight Has Moved Up the Stack

One revealing passage in Anthropic’s essay says that “[p]olicymakers have not tightened loopholes on the CCP’s access to compute.” But “access to compute” increasingly means far more than access to physical chips. It includes hosted AI services, distributed cloud infrastructure, stolen or synthetic identities used to evade know-your-customer (KYC) controls, and criminal proxy markets that can generate thousands of fraudulent accounts.

It also includes orchestration frameworks that combine smaller or older models into coordinated systems, as well as access to the outputs of frontier models through distillation attacks. In other words, the relevant threat is not always someone buying forbidden chips. Sometimes it is someone abusing a deployed model at scale.

Ironically, Anthropic’s own essay acknowledges this problem. It correctly identifies organized efforts to extract capabilities from hosted systems through mass account generation, structured querying, and automated harvesting.

But that observation undercuts the claim that hardware controls are the decisive battleground. If the principal threat vector is adversarial interaction with deployed systems, then the strategic center of gravity shifts upward from the chip layer to the capability layer. That is where the real contest increasingly lives.

This shift also exposes an application programming interface (API) blind spot. An API is the software doorway that lets outside users and applications interact with a model. Advanced capabilities can be reached globally through account arbitrage, distillation, synthetic identities, and evasion of customer-screening protocols. More rigorous end-use monitoring for APIs is technically possible, but commercially uncomfortable for high-growth labs. It requires limiting the same global access and revenue streams that make hosted frontier models so valuable.

That incentive helps explain why compute controls are often easier to champion politically than strict, auditable controls on model access. It is simpler to talk about denying chips than to explain how one will monitor millions of users, accounts, prompts, integrations, and downstream applications without crushing legitimate use.

Nonetheless, the emerging AI security architecture increasingly resembles cybersecurity, fraud prevention, anti-money-laundering systems, and counterintelligence operations more than traditional nonproliferation regimes. In this world, the key question is not simply who owns chips or trains model weights. It is who can access deployed systems, how they use them, and whether defenders can detect abuse in time.

That makes anomaly detection, threat telemetry, identity verification, and orchestration controls central. Unsurprisingly, this is where frontier firms are already pouring enormous effort.

The AI arms race is likely to resemble cybersecurity itself: a perpetual contest between attackers and defenders in which no side ever achieves total control. That does not make hardware irrelevant. But it does mean the decisive terrain is shifting away from pure hardware denial.

Dual Use Is Not a Blank Check

Another theme running through Anthropic’s essay is the claim that AI is a “dual-use technology.” That characterization is not wrong, per se, but it risks doing too much rhetorical work. Every sufficiently general technology is dual use. Electricity powers hospitals and weapons factories. Cryptography protects dissidents and criminal enterprises. The internet enables education, commerce, espionage, and fraud.

The existence of dual-use risks does not, by itself, justify highly centralized control. Nor does invoking an “arms race” automatically establish the case for expansive regulatory authority over general-purpose intelligence systems.

One danger in today’s policy debate is that legitimate security concerns become the rationale for an increasingly centralized, government-defined AI ecosystem. Such a regime would do more than limit misuse. It would shape which systems can be built, who may build them, which values become embedded in them, who may access them, and even which forms of machine cognition remain permissible.

Centralization also creates a straightforward market-structure problem. Licensing regimes, rigid compute thresholds, and data-center moratoria do not affect all market participants equally. They tend to favor firms that already possess capital, infrastructure, legal resources, and regulatory access, while raising barriers for startups, open-source communities, and downstream developers.

Even when framed as safety measures, these policies can function as ladder-pulling mechanisms, converting legitimate national-security concerns into incumbent protection. The risk is particularly acute in AI because the technologies, business models, and competitive landscape remain highly fluid.

History offers little reason to believe governments excel at managing rapidly evolving technological ecosystems through centralized control. More importantly, centralized institutions tend to outlive the emergencies that justified their creation. Temporary powers have a habit of becoming permanent ones.

Policy should also avoid treating China—or any other adversary—as a monolith. Private firms, state-owned enterprises, military-linked entities, and opportunistic commercial actors operate under different incentives and constraints. Policies that ignore those distinctions are more likely to block beneficial activity while missing the specific channels that pose genuine security risks.

The Best Defense Is More AI

One striking limitation in many AI-risk arguments is that they treat AI primarily as an accelerant for offensive capabilities. Anthropic notes that “[a]dvanced AI models will be able to compress R&D cycles in semiconductors, biotech, and advanced materials.” True enough. But AI will also improve defensive cybersecurity, malware analysis, anomaly detection, fraud prevention, vulnerability remediation, supply-chain auditing, biological-threat detection, infrastructure resilience, and autonomous defensive operations.

As models become more commoditized and the technology more diffuse, defensive capabilities will matter more. A dual-use framing that criminalizes—or tightly controls—the development of advanced capabilities risks leaving Americans at a serious disadvantage against emerging threats.

The same agentic systems that can accelerate attacks can also harden networks, monitor behavior, identify adversarial coordination, and defend users at machine speed. That matters because the likely long-run equilibrium is not a world in which dangerous capabilities disappear. It is one in which defensive capabilities proliferate alongside offensive ones.

The cybersecurity evidence also counsels against step-function panic. AI will improve vulnerability discovery, exploit adaptation, and triage, but those gains are more likely to accumulate as continuing pressure than to appear overnight as a clean break in the offense-defense balance. If frontier access is gated through regulatory delay or cartelized compliance costs, the defensive shortfall will fall hardest on smaller firms, open-source developers, and ordinary users who cannot buy bespoke security infrastructure.

This leads to a larger and still underdeveloped question: What does defensive AI look like at the individual level? Much of the current debate assumes security must come from centralized institutional control. Another possibility is emerging: the democratization of personalized defensive intelligence.

In practice, individuals may increasingly rely on persistent AI systems that function as anti-fraud and personal-security tools, monitoring the edges of their digital identities 24 hours a day. These systems would behave less like chatbots and more like always-on defensive infrastructure.

Cyberpunk fiction has a recurring concept of ICE and counter-ICE: intelligent defensive systems continuously contesting hostile intrusion attempts. The analogy feels less fictional by the day. As AI systems become commoditized, the strategic advantage may shift away from merely possessing powerful models and toward possessing trustworthy, aligned, defensive, and highly personalized systems.

In that world, raw model capability may not be the key competitive advantage. Public policy should therefore be careful with sweeping categories like “dual use,” which risk chilling beneficial development and deployment. The United States needs companies building trust systems that preserve individual autonomy while strengthening defensive architectures. Just as critically, those systems must not become entry points for future authoritarian control. That means continuing to develop these technologies in a distinctly American style: decentralized, private, and bottom-up.

A centralized regime that dictates which models can be built and deployed will not get us there. These questions are likely more important than whether an adversary possesses somewhat weaker chips.

Open source, then, is not merely a consumer convenience or a developer preference. It is a strategic lever. Broad domestic open-source availability commoditizes the model layer, prevents deployment from hardening around a few hosted APIs for regulatory reasons, gives emerging firms a distribution path against incumbent network effects, and keeps the global developer base oriented toward U.S.-aligned tools. If domestic open-source development is chilled, developers will not stop building; they will migrate to foreign open-source alternatives.

In short, open source is a strategic playbook, not merely a development norm. Suppressing domestic open-weight competition in the name of export controls would not eliminate open models. It would make foreign open ecosystems more attractive as the default base layer for global developers. 

The Future Is Defense, Not Denial

None of this means export controls—or analogous controls at the capability layer—are useless. There may be narrow domains where targeted restrictions remain sensible, particularly at the bleeding edge. Nor does it mean capability-layer governance will be easy. The better approach is a sliding scale: delay access to the most advanced hardware and model capabilities where delay genuinely matters, but avoid sweeping embargoes that permanently push global developers toward alternative technology stacks.

The capability layer is likely to remain a permanent adversarial battleground. The future is unlikely to resemble nuclear nonproliferation. It will look much more like cybersecurity: an ongoing contest between attackers and defenders in which neither side ever achieves complete control.

The appropriate response is not tighter chip controls alone, nor a government-administered mother-may-I regime for developing and deploying frontier AI systems. It is a policy framework that enables U.S. firms to build resilient systems, democratize defensive intelligence, and harden the places where AI is actually used. In practice, that means preserving reliance on U.S. platforms, governing real access vectors, and making defensive AI broadly available so that security scales alongside the threat.

It also means ensuring that individuals, firms, and institutions possess the tools they need to defend themselves in an increasingly AI-saturated world.

The central challenge of AI policy is not preventing powerful capabilities from existing. That ship has largely sailed. The challenge is ensuring that defensive capabilities diffuse at least as quickly as offensive ones.

If AI’s future looks more like cybersecurity than nonproliferation, then the goal should not be to monopolize intelligence. It should be to make the defense scale faster than the attack.

The post You Can’t Export-Control the Future: The Case for Defensive AI appeared first on Truth on the Market.

The Federal Court of São Paulo has now suspended the R$250,000, or about $50,000, daily fine that Brazil’s Administrative Council for Economic Defense (CADE) imposed on Meta in its WhatsApp AI-chatbots case, sending the parties into conciliation instead. The judicial decision remains under seal, but it marks the latest turn in a remarkable four-month saga. 

CADE opened an administrative inquiry and imposed an interim measure. A federal court initially suspended that measure, then reinstated it after a closer look. CADE then imposed a daily fine after concluding that Meta had not adequately complied. Now, that fine has been paused, too. 

The merits remain unresolved, making this a useful moment to ask what the dispute is really about—and where the harder questions lie. 

The short version is that CADE’s dominance theory rests on relatively firm ground. Its theory of harm, by contrast, depends on assumptions about AI markets that the available evidence is unlikely to support. In that sense, the court’s decision to pause the fine and require conciliation may create a useful off-ramp in an important antitrust dispute. 

How a WhatsApp Policy Change Became an International Incident

The case began when Meta amended its WhatsApp Business Solution Terms to bar third-party general-purpose AI providers from using the WhatsApp Business API as their primary functionality. Two AI-chatbot startups—Madrid-based Factoría Elcano, which operates Luzia, and Montevideo-based Brainlogic AI, which operates Zapia—filed a complaint with CADE’s General Superintendence (SG). They argued that the change amounted to abusive exclusionary conduct under Article 36 of Brazilian Competition Law (BCL) and sought an interim measure to preserve the status quo ante: free access to the API. 

Because Meta rolled out the policy globally, the dispute has not stayed within Brazil’s borders. The Italian Competition Authority (AGCM), which had already opened an investigation into Meta AI’s integration with WhatsApp, expanded its probe to cover the new Business Terms and ordered Meta to suspend them in Italy. 

The European Commission opened a parallel investigation under Article 102 of the Treaty on the Functioning of the European Union (TFEU), covering the European Economic Area (EEA) outside Italy. After Meta replaced the outright restriction with a paid-access policy of $0.06 per message, the commission issued a supplementary statement of objections, preliminarily finding that the new pricing produced the same exclusionary effect as the original ban. Meta announced last month that, “[a]s part of ongoing discussions with the European Commission,” general-purpose AI chatbots operating in the EEA would receive free access to the WhatsApp Business API for one month—effectively freezing the new terms to avoid a European Union interim measure. 

Other regulators are watching, too. Investigations are underway before the Common Market for Eastern and Southern Africa (COMESA) Competition Commission, whose director has publicly ruled out an interim injunction and reportedly said COMESA “will not be influenced by the European Commission’s recent statement of objections” challenging Meta’s chatbot fees. The Turkish Competition Authority is also investigating. Onyeka Aralu has a useful recent Truth on the Market piece on the COMESA proceeding. 

In Brazil, the dispute has unfolded across three parallel CADE proceedings. The first is the administrative inquiry, in which the SG is examining whether Meta’s conduct violates Article 36 of the BCL. In opening that inquiry, the SG also imposed a preventive measure under Article 84 of the BCL, suspending the new terms and preserving the prior arrangement, under which Meta did not charge third-party AI chatbots for API access. 

Meta appealed to CADE’s Tribunal—the second proceeding, known as the Recurso Voluntário—and also went to court. A federal court briefly suspended the measure, then reinstated it after concluding that the record contained enough indicia of anticompetitive conduct to make the measure proportionate and reasonable. With the interim measure restored, CADE’s Tribunal unanimously upheld it, following Commissioner Carlos Jacques’ opinion. 

The third proceeding concerns sanctions. Meta filed a compliance plan in March proposing to readmit AI chatbots to the API, but at a charge of $0.06 per message—the same policy it had introduced in Europe. The SG found that the practice could amount to a “constructive refusal to deal,” meaning access formally exists, but only on terms that make it practically unusable. It concluded that the interim measure required preserving the status quo ante of unlimited, free access and imposed a daily fine of R$250,000, or about $50,000, until Meta restored that arrangement. 

Meta appealed, but the Tribunal unanimously upheld the fine. Jacques’ opinion drew on Erik Hovenkamp’s work on “secondary refusals” and on the European Commission’s supplementary statement of objections, which the opinion framed as “international context as a reinforcement of rationality.” 

Six days later, the Federal Court of São Paulo suspended the fine and ordered Meta and CADE into conciliation. That opened a path toward settlement on terms reminiscent of the agreements recently reached with Google and Apple. 

The Easy Market and the Hard One 

CADE’s case rests on a two-market analysis. In the decision opening the administrative inquiry and imposing the interim measure, the SG declined to define the relevant markets precisely, leaving that question open because the investigation was still in its early stages. Jacques took the same approach in his opinion on Meta’s appeal, refraining from formally delineating the markets. 

Both decisions nonetheless drew broadly on the complainants’ proposed market definitions. The first is a Brazilian market for instant-messaging services, where WhatsApp allegedly holds an uncontestable dominant position and where the challenged conduct occurs. The second is a market for AI services and solutions, whose geographic scope remains open, where Meta AI competes with rivals, including the complainants. 

If CADE ultimately adopts those definitions, WhatsApp’s dominance in Brazilian instant messaging would be difficult to dispute. According to the complainants’ figures, WhatsApp has roughly 148 million users in Brazil, is installed on 99% of smartphones, and is used daily or nearly daily by 97% of users. Those numbers sit comfortably above the 20% rebuttable presumption of dominance established by Article 36, Section 2 of the BCL and would likely support characterizing WhatsApp as a virtual monopoly. 

The downstream market looks very different. There, the relevant market would be AI chatbots and assistants, where Meta competes against the complainants and other firms through Meta AI. The theory of harm combines offensive leveraging—using dominance in messaging to gain an advantage in AI assistants—with self-preferencing, by embedding Meta AI directly within WhatsApp. 

Jacques framed this theory through the lens of ecosystem orchestration, drawing on CADE’s recent Apple App Store case, for which I recommend Mario Zúñiga’s recent analysis. He also invoked the familiar “embrace, extend, extinguish” narrative associated with the 2001 decision in United States v. Microsoft Corp. 

The upstream side of the story is plausible enough. The harder questions arise downstream, in the AI market—a part of the case CADE has not yet meaningfully examined. 

The Bottleneck That May Not Be One

The theory of harm depends heavily on defining the AI-chatbot market narrowly—something like “general-purpose AI assistants distributed through consumer-messaging apps in Brazil.” Broaden the lens to AI assistants generally, which better reflects how users interact with these tools and how AI services are distributed, and the theory becomes much harder to sustain. 

The evidence points away from a simple foreclosure story. As Giuseppe Colangelo documents in his recent paper, “Don’t Catch Me (Too Early) If You Can,” OpenAI’s ChatGPT accounted for roughly 86% of the global AI-chatbot market from April 2024 to March 2025, before falling to about 65% by January 2026 amid increased competitive pressure from Google’s Gemini, which surpassed the 20% market-share threshold. Meta AI, by contrast, remained below 1% of global chatbot traffic. 

Dirk Auer makes a related point in his recent “Brussels’ AI Squeeze” post: Generative AI firms are aggressively expanding across distribution channels. Anthropic has Claude Cowork and Office plugins. Perplexity is available through WhatsApp, Telegram, and X, each with its own dedicated number. OpenAI offers a WhatsApp number that anyone can message. Microsoft distributes AI through Windows, Office, and GitHub. Google does the same through Search, Android, and Workspace. Even the complainants here, Luzia and Zapia, are available through proprietary apps and websites, though WhatsApp remains their main channel for Spanish- and Portuguese-speaking users. 

In a market where multichannel distribution is the norm, treating any single messaging app—even one as dominant as WhatsApp—as a foreclosable bottleneck is a stretch. So is the broader Big Tech-centric theory that this market will predictably tip because of strong network effects, decisive data-feedback loops, and winner-take-all dynamics inherited from earlier platform markets. The actual evolution of AI markets points in a messier—and more competitive—direction. 

Judge Amit Mehta made a similar point in United States v. Google, when he declined to impose a sweeping remedy that would have barred Google from self-preferencing Gemini in Chrome: 

The bar on self-preferencing also goes too far in that it would hamstring Google’s ability to compete. Take, for example, Plaintiffs’ proposal to prohibit Google from self-preferencing Gemini in Chrome. Such a restriction would set Google apart from its competitors. It is commonplace for companies in the GenAI space to leverage their own products to distribute their GenAI technologies. 

None of this means Meta cannot abuse its dominant position in messaging. It does mean that the theory of harm CADE has imported from the AGCM and the European Commission rests on assumptions about AI-market structure that the data do not support. The AI market into which Meta allegedly is leveraging its messaging dominance appears far more competitive than CADE’s current theory allows. 

The Price Sheriff Problem 

The Federal Court of São Paulo’s decision to pause the fine appears to turn on a tension running through Jacques’ opinion. 

The preventive measure required Meta to restore the status quo ante of access to the WhatsApp Business API. The SG and CADE’s Tribunal interpreted that obligation to mean AI chatbots could resume operating on WhatsApp under the same non-onerous conditions that existed when they were classified under the “Customer Service” category. 

Meta complied only in part. It allowed AI chatbots back onto the platform, but reclassified them under the paid “Marketing” category, which Brazilian small and medium-sized enterprises also use for notifications. The SG concluded that this failed to restore the status quo ante and amounted to the functional equivalent of a constructive refusal to deal. Jacques relied on Hovenkamp’s recent work on “secondary refusals” to reach a similar conclusion. CADE then imposed the daily fine. 

Yet Jacques’ opinion in the sanctions proceeding also emphasized that: 

In fact, it is not—and never has been—the role of the competition authority to determine what the fairest or most appropriate price for a specific market would be. CADE cannot be expected to dictate what price should be charged. Its institutional role is to identify guiding parameters to ensure that certain conduct does not have harmful effects on the market, such as raising barriers to entry or excluding competitors from the market. 

That principle is unobjectionable. The difficulty is that, in practice, the order appears to require exactly a price-of-zero outcome. 

The API category at issue was designed, according to Meta, for small and medium-sized enterprises’ messaging services. Under CADE’s interpretation of the interim measure, Meta may charge those firms for access through the Marketing category, but may not charge AI firms using the same infrastructure. As Meta frames the issue, the order prevents it from charging companies such as OpenAI (valued at roughly $850 billion) or Microsoft (valued at roughly $2.9 trillion), even though those firms account for about 98% of the contested API traffic. 

That tension appears to have helped motivate the court’s decision to order conciliation. CADE has sought to distance itself from the image of a “price sheriff,” even as it has become increasingly involved in disputes over platform pricing. The same issue surfaced in CADE’s recent settlement with Apple, where the authority effectively shaped the fee structure governing access to the iOS ecosystem. 

At the same time, CADE’s concern seems less about whether Meta may charge for access and more about Meta’s failure to restore the original arrangement. A similar nuance appeared in the Apple case, where CADE expressly recognized Apple’s right to receive reasonable compensation for granting developers access to its ecosystem. 

That leaves the real question: What is the proper baseline? Is it no tariff at all, or some commercially reasonable tariff—likely below the current Marketing rate, but above zero? The court’s conciliation order may provide both the space and the incentive to answer that question. 

Two Questions, One Case 

It remains unclear where this case ultimately will land, and it may be far from over. The administrative inquiry continues on the merits. The European Commission is moving forward in parallel, albeit with a somewhat more measured procedural posture. Meanwhile, the Federal Court of São Paulo’s conciliation order creates a path toward negotiated terms, rather than an endless cycle of fines and counter-fines. 

In the end, two questions are likely to determine the outcome. 

The first is how CADE defines the relevant AI market when it reaches the merits. A narrow market centered on AI assistants distributed through messaging apps points in one direction; a broader market for AI assistants generally points in another. 

The second is whether the emerging convergence among CADE, the AGCM, and the European Commission produces a coherent enforcement standard for AI distribution channels. Or whether, instead, it illustrates Auer’s warning—borrowing Robert Bork’s phrase—of a “policy at war with itself,” in which a platform is simultaneously constrained from monetizing through advertising, from sharing data across services, and from charging for API access. 

Brazil is an especially important testing ground for that question because WhatsApp occupies a uniquely central place in daily life. The answer may determine not only how platforms distribute AI, but whether competition authorities can develop a theory of AI-market competition that fits the market as it actually exists, rather than the one they expected to find.

The post Brazil, Bots, and the Price of Free appeared first on Truth on the Market.

The company’s reported initial public offering (IPO) has already drawn significant concern from institutional investors and corporate-governance observers. That concern is understandable. SpaceX reportedly seeks to raise as much as $75 billion at a valuation exceeding $2 trillion, potentially making it the largest IPO in history.

SpaceX is also no ordinary issuer. It is one of the most influential companies in the space industry, with billions of dollars in government contracts, considerable political influence, and a central role in both national-security and commercial-space infrastructure.

A May 13 letter from the New York State Comptroller, the New York City Comptroller, and the California Public Employees’ Retirement System (CalPERS) criticizes the reported governance package as “novel and extreme.” The letter highlights several concerns: perpetual super-voting shares, restrictions on removing Elon Musk, mandatory arbitration of shareholder claims, controlled-company status, Texas-law barriers to derivative litigation, and the concentration of the chief executive officer, chief technology officer, and chair roles in Musk, even as he simultaneously leads several other major companies.

Those objections are serious. But the SpaceX controversy is not merely about Musk, founder control, or the outer boundaries of shareholder rights. It also highlights a deeper shift in corporate law. Companies are no longer simply choosing where to incorporate. Increasingly, they are choosing among competing governance philosophies.

My forthcoming article, “New Corporate Geography,” argues that incorporation decisions increasingly reflect more than a preference for a particular chartering state. They also reflect a firm’s operational risks, litigation exposure, regulatory environment, investor base, and governance strategy. SpaceX offers a real-time example of that shift. The reported offering is not simply a test of whether investors will buy a founder-controlled company. It is a test of whether public markets will accept a governance structure built around Texas statutory ordering, federal securities disclosure, and investor choice, rather than the more familiar model of Delaware-style fiduciary review.

It’s Not Just SpaceX. It’s Texas vs. Delaware.

Much of the commentary on SpaceX assumes a Delaware baseline. That is unsurprising. Delaware has long supplied the dominant vocabulary of public-company governance. Its corporate law is flexible and contract-friendly, but it is also infused with equity—the old judicial tradition that courts may intervene when formally legal conduct is unfair or abusive.

That matters because corporate governance is not just about what a company’s charter says. In Delaware, private ordering operates against a backdrop of fiduciary duties, entire-fairness review, enhanced scrutiny, and the Court of Chancery’s power to police inequitable uses of corporate machinery. Put less lawyerly: Delaware lets companies write many of their own rules, but it keeps a judge nearby.

The familiar Delaware principle, stated in Schnell v. Chris-Craft, is that inequitable action does not become permissible simply because it is legally possible. Delaware is not anti-contract. But its contractarianism operates within a fiduciary culture that gives courts meaningful room to ask whether directors and controllers have used corporate rules unfairly.

Texas corporate law is moving in a different direction. Texas generally does not recognize a broad common-law duty of good faith and fair dealing in ordinary commercial contracts. Its recent corporate-law reforms emphasize statutory authorization, procedural predictability, and ex ante governance design—that is, rules chosen in advance, rather than reviewed later through open-ended equitable standards.

Senate Bill 29, enacted in 2025, created new governance tools for Texas corporations, including presumptions favoring directors and officers of public and electing corporations, heightened pleading requirements, internal-forum provisions, jury-trial waiver authority, narrower inspection rights, and derivative-action ownership thresholds of up to 3% of outstanding shares. Derivative suits are lawsuits shareholders bring on behalf of the corporation, often to police alleged fiduciary misconduct. Raising the ownership threshold can sharply limit who may bring them.

That distinction matters because the SpaceX debate is often framed as a shareholder-protection problem. That framing is not wrong, but it is incomplete. The reported SpaceX structure is better understood as a test of whether public markets will accept a disclosed governance arrangement that substitutes contractual and statutory ordering for many of the litigation rights and equitable protections traditionally associated with public-company investing.

Federal securities law reinforces the point. The public-offering regime is primarily disclosure-based, not merit-based. Regulation S-K requires companies to disclose material risk factors that make an investment speculative or risky. It generally does not prohibit investors from buying into a risky, founder-dominated, asymmetrical governance structure. The premise is not that investors will make wise choices. The premise is that they should receive material information before making them.

The Securities and Exchange Commission’s (SEC) recent position on mandatory arbitration points in the same direction. In September 2025, the SEC issued a policy statement clarifying that a mandatory-arbitration provision for federal securities claims would not, by itself, prevent acceleration of a registration statement’s effectiveness. In plain English: such a provision would not automatically block a company from going public. The SEC’s position does not decide whether every such provision is enforceable. But it reflects a disclosure-centered approach to public offerings: disclose the provision, disclose the consequences, and let investors decide whether to buy.

Texas law makes that disclosure-and-consent model more consequential. Under SB 29, eligible Texas corporations may adopt derivative-action ownership thresholds of up to 3% of outstanding shares. For a company valued in the trillions, that threshold would be functionally prohibitive for nearly all shareholders. The pension funds’ letter makes the point directly, warning that the threshold could require a shareholder to hold billions of dollars in stock before bringing a derivative claim.

That objection should not be dismissed. Derivative litigation remains one of the principal mechanisms through which shareholders police fiduciary misconduct. A threshold that only the largest institutions can satisfy meaningfully changes the shareholder-enforcement landscape.

But that change also illustrates the broader Texas model. Texas is not merely adjusting shareholder-litigation rights in isolation. It is building a governance framework that favors statutory clarity, advance ordering, and limits on litigation over open-ended equitable review.

What If Investors Say Yes?

This is where the SpaceX debate becomes part of the broader “DExit” conversation—the debate over whether companies are leaving, or should leave, Delaware for other states. Recent reincorporation fights are often framed as contests over court quality, doctrinal predictability, or judicial expertise. That framing is too narrow.

A firm with unusual investor demand, a founder closely identified with the company, national-security significance, and high expected growth may value a legal regime that reduces shareholder-litigation optionality. Whether investors should accept that tradeoff is a separate question. The point is that Texas gives companies a statutory vocabulary for making it explicit.

For many firms, Delaware’s model remains highly valuable. Delaware offers an experienced judiciary, a deep body of precedent, and well-developed doctrines governing conflicted transactions, board process, controller conduct, and fiduciary oversight. Those features reduce uncertainty in many contexts. They also lend legitimacy to public-company governance because investors understand that formal compliance may still be subject to equitable review.

For other firms—particularly founder-controlled companies with unusually strong investor demand—those same backstops may look less like investor protection and more like litigation uncertainty. Texas offers a different proposition: if the statute permits a governance structure, the governing documents adopt it, and investors are told what they are buying, the bargain should generally stand.

That does not mean investor choice is always meaningful. The pension funds’ strongest argument concerns index inclusion. If SpaceX becomes large enough to enter major equity indexes, passive funds and public pensions may hold its stock as a practical matter, even if they object to the governance terms. In a market increasingly dominated by index investing, consent is not always an individualized act of agreement. Sometimes it is a consequence of market structure.

That concern deserves serious attention. But it does not transform every restrictive governance provision into a legally illegitimate one. Instead, it points to the real policy question: Should public-company governance be constrained through substantive fairness rules, or through disclosure, pricing, and investor choice?

If the answer is substantive fairness, regulators, exchanges, index providers, and institutional investors should say so directly. They should identify which governance structures are too restrictive for public markets, even when investors are willing to buy them. That would move the system closer to merit regulation—the idea that regulators should judge whether an investment is acceptable—and further from the traditional disclosure-based model.

If the answer is disclosure and consent, then SpaceX may become the most prominent example yet of the new geography of corporate governance. In that geography, incorporation decisions are not merely technical choices about a chartering state. They are choices among competing legal infrastructures: equity and fiduciary review in one jurisdiction, statutory ordering and litigation filters in another, with federal securities disclosure layered across both.

What the Market Will Tolerate

The SpaceX IPO should not be read merely as a Musk story. It is a test of investor autonomy, shareholder enforcement, and the changing geography of corporate law. Public markets may reject the bargain. Investors may demand a discount. Index providers may resist inclusion. Regulators may eventually conclude that some governance terms go too far.

But if investors knowingly buy the stock, the offering will reveal something important about charter competition: incorporation is increasingly becoming a mechanism for sorting firms and investors across competing governance infrastructures.

In that emerging geography, governance rights are not always the only product being sold. Sometimes investors are buying exposure, upside, identity, and access.

The SpaceX IPO is ultimately a referendum not on Elon Musk, but on how much governance choice public markets are willing to tolerate.

The post SpaceX and the New Geography of Corporate Governance appeared first on Truth on the Market.

The consumer welfare standard asks whether business conduct harms the competitive process, typically by raising prices, reducing output, degrading quality, or slowing innovation. Neo-Brandeisianism, by contrast, is more skeptical of concentrated economic power as such. It worries not only about prices and output, but also about whether large firms wield too much political, social, or economic influence.

For all its political momentum, neo-Brandeisianism has had a much harder time reshaping antitrust law than reshaping antitrust rhetoric. Courts have largely continued to apply economically grounded standards focused on competitive effects, efficiency, innovation, and consumer welfare. As Lazar Radic and Nicolas Petit recently observed, neo-Brandeisianism has had far greater influence on the political conversation surrounding antitrust than on antitrust doctrine itself. 

Strategic rivalry between the United States and China has now added a new dimension to that debate. Innovation, technological leadership, and economic performance have become matters of national strategy, particularly in industries such as semiconductors, artificial intelligence, cloud computing, and advanced manufacturing. Ironically, that shift has strengthened the case for an antitrust framework centered on efficiency, innovation, and dynamic competition—the very considerations at the heart of the consumer welfare standard. 

When Antitrust Goes Geopolitical

This shift matters for antitrust policy. Critics have long argued that the consumer welfare standard is too narrow because it focuses on economic outcomes rather than broader social goals. Yet geopolitical competition has underscored the importance of precisely the factors that consumer-welfare analysis emphasizes: innovation, productivity growth, technological leadership, and efficient use of resources. These are no longer merely economic objectives. They are increasingly sources of national strength. 

That does not mean antitrust should become industrial policy in a wig and robe. Competition authorities should not promote national champions or sacrifice competition principles in pursuit of geopolitical objectives. Competition law and industrial policy serve different functions. Governments may legitimately pursue industrial-policy goals through subsidies, procurement policies, research funding, export controls, and investment incentives. Competition law, by contrast, should remain focused on preserving competition, market contestability—the ability of new rivals to challenge incumbents—and innovation. 

Even so, geopolitical realities reinforce the case for an antitrust framework that prioritizes efficiency and innovation, rather than treating scale itself with suspicion. The point is especially clear in artificial intelligence and semiconductor markets, which feature enormous fixed costs, significant economies of scale, intense global competition, and fast-moving technological ecosystems. In these sectors, large firms often play a critical role in financing research, building infrastructure, and commercializing new technologies. Indeed, some of the United States’ most important advantages in artificial intelligence depend on firms that would plainly qualify as “Big Tech.” 

Why Size Isn’t the Right Question

That reality creates an obvious tension for neo-Brandeisian approaches that treat bigness itself as suspect. If technological leadership depends partly on globally competitive firms operating at scale, policymakers may grow less willing to treat size as a problem in itself. That does not mean large firms get a free pass. Exclusionary conduct, anticompetitive acquisitions, and practices that suppress innovation remain legitimate antitrust concerns. 

The right question is whether particular conduct harms competition and innovation—not whether a firm has simply become large. That distinction matters especially in AI markets, where the most efficient business structures and technical architectures remain uncertain. Nobody yet knows exactly which models, platforms, chips, cloud systems, or organizational forms will prove most effective. Competition policy should therefore focus on preserving innovation rivalry, rather than trying to engineer market outcomes—or market structures—through highly prescriptive interventions. 

Recent survey evidence also suggests public attitudes toward artificial intelligence are more nuanced than often assumed. Although many Americans worry about artificial intelligence’s social consequences, RAND survey data indicate that a significant share of the public views U.S. leadership in artificial intelligence as strategically important and worries about China overtaking the United States in AI development. In other words, skepticism of large technology firms may increasingly coexist with the recognition that technological leadership is now part of geopolitical competition. 

The Consumer Welfare Standard Strikes Back 

To be sure, the political landscape remains fluid. Anti-bigness arguments continue to influence policy debates and are no longer confined to the neo-Brandeisian movement. Vice President J.D. Vance has expressed sympathy for concerns about excessive concentration, and some competition officials continue to advocate heightened scrutiny of large technology platforms based less on demonstrable competitive harms than on their size and economic importance. 

There are also reports that Sens. Amy Klobuchar (D-Minn.) and Chuck Grassley (R-Iowa) may soon reintroduce the American Innovation and Choice Online Act. Modeled in part on the European Union’s Digital Markets Act, the bill would impose a range of obligations on designated technology companies, limiting their ability to expand into adjacent markets, including, presumably, artificial intelligence. Among other things, it would restrict some forms of vertical integration—that is, combining complementary stages of production or distribution within the same company—and require covered platforms to share investments and capabilities through interoperability mandates, which require systems to work with rivals’ products or services. None of these provisions would require a showing of harm to consumers. The future direction of U.S. antitrust policy therefore remains contested. 

Even so, the broader trend is becoming harder to ignore. Geopolitical competition is pushing policymakers to focus on innovation, productivity, technological development, and dynamic efficiency. Across the Atlantic, Mario Draghi’s report on European competitiveness identified a lack of scale as a central weakness, while the European Commission’s draft merger guidelines increasingly treat scale and innovation as complementary competitive virtues. These developments sit more comfortably within the consumer welfare tradition than within approaches that regard economic scale as inherently suspect. 

The irony is difficult to miss. Geopolitical competition was supposed to make the consumer welfare standard look obsolete. Instead, it has highlighted the importance of the very factors the framework has emphasized all along: innovation, efficiency, and economic dynamism. In a world defined by technological rivalry, consumer welfare looks less like a narrow economic objective and more like a strategic necessity.

The post How China Accidentally Made Consumer Welfare Cool Again appeared first on Truth on the Market.

Brazil’s Google News inquiry belongs firmly in the latter category. On April 3, Brazil’s Administrative Council for Economic Defense (CADE) announced that its Tribunal had unanimously decided to send a seven-year-old administrative inquiry concerning Google’s use of journalistic content—whether for indexing information, displaying “snippets,” or, more recently, generating “AI Overviews”—back to the agency’s General Superintendence (SG/CADE). This time, the Tribunal instructed SG/CADE to open a formal administrative proceeding and conduct a deeper investigation. 

The shift followed the vote of acting President Diogo Thomson de Andrade (the “Thomson vote”) and a concurring opinion (voto-vista) by Commissioner Camila Cabral Pires-Alves (the “Pires-Alves vote”). The rapporteur, former Commissioner Gustavo Augusto Freitas de Lima, had originally voted to dismiss the matter, but later amended his vote to join the concurring opinion. 

A unanimous vote may sound decisive. In this case, it is almost the opposite. Rather than close a stale inquiry launched in 2019, whose original factual premises have changed beyond recognition, CADE chose to keep the matter alive while substituting a new theory of harm and, in some respects, a new set of facts. There are several reasons to view that move with concern and skepticism. 

A New Case in an Old File

A brief procedural recap is in order.

The inquiry traces back to another well-known Brazilian case: Google Scraping. CADE did not open the matter in 2019 through SG/CADE acting ex officio—that is, on its own initiative. Instead, the Tribunal itself launched the inquiry while adjudicating the Google Scraping proceeding. 

The trigger was Globo/G1’s 2019 response to a request for information, or RFI. Issued by Brazil’s largest media conglomerate, the RFI asked whether Google had used Globo’s content on its news platform. Globo responded: 

As for the period between 2011 and 2012, we know that G1 content appeared in search results, but we are unable to determine at this point [in 2019] how Google’s news platform worked at the time, nor what content was displayed. As for the subsequent period [2013 onwards], the answer is yes—the platform provided links, headlines, and photos of G1 content. However, we cannot say with certainty exactly when this content began to be displayed.

Globo’s central allegation was straightforward: Google News and Google Search displayed headlines and snippets derived from third-party news content without compensation. The claim was framed as both exclusionary and exploitative conduct. In antitrust terms, an exclusionary claim alleges that a firm harmed rivals or shut them out of the market; an exploitative claim alleges that a firm used its market power to impose unfair terms. In response, SG/CADE opened the inquiry through Order No. 4/2019. 

After four years of investigation, CADE’s Department of Economic Studies (DEE/CADE) issued Technical Note 24/2023, concluding that the evidence failed to support any of the proposed theories of competitive harm. The findings were unequivocal: no incentive to foreclose rivals, no essential-facility concerns, and—crucially—a positive net flow of traffic from Google to publishers, with “insufficient evidence that snippets led consumers to stop accessing the links” (Thomson vote, ¶36). SG/CADE reached the same conclusion in Technical Note 70/2024, recommending dismissal for lack of evidence of an antitrust violation—or, put differently, lack of competitive harm (Thomson vote, ¶63). 

Ordinarily, that would have ended the matter. 

Instead, on March 28, 2025, Commissioner Camila Cabral Pires-Alves issued Decision No. 9/2025, proposing that the Tribunal revisit the case through a procedural mechanism known as avocação. Under Brazilian competition-law procedure, a commissioner who disagrees with SG/CADE’s recommendation may ask the Tribunal to substitute its own judgment for that of the investigative authority. In practice, avocação allows the Tribunal to review and potentially overturn SG/CADE decisions, including dismissals in conduct cases and unconditional approvals in merger reviews. 

The Tribunal accepted Pires-Alves’ proposal and randomly assigned the case to Freitas de Lima as rapporteur (Thomson vote ¶¶66, 69). At CADE’s 249th Ordinary Judgment Session on June 11, 2025, Freitas de Lima voted to affirm SG/CADE’s recommendation of dismissal. 

His reasoning tracked the agency’s earlier findings. Snippets, he concluded, “generate additional traffic for the portals, effectively functioning as a form of free publicity” (Thomson vote ¶80). A leveraging theory failed because Google “does not operate directly” in the downstream market for news portals (Thomson vote ¶81). The practice was justified under the rule of reason because indexing is “a central element of the business model of search engines, facilitating users’ rapid access to information and generating gains in efficiency and utility” (Thomson vote ¶82). Finally, economic dependence by publishers, standing alone, does not constitute a competition-law violation (Thomson vote ¶86). 

By that point, DEE/CADE, SG/CADE, and the assigned rapporteur had all independently reached the same conclusion. 

The case nevertheless took another turn. 

At that same session, Acting President Diogo Thomson de Andrade filed a request for further consideration (Thomson vote ¶94) and launched what he described as a “supplementary investigation” (Section 2 of his vote). He issued additional decisions, sought further information from Google, sent RFIs to seven publisher associations (Thomson vote ¶99, Table 4), and explicitly invited “civil society in general to submit technical and factual contributions” (Thomson vote ¶101). 

The effort generated 10 additional submissions from Brazilian media outlets—including Zero Hora, A Gazeta, and Folha de S.Paulo—as well as advocacy organizations such as the Open Markets Institute’s Center for Journalism & Liberty, Article 19 Brazil, Reporters Without Borders, Sleeping Giants Brasil, and Foxglove (Thomson vote ¶102, Table 5). Many of these organizations are sophisticated advocacy groups with established positions on platform regulation. 

The timing matters. 

Google launched AI Overviews in Brazil only in August 2024 (Thomson vote ¶154), after SG/CADE had already prepared Technical Note 70/2024 and recommended dismissal. The original 2019 inquiry therefore could not have generated evidence concerning AI Overviews. In substantive terms, Thomson’s supplementary investigation looks less like a continuation of the original case than a new investigation into a new product feature, attached to an old file whose original premises—scraping, snippets, and headlines—had already failed on the evidence. 

Thomson is candid about the institutional rationale behind this approach. He writes that the supplementary investigation “was not guided solely by the aim of strengthening the evidence in the strict sense” (Thomson vote ¶111). Rather, it reflected “a broader institutional understanding of the need to increase the openness of competition policy to the involvement of civil society and of actors who, although not directly competing in the market under scrutiny, represent diffuse interests.” Later, he contrasts this approach with analysis driven primarily by technical specialists, arguing that doing otherwise would create “an artificial barrier to the expression of diverse perspectives” and risk “a kind of closure of the marketplace of ideas” (¶¶117-118). 

That rationale raises important concerns.

Treating SG/CADE’s dismissal recommendation as the starting point for a year-long, advocacy-driven supplementary investigation risks transforming the agency from a competition-law enforcer into a stakeholder-management forum. There is, of course, a place for broad debate about competition policy. Competition-law proceedings serve a narrower purpose. They assess specific harms under defined legal standards, using evidence-based tools intended to reduce error costs—the risk that an agency either condemns lawful conduct or misses genuinely harmful conduct. 

CADE should not hesitate to pursue cases when the evidence demonstrates harm to competition. But here, the evidence assembled by the agency’s own economic experts points in the opposite direction. Against that backdrop, the unusual duration and procedural trajectory of this case raise legitimate questions about the fairness and neutrality of the process. 

The Return of the Phantom Menace

The most consequential—and ambitious—move in the case is substantive. Thomson acknowledges that CADE’s technical bodies have thoroughly examined the traditional theories of harm. In his view, the case deserves a second life through a different lens: exploitative abuse.

As Thomson explains:   

Thus, the theories of harm already explored by the DEE/Cade – increased costs for rivals, leveraging of a dominant position, scraping of journalistic content that allows access to information without redirecting to publishers, and tying – as well as by the SG/Cade – predatory innovation and blocking of essential inputs, retention of traffic with a view to increasing advertising revenue, and self-preferencing in the event of traffic diversion or retention for one’s own benefit – and subsequently analysed in the opinion of the Reporting Commissioner, have, in my view, already received sufficient analytical treatment in the course of the proceedings and in previous decisions.

There remains, however, a need for a more detailed examination of the exploitative aspect of the conduct, even though this has been addressed in part in the aforementioned statements. For this reason, I shall focus here on the aspect of abuse of a dominant position that I consider relevant, specifically with regard to alleged exploitative abuse (or abuse of economic dependence), whether or not associated with elements of exclusionary abuse. (emphasis in original) (Thomson vote, ¶¶ 502-503).

Thomson reaches this conclusion through a particular reading of Article 36 of Brazil’s Competition Law (Law No. 12,529/2011). Having acknowledged that the conventional abuse theories failed on the evidence, he argues that Brazilian competition law nevertheless reaches the conduct through what both he and Pires-Alves call the “typological openness” of Article 36—that is, the statute’s non-exhaustive list of anticompetitive practices.

Pires-Alves expressly embraces this approach. “I concur with Commissioner Diogo [Thomson]: it seems more appropriate to frame the theory of harm, at the forefront, as exploitative” (Pires-Alves vote ¶10). She notes that “the trajectory of exploitative abuses in [Brazil] was marked more by institutional caution than by conceptual rejection” (Pires-Alves vote ¶11, citing Barbosa & Kastrup, 2021) and argues that Article 36’s typological openness permits scrutiny of “unilaterally imposed unfair conditions on dependent partners, including when the extraction of value does not manifest in the form of direct monetary price” (Pires-Alves vote ¶11).

Pires-Alves also acknowledges that the exclusionary case has largely failed. In her view, that failure does not doom the exploitative theory: 

The current absence of a stronger showing of exclusion does not neutralize the exploitative theory of harm. It is preferable to acknowledge that the issue admits a primarily exploitative reading, without ruling out possible exclusionary effects, and that the current state of the investigation does not yet point to a definitive conclusion in this regard. (Pires-Alves vote ¶14). 

That argument faces a significant obstacle: exploitative-abuse cases are competition law’s phantom menace. They are theoretically available in many jurisdictions, but rarely enforced in practice. 

European Union law permits exploitative-abuse claims under Article 102(a) of the Treaty on the Functioning of the European Union (TFEU). Yet the European Commission’s own enforcement guidance prioritizes exclusionary conduct, and exploitative-abuse cases remain uncommon. U.S. antitrust law has gone further. Since the U.S. Supreme Court’s 2004 decision in Verizon Communications Inc. v. Law Offices of Curtis V. Trinko, LLP, monopoly pricing itself has generally been treated as lawful absent anticompetitive conduct. As the Court explained: 

The mere possession of monopoly power, and the concomitant charging of monopoly prices, is not only not unlawful; it is an important element of the free-market system. The opportunity to charge monopoly prices—at least for a short period—is what attracts “business acumen” in the first place; it induces risk taking that produces innovation and economic growth. To safeguard the incentive to innovate, the possession of monopoly power will not be found unlawful unless it is accompanied by an element of anticompetitive conduct.

In Justice Antonin Scalia’s view—and, we would argue, the correct one—the prospect of earning monopoly profits is a feature, not a bug, of a competitive economy. The possibility of charging supracompetitive prices rewards risk-taking, investment, and innovation. Brazilian law reflects a similar principle. Article 36, §1 of the Brazilian Competition Law provides that “[a]chieving dominance in a market by natural process and by being the most efficient economic agent in relation to competitors does not characterize” an antitrust violation. Antitrust enforcement should therefore focus on supracompetitive prices achieved through anticompetitive conduct, not those resulting from superior efficiency, innovation, or business acumen. 

Even setting that debate aside, exploitative abuse is notoriously difficult to define and prove. Drawing a principled line between a “reasonable” price and an “excessive” one—or determining the “fair” allocation of economic benefits—is no easy task. As Massimo Motta and Alexandre de Streel observe, “the proof of an excessive price, or in other words the search for the competitive price, may be like a quest for the Holy Grail.” 

For that reason, even scholars who support the use of antitrust law against excessive pricing generally regard it as a remedy of last resort. Motta and De Streel argue: 

It is important that in those guidelines [referencing the European Commission’s Guidelines on Article 82, now Article 102 TFEU], the Commission commits itself to limit the use of its broad legal power and make explicit that excessive pricing actions should be an option of last resort for antitrust authorities, and that they should be used only when other routes fail. (emphasis added and footnotes omitted) (Motta & De Streel, p. 42). 

The Thomson-Pires-Alves approach points Brazilian law in the opposite direction. 

As Giuseppe Colangelo has explained, the “free-riding” narrative underlying many claims by newspapers and other media organizations against digital platforms “serves chiefly as a rhetorical tool, strategically used to legitimize a mandated transfer of revenue that policymakers appear ready to pursue even in the absence of clear evidence that platform-based business models have caused economic harm to press publishers.” 

Once that theory is accepted—and once Article 36’s “typological openness,” combined with notions of “structural dependence,” becomes a substitute for traditional dominance-and-foreclosure analysis—the doctrine is unlikely to stop with Google. Any platform with allegedly dependent counterparties, a category broad enough to encompass countless ordinary commercial relationships, could become a target for exploitative-abuse claims untethered from measurable competitive effects. 

As Craig Conrath has explained:

[I]t is necessary to distinguish between contract power and monopoly power, between post-contractual opportunism and abuse of dominant position. … Therefore, whenever there is a claim of monopolistic practice by a party in a contractual relationship with the person accused of a monopolistic practice, it is important not to ask the question, “Does this party now have any other choices?” This is a question of contract law. Instead, one should ask the relevant antimonopoly law question: When the contract was made, did this party have other choices? (Practical Handbook Of Antimonopoly Law Enforcement For An Economy In Transition, § 5, at 5 (1995)).

That distinction matters here. The mere fact that a business has become economically dependent on a successful platform does not transform a contractual dispute into an antitrust violation.

It is also worth remembering what platforms such as Google News and other content aggregators actually do. They collect, organize, and direct users to relevant information, reducing search costs—the time and effort consumers would otherwise spend finding news and other content on their own. That benefit should not vanish from the analysis simply because some publishers dislike the commercial terms of the relationship.

Dominance Without a Market

Even if one accepts the direction this case has taken, a threshold question remains: In what market, exactly, is Google supposed to be dominant for purposes of an AI Overviews investigation?

There are at least three possible answers, and the case looks weak under all of them.

One possibility is that AI Overviews are simply a new feature within Google Search. On that view, the relevant market remains general search, where Google’s share in Brazil has exceeded 90% for years, albeit with some recent erosion. If so, the AI-related concerns are merely an extension of the original snippets-and-scraping investigation.

The problem is that this framing ignores the competitive threat that generative artificial intelligence poses to Google itself. In the remedies phase of United States v. Google LLC, after finding that Google possessed monopoly power in general search, Judge Amit Mehta declined to impose structural remedies, such as divestiture of Chrome or Android. A key reason was the emergence of generative AI. Mehta concluded that AI had “changed the course of this case” and posed a meaningful threat to Google’s position. Against that backdrop, it is difficult to reconcile a theory that treats Google’s efforts to compete in AI as evidence of deepening monopoly power.

A second possibility is that AI Overviews compete in a distinct market for generative-AI assistants and conversational search tools. Users who seek information through ChatGPT, Claude, Perplexity, Grok, or Gemini are often making functionally similar choices.

In that market, Google is not dominant. It is a challenger. Gemini has gained traction, but it still trails ChatGPT and faces strong competition from Anthropic, Perplexity, xAI, and others. More importantly, the market itself remains highly dynamic. Google, Microsoft, OpenAI, Anthropic, Perplexity, and other firms are rapidly experimenting with how AI-driven search works, how sources are cited, how referral traffic is generated, and how these products are monetized.

Treating any particular implementation as the basis for an antitrust proceeding risks freezing a temporary design choice into law. In markets evolving this quickly, false positives are especially costly. Regulators may deter features that consumers value, publishers benefit from, and competitors do not challenge in practice.

Recent empirical evidence reinforces this point. In “The Emerging Market for Intelligence” (2025), Aydin Demirer, Andrey Fradkin, Steven Tadelis, and Yiming Peng document extraordinary turnover in generative-AI markets. Drawing on usage data from OpenRouter and Microsoft Azure, they find that the number of available large-language models grew from 253 to 651 in 2025 alone, while the number of model creators nearly doubled and the number of inference providers more than tripled. Intelligence-adjusted prices at the technological frontier fell by roughly 1,000x since 2023. Leadership shifted repeatedly among Anthropic, Google, and xAI models throughout the year, and “the top 10 models today accounted for just 20% of market share four months ago and did not even exist ten months ago.” Their conclusion is straightforward: “No single model dominates across use cases.” 

That is not what an entrenched market looks like.

A third possibility is a broader market definition encompassing search engines, AI assistants, social media, email, and other channels through which publishers reach readers. From the perspective of news organizations, the relevant question is not whether Google exists in isolation, but whether publishers have meaningful alternatives for distribution and audience acquisition.

The Thomson vote attempts to avoid this inquiry through a dependence-based framework:

In my view, the core of the case does not lie in precisely and exhaustively determining whether the relevant market should be defined, with mathematical precision, as general search, thematic news search, digital content distribution, or any other intermediate formulation. The core issue is to determine whether, regardless of the formal market definition adopted, publishers operate in a position of structural dependence on Google for purposes of discovery, distribution, monetization, and the management of their own economic risk.

This emphasis on dependence, rather than definitional formalism, is consistent with Padilla’s (2024) approach. If the test for exploitative abuse is to focus on the harm caused by dominance, it would make no sense to require, as an absolute prerequisite, an exhaustive mapping of all digital attention channels. The legally relevant point is another: whether the dominant firm is capable of imposing terms that the trading partner would not accept in a minimally competitive environment. (Thomson vote, ¶¶591-592)

That approach raises serious concerns.

For one thing, Brazilian law still requires dominance to establish abuse of dominance. Unlike jurisdictions such as Germany or Japan, Brazil has no standalone doctrine of abuse of economic dependence. More fundamentally, the reasoning risks becoming circular. The challenged contractual arrangement becomes evidence both of the firm’s alleged market power and of the harm that power supposedly produces. Competition law ordinarily requires separate proof of market power, improper conduct, and competitive harm. 

The Case for Closing the Case

Commissioner Gustavo Augusto Freitas de Lima’s approach would likely have produced the better outcome.

As noted above, Freitas de Lima—the case’s original rapporteur—initially voted to uphold SG/CADE’s recommendation of dismissal, relying heavily on the DEE’s technical analysis, which found no evidence of competitive harm arising from the challenged conduct. Following Thomson’s supplementary investigation, Freitas de Lima modified his position in one important respect. He agreed that AI Overviews might warrant scrutiny, but as a distinct investigation separate from—albeit related to—the original Google News matter. He explained:

§ 133. Accordingly, following the analysis conducted based on the submissions provided and taking into account the study prepared by the DEE, I find that there is no evidence of a material violation of the economic order with respect to the conduct originally investigated, which consisted of the collection, indexing, and display of journalistic content through snippets and related mechanisms in the search engine.

§ 134. I emphasize, however, that the use of artificial intelligence tools integrated into the search engine may imply a significant functional difference in the conduct, with potential competitive impacts distinct from those observed in the traditional search model. Considering the evidence gathered, and without delving into the merits of the matter, I believe there is sufficient evidence to justify the initiation of a new investigation to specifically address the innovations introduced by Google’s Artificial Intelligence Overview (AI Overview).

§ 135. In light of the foregoing, I vote to dismiss the present Administrative Inquiry regarding the original conduct, pursuant to Article 144, §3, I, of RICADE [CADE’s Internal Regulation] and Article 67, §2, I, of Law No. 12,1529/11. As for the conduct related to the use of artificial intelligence in the search ecosystem (AI Overview), I concur with the proposal contained in the dissenting opinion of Commissioner Diogo Thomson, under the terms outlined therein, understanding that the case should be returned to the SG for the opening of a new specific investigation.

If CADE genuinely believes that Google’s deployment of AI Overviews in Brazil raises competition concerns, then the procedurally sound response is the one Freitas de Lima outlined—not keeping a 2019 inquiry on procedural life support.

CADE should close the original inquiry. The factual premises that gave rise to it have been overtaken by technological change, and the evidentiary record assembled over seven years supports dismissal. Closing the case would not prevent future enforcement. It would simply require the agency to articulate a new theory of harm, define the relevant market, and build a new evidentiary record tailored to the conduct actually under scrutiny.

CADE should also resist the temptation to embrace a broad exploitative-abuse doctrine. Even in jurisdictions that recognize such claims, exploitative abuse remains a tool of last resort. It is difficult to define, difficult to prove, and even more difficult to remedy in a predictable and administrable way.

AI Overviews may or may not raise genuine competition concerns. That question deserves serious analysis. But if CADE wants to investigate the future of search, it should start a new case—not keep rewriting an old one.

The post Brazil’s Google News Case and the Art of Not Letting Go appeared first on Truth on the Market.

The European Union’s Digital Markets Act (DMA) is the clearest recent example. The DMA did not emerge because economists suddenly discovered universal truths about digital markets. It reflects a specific regulatory judgment: that the costs of underenforcement in digital markets outweigh the risks of overenforcement, and that values like fairness and contestability can justify departing from effects-based analysis. Whether those tradeoffs make equal sense for countries like India is a very different question.

India’s Draft Digital Competition Law, still in gestation, appears poised to follow the same path. Both regimes treat tying and bundling by dominant digital platforms as presumptively illegal. Both impose categorical ex ante prohibitions on practices economists have long recognized as frequently efficiency-enhancing.

In doing so, both misunderstand not only the economics of digital markets, but also the basic reality of how digital products are built. The laws’ treatment of tying and bundling make that failure especially vivid. They also make it especially consequential for a country like India, whose economy increasingly depends on technological innovation and is now being reshaped by artificial intelligence (AI). 

Brussels Is Not Bangalore

India’s Draft Digital Competition Law emerged, in large part, from the gravitational pull of the Brussels Effect. The two government reports that anchor the draft law—the Parliamentary Standing Committee’s 2022 report and the Ministry of Corporate Affairs’ 2024 follow-up—borrow directly from the DMA’s architecture. 

On tying and bundling, the first report effectively copy-pasted Article 5(8) of the DMA wholesale. The draft article provides: 

The Committee thus are, of the view that a SIDI should not force business users or end users to subscribe to, or register with, any further service as a condition for being able to use, access, sign up for registering with any of that platform’s core platform service. 

That language imposes an absolute bar on gatekeepers requiring users of one core platform service to subscribe to, register with, or use another core platform service as a condition of access. No exceptions. No weighing of effects. No room for the kind of case-by-case economic analysis that has characterized competition jurisprudence in both the EU and India for the better part of three decades. 

Before accepting this as a reasonable model, India should ask a question the drafters appear not to have considered: Was the DMA designed for India, or for Europe? 

The answer is plainly Europe—and perhaps not even all of Europe. The DMA’s per se approach to bundling reflects a deliberate regulatory tradeoff that makes sense, if it makes sense at all, only within the EU’s specific political economy. Of the top 100 technology companies in the EU by market capitalization, 74 are American. The DMA is, at least in part, an instrument of digital sovereignty: a mechanism to constrain the dominance of U.S. platforms and clear space for European tech firms. Whether that goal justifies the economic costs is a normative question Europeans must answer for themselves. It is emphatically not India’s problem to solve. 

At a more technical level, India’s draft law, like the DMA, wrongly assumes that all gatekeepers operate on similar business models and that digital markets are uniformly “winner-take-all.” The result is a one-size-fits-all set of 18 obligations that risks significant error costs—that is, the economic harm caused when regulators mistakenly condemn beneficial conduct or miss harmful conduct. 

India’s digital economy has its own competitive dynamics: fierce regional differentiation, strong multi-homing, and platform markets that look nothing like the winner-take-all archetypes animating Brussels’ regulatory imagination. In India’s taxi market, for example, Uber does not dominate everywhere. Goa Miles dominates in Goa, Namma Yatri in Bangalore, and Ola competes fiercely nationwide. 

Applied to India, DMA logic could sweep in many homegrown technology companies with regional strength and varied business models, categorizing them as gatekeepers despite their very different competitive realities. 

Case-by-Case Beats Copy-and-Paste

Here is what proponents of India’s Draft Digital Competition Law rarely acknowledge: India already has a sophisticated, effects-based framework for adjudicating tying and bundling cases. And it works. 

Under Section 4(2)(d) of the Competition Act, 2002, tying by a dominant firm is prohibited when three conditions are met: the tying and tied products are distinct; access to one product is conditioned on acceptance of the other; and that conditionality forecloses competition in the tied market. Critically, any finding of anticompetitive conduct also requires proof of an “Appreciable Adverse Effect on Competition.” That standard obliges competition authorities to weigh both pro- and anticompetitive effects ex post. 

Despite periodic suggestions that Section 4 functions as a quasi-per se prohibition, the Supreme Court settled the issue definitively in Competition Commission of India v. Schott Glass India Pvt. Ltd. (May 2025). The Court held unequivocally that a per se reading of Section 4 is wrong. Demonstrable market effects must be shown, and the analysis must account for both competitive harm and legitimate business justifications. 

The practical record shows the framework works as intended. When the Competition Commission of India (CCI) found Google liable for tying its Play Billing System to app distribution, it did so only after conducting a careful effects-based analysis showing concrete harms to innovation, developer autonomy, and ultimately consumer welfare. When similar claims were brought against Apple’s in-app payment rules, the same framework produced a similarly reasoned analysis. 

The CCI did not need a categorical prohibition to identify abusive conduct. It needed the right analytical tools, and it already had them. 

The real question, then, is not whether India needs a competition-law framework capable of handling digital tying. It already has one. The question is whether replacing careful case-by-case analysis with categorical prohibitions would improve that framework. 

It would not. 

Sometimes Integration Is the Product

Tying arrangements generate efficiencies across multiple dimensions. They can reduce production, distribution, and licensing costs—often substantially—when demand for bundled components is positively correlated. They can improve product quality by solving compatibility problems that arise when consumers combine components from different providers. They reduce consumer search costs, which become increasingly burdensome as products grow more technically complex. And they can solve the problem of double marginalization: when two monopolists sell complementary products independently, each adds its own monopoly markup; vertical integration through tying eliminates the duplicated markup and can reduce final prices for consumers. 

These are not abstract artifacts of the Chicago School. They are widely documented empirical regularities. 

In digital markets, these efficiencies do not weaken. They intensify. 

Software products typically involve high fixed costs and near-zero marginal costs, which magnifies the gains from cost reduction. Compatibility problems in complex technological products are also far more severe than in traditional manufacturing markets. A smartphone operating system, cloud platform, or AI ecosystem is not a kitchen mixer with interchangeable attachments.

The value of reducing consumer search costs is especially high in markets where many users lack the technical expertise to assemble and optimize products independently. That matters in India, where digital adoption has expanded rapidly, but digital literacy remains uneven. Much of India’s 1.4 billion-person population still lacks the technical knowledge to navigate highly modular digital ecosystems efficiently. 

Yet these efficiencies are largely absent from the DMA’s framework and from India’s draft law. That omission creates substantial error costs by treating potentially beneficial integration as presumptively suspect. 

There is also a category of efficiency unique to technological markets that antitrust doctrine has long struggled to describe clearly—and that categorical per se rules are structurally incapable of recognizing at all. It requires its own conceptual vocabulary. 

When 1 + 1 = 3

Systems theory uses the concept of emergence to describe a specific phenomenon: when components combine into a system, the system can acquire properties that are not present in—and cannot be predicted from—the individual parts. 

The classic examples are physical. Water is wet; hydrogen and oxygen, individually, are not. A murmuration of starlings produces fluid, coordinated movement that no individual bird directs. Ant colonies build and maintain complex infrastructure without any ant understanding the overall design. The emergent property—wetness, coordinated flight, colony architecture—belongs to the system as a whole. Break the system apart, and the property disappears. 

The same phenomenon appears in technology.

Technological integration, properly understood, occurs when two or more components are combined at a sufficiently deep technical level that the integrated system acquires a genuinely new capability—something neither component can perform independently, and something that cannot be recreated simply by running both side by side. 

That distinction matters. Bundling two apps onto a home screen is not technological integration. Nor is preinstalling a search engine alongside a browser in the relatively superficial sense addressed in the Google Android case. Technological integration, in the emergent sense, occurs when components are architected together so that their interaction produces a new system-level capability. 

The result is not merely Component A plus Component B. It is Product C—with distinct functionality, distinct user value, and often distinct consumer demand. 

Apple’s Face ID illustrates the point clearly. Three components work together: the TrueDepth infrared camera array, which captures detailed three-dimensional facial geometry; a neural-network recognition system, which processes and matches that data against a stored template in real time; and the Secure Enclave’s cryptographic architecture, which stores authentication data and executes verification without exposing sensitive information to the operating system.

Each component can be described independently. None, standing alone, performs secure biometric authentication.

The authentication capability emerges from their integration. Remove the Secure Enclave, and the system loses its trusted execution environment. Remove the neural network, and the camera loses its matching capability. The emergent system—secure, real-time, privacy-preserving biometric authentication—depends on the integration itself. Put simply, it does not survive unbundling. 

A second example comes from medical diagnostics. A medical-imaging tool integrated with a convolutional neural network trained on longitudinal patient data can forecast disease risk decades in advance. The imaging tool alone captures anatomical structure. The neural network alone classifies patterns. But the integrated system, trained on the relationship between imaging features and long-term clinical outcomes, produces predictions neither component could generate independently. The diagnostic capability emerges from the integration. 

Increasingly, AI-driven product design follows the same logic. When a large language model is integrated architecturally into a productivity suite—not as a chatbot bolted onto the side of an existing application, but as a capability woven into document editing, email composition, data analysis, and workflow automation—the result is a fundamentally new kind of tool. 

The product is no longer simply “word processor plus AI.” It becomes a qualitatively different productivity environment whose capabilities depend on deep technical co-design between the components. 

This is precisely why the standard “separate product test” in tying analysis—which asks whether the tied and tying products had independent pre-existing demand—is the wrong framework for evaluating emergent integrations. 

The test is inherently backward-looking. It asks whether the products existed separately before integration occurred. Emergence, by definition, creates something new. The integrated system could not have generated prior independent demand because it did not previously exist as an integrated system. Applying a backward-looking demand test to forward-looking innovation will systematically misclassify valuable integrations as presumptively suspect. 

Before the DMA, European courts in the Microsoft litigation gestured toward a more sensible framework when they held that technological integration must produce “superior technological efficiency” specifically dependent on the tie. That standard imperfectly captures the logic of emergence. It asks not whether products were historically separate, but whether integration produced something genuinely new and better. 

It is an imperfect test, but at least it asks the right question.

The DMA abandoned that approach entirely. India’s draft framework, with its backwards-looking “integral to the core service” exemption introduced in the 2024 report, risks making the same mistake.

Innovation, but Only If It Already Exists

The Ministry of Corporate Affairs’ 2024 report acknowledged, at least in principle, that a blanket prohibition would be too blunt. It proposed a narrow exemption for ties that are “integral to the provision of the Core Digital Service,” leaving the Competition Commission of India to specify, through future regulations, what qualifies.

That is better than nothing. It is not nearly good enough.

The amended draft provision provides:

15. Tying and Bundling: A SIDI shall not require or incentivize business users or end users of the identified Core Digital Service to use on or more of the Systemically Significant Digital Enterprise’s other products or services, or those of: (a) Related parties; or (b) Third parties with whom the Systemically Significant Digital Enterprise has arrangements for the manufacture and sale of provision of services Alongside the use of the identified Core Digital Service, unless the use of such products or services is integral to the provision of the Core Digital Service.

Explanation: For the purpose of this section, the Commission may specify, by regulations, the nature of products or services that may be considered ‘integral’ to the provision of a Core Digital Service.

The exemption is backward-looking by design. If a product feature was not already part of the core service, it may not count as “integral” to that service—no matter how transformative the integration may be. This freezes product definitions at the moment of designation and penalizes precisely the kind of forward-looking, emergent innovation that drives value in digital markets.

The rule effectively tells firms that technological improvement through integration is presumptively unlawful unless it mirrors the historical baseline. When innovation arises through emergence—where combining products A and B creates a new product C—a legal framework that tethers the value of C to the original provision of A risks pixelating emergent innovation, reducing a novel creation to its constituent inputs. 

The exemption also introduces a new and more opaque form of legal uncertainty. By delegating the definition of “integral” to future regulation, it replaces judicial standards with discretionary administrative determinations. Those determinations may be made without transparent market studies, meaningful appellate mechanisms, or stable limiting principles—and may be reclassified at any point.

That is not an improvement over the effects-based framework. In important respects, it is worse: less predictable, more susceptible to capture, and further removed from principled economic analysis. 

Crucially, the exemption makes no room for economic justification. It does not ask whether a tie reduces search costs, counters double marginalization, resolves compatibility problems, or produces superior technological efficiency. Those are precisely the factors Indian courts have long weighed in Section 4 cases. 

Under the draft law, they disappear entirely. 

The exemption is not a return to economic reasoning. It is a fig leaf on top of a prohibition. 

The AI Problem the DMA Cannot See

The inadequacy of both the DMA and India’s draft framework becomes clearest in frontier cases.

Consider, hypothetically, Microsoft’s integration of OpenAI’s model into Microsoft 365 through Copilot. Under the DMA, the threshold question is whether Microsoft’s AI offering qualifies as a “core platform service,” and whether Article 5(8) therefore applies at all. If AI-generated content tools do not qualify as core services, the analysis reverts to traditional Article 102 doctrine.

At that point, the familiar problems return. Does an iterative, self-learning AI assistant have independent demand apart from the productivity suite into which it is integrated? The answer is far from obvious. So, too, is whether the integration produces the kind of “superior technological efficiency” recognized in the pre-DMA Microsoft jurisprudence—the standard European courts developed before the DMA largely swept it aside.

Those are difficult questions. A per se rule simply refuses to ask them. 

Under India’s draft framework, the result is arguably worse. Microsoft Office would likely qualify as the core digital service. Copilot would be supplied through a commercial arrangement with a third party. On a literal reading of Section 15 of the draft law, the integration would therefore be unlawful unless Copilot were “integral to the provision” of Office.

But Office plainly functions without Copilot.

The rule would thus classify a genuine AI-driven productivity innovation as an anticompetitive tie in complete disregard of emergence. Not because the integration forecloses competition. Not because it harms consumers. But because the capability did not exist in the product before. 

That is not effects-based competition analysis. It is regulatory formalism masquerading as technological neutrality. 

Europe’s Regulatory Politics Are Not India’s Future

India should resist the DMA’s pull—not because tying is always benign, but because the economics of tying are always contextual. That is especially true in the age of AI. 

The right framework starts from the premise that tying warrants scrutiny, not automatic condemnation. A structured effects-based analysis should weigh demonstrated harms—leveraging, foreclosure, and price obfuscation—against the full range of pro-competitive efficiencies, including technological integration in the emergent sense described above. 

The European Commission’s pre-DMA approach—which asked whether an integration produced “superior” functionality genuinely dependent on the tie—offers a workable and economically coherent template. India’s Competition Act, updated to codify explicit exemptions for emergent technological integration and to clarify the role of multi-homing and switching costs in foreclosure analysis, would be far better equipped to handle AI-era tying cases than any per se rule imported from Brussels. 

Multi-homing matters here, especially in India. Users routinely rely on multiple platforms simultaneously across food delivery, payments, ride-hailing, messaging, and e-commerce. That reality substantially weakens the lock-in logic underlying many foreclosure theories. Strengthening data portability through frameworks like the Data Empowerment and Protection Architecture (DEPA) would weaken switching costs further, limiting the anti-competitive potential of even aggressive tying strategies. 

The goal of competition law, as the Indian Supreme Court reaffirmed in Competition Commission of India v. Schott Glass India Pvt. Ltd., is to protect competition through consumer welfare—not to protect competitors from competition. 

Protecting competition in India’s digital economy means preserving incentives to innovate through integration, not sacrificing them in pursuit of regulatory symmetry with the EU. 

India does not need a digital competition regime that mistakes integration for abuse, innovation for foreclosure, and product improvement for coercion. 

It certainly does not need to reinvent the License Raj for the AI era.

The post The Bundle of All Fears: India’s Risky War on Integration appeared first on Truth on the Market.

That is the central tension in the Commission’s April 16 preliminary findings under Article 8(2) of the Digital Markets Act (DMA), which specify how Alphabet must comply with Article 6(11)’s data-sharing obligations. The consultation closed May 1, and a final implementing act is due by July 27.

The proposed measures are detailed, and they reflect a serious effort to reconcile the DMA’s data-access mandate with the General Data Protection Regulation’s (GDPR) anonymization requirement. In particular, the Commission proposes a two-layer regime: a technical anonymization pipeline—attribute suppression, allowlisting, length thresholds, metadata generalization, and “mini-sessionization”—backed by contractual restrictions and recurring audits.

The problem is that the regime works only if both layers hold. And each layer depends heavily on trust in the other.

This post examines two questions the Commission has not adequately answered.

First, are the technical measures sufficient—on their own terms—to render the Search Dataset anonymous under European Union law? Put differently, do they ensure that re-identification “appears in reality to be insignificant,” under the Court of Justice of the European Union’s (CJEU) standard in Breyer, later reaffirmed in Single Resolution Board (SRB)?

Second, are the contractual and audit mechanisms robust enough to handle the realistic range of recipients? That includes recipients who are technically competent, commercially sophisticated, and not formally designated as hostile, but who may still behave adversarially in practice.

My short answer to both questions is no. More importantly, the two weaknesses reinforce each other.

The Commission has shifted meaningful risk-bearing work from the technical layer to the contractual layer, and from the contractual layer to private enforcement after the fact. If the audit process cannot be trusted, the anonymization process cannot be trusted. If the anonymization process cannot be trusted, the data was never lawful to share.

I will explain why both lines of defense are weaker than the consultation document suggests. In doing so, I draw on my earlier analyses of the coming  GDPR/DMA Article 6(11) conflict, my coverage of the 2024 and 2025 DMA compliance workshops, and my comparison of the EU DMA regime with Judge Amit Mehta’s user-side data-sharing remedy in U.S. v. Google. I also draw on the International Center for Law & Economics (ICLE) comments submitted to the Commission during the consultation by Geoffrey Manne, Dirk Auer, and Mario Zúñiga regarding Alphabet’s Article 6(11) obligations.

The Recital Can’t Save the Rule

Start with the legal standard. Article 6(11) of the DMA requires gatekeepers to provide access to “ranking, query, click and view data” on fair, reasonable, and nondiscriminatory (FRAND) terms. It then adds the crucial condition: “[a]ny such query, click and view data that constitutes personal data shall be anonymised.” Recital 61 further states that anonymization should occur “without substantially degrading the quality or usefulness of the data.”

The recital is doing more work in the Commission’s draft than it can bear.

As Peter Craddock has argued, and as Mark Leiser similarly argues in his consultation submission, the operative provision is unconditional: personal data “shall be anonymised.” The recital’s utility caveat operates within the anonymization requirement, not against it. If a given technique sufficiently anonymizes the data, the gatekeeper should prefer the version that preserves more utility. But if no available technique can adequately anonymize the data at a given utility level, the legal answer is to suppress the data—not to weaken the anonymization standard. The Commission’s Article 8(2) specification power does not extend to creating a softer, DMA-specific definition of “anonymisation.”

So what does anonymization require?

Under CJEU case law—most notably Breyer, and now SRB—the test is “relative.” The same dataset can be personal data for one entity and anonymous data for another, depending on the “means reasonably likely to be used” by the recipient, or by anyone else whose capabilities must realistically be considered.

Importantly, Breyer limits the analysis to lawful means of identification. The question is whether the risk of identification “appears in reality to be insignificant.” Opinion 05/2014 of the Article 29 Working Party supplies the operational framework: anonymization must prevent singling out, linkability, and inference.

The Commission’s two-layer approach—technical restrictions plus contractual controls—is, charitably read, an attempt to satisfy SRB by tightly constraining the recipient environment, so that re-identification tools available to recipients no longer count as “reasonably likely” means. The draft Joint Guidelines on the interplay between the DMA and the GDPR endorse this combined approach. (See paragraphs 180-181.)

The hard question—and the one the Preliminary Measures largely glide past—is whether that combination actually delivers what the DMA requires.

The Technical Pipeline’s Four Big Problems

To be fair, the five-step pipeline described in Section 3.1 of the Preliminary Measures is more sophisticated than Google’s earlier frequency-thresholding implementation. Third parties complained that Google’s original approach was so restrictive that it yielded little useful data. At the 2025 compliance workshop, DuckDuckGo and Seznam argued that 99% of distinct queries were excluded. 

As Alba Ribera Martínez explains, the Commission’s newer regime is more nuanced. It combines an entity-based allowlist—more than 50 signed-in users issuing queries containing a given entity over 13 months—with length-based suppression, metadata generalization, and “mini-sessionization.” That is more robust than Google’s original “30 globally signed-in users per exact query” filter, at least for queries that are lexically rare but semantically common.

Still, the pipeline has at least four structural weaknesses the Commission has not adequately addressed.

The detector layer is brittle at scale

The technical pipeline relies heavily on “personal data detectors” to identify names, addresses, and phone numbers before queries are split into entities. (See paragraph 22(a)(1) of the Preliminary Measures.) As Craddock explains, names are not a tractable detection problem at internet scale.

Capitalization conventions vary by user habit. A search for “james brown” may evade a capitalized-name detector. Naming conventions vary by culture: “Charles de Gaulle,” “LeBron James,” and “Sk?odowska-Curie” all behave differently. Meanwhile, countless ordinary words are also common surnames: “Cook,” “Smith,” “Rose,” “Bill.”

Across billions of search queries, even a low false-negative rate produces millions of records in which personal data slips through. Tightening the detectors creates the opposite problem: more false positives and more over-suppression of legitimate queries. Either way, the system degrades.

Notably, the Preliminary Measures contain no acceptable-error-rate metric, no public benchmark against which detector performance can be audited, and no contingency plan for inevitable misclassifications.

There is also a deeper problem, and this is where Leiser’s critique becomes especially powerful. Search queries routinely contain personal data about people other than the searcher: relatives, colleagues, public figures, complainants, alleged wrongdoers, former partners, and so on.

The Commission’s pipeline effectively assumes that the only relevant data subject is the user entering the search query. Article 6(11) does not support that assumption. The operative provision states that “any such query, click and view data that constitutes personal data shall be anonymised.” Under Article 4(1) of the GDPR, “personal data” is defined by reference to the person the data relates to—not the person who generated it.

To be sure, the “such data” referenced in Article 6(11) originates from end-user activity. But the phrase “constitutes personal data” does not limit the relevant data subject to the end user. Recital 61 opens by discussing “personal data of end users,” and the Commission appears to rely heavily on that wording. But the recital’s own anonymization test immediately drops the end-user qualifier, referring instead to whether the information relates to “an identified or identifiable natural person.”

In any event, a recital cannot narrow the operative provision, as Craddock notes. Reading Article 6(11) to authorize disclosure of identifiable third-party information—with no notice, no controller chain, and no remedy for the affected third party—would also sit uneasily with Article 52(1) of the Charter of Fundamental Rights, given the Article 7 privacy rights of those named individuals.

The mechanism is easy to see.

Imagine a user searches: “Jean Dupont infidelity divorce Brussels.” “Jean Dupont” is a placeholder for a moderately well-known professional whose name has appeared often enough in Google Search queries to land on the Commission’s allowlist. The detector recognizes the name. The entity clears the 50-user threshold. The remaining terms—“infidelity,” “divorce,” and “Brussels”—are common enough to survive filtering. The query length falls below the 95th-percentile threshold. The record flows into the export dataset.

Recipients then receive the full query text, along with country, language, device information, and an S2-cell-level location indicator. The searching user’s metadata may satisfy the Commission’s k=50 anonymity threshold. Jean Dupont’s data does not. The pipeline does not meaningfully treat him as a data subject at all.

The result is disclosure of identifiable information about Jean Dupont’s family life—and potentially his sexual conduct, depending on the inferences recipients draw—to entities with which he has no relationship, no contract, and no GDPR controller chain. Under the Commission’s design, he receives neither notice nor remedy.

Nor does the detector layer solve this. Even a perfect name detector would suppress third-party data only if the name itself were blocked, and only if suppression extended to the rest of the query string. The Commission’s pipeline does neither. If the name appears on the allowlist, the surrounding query text remains intact.

In fact, the allowlist may invite leakage. The more publicly salient a person is, the more likely his name appears on the allowlist, and the easier it becomes for any search mentioning him—by any user, for any reason—to flow into the daily export.

The metadata thresholds do not solve composition attacks

The Commission also requires that at least 50 signed-in users share the same combination of inferred language, location, and device type. (See paragraph 31.) This is essentially a k-anonymity guarantee. The problem is that k-anonymity has a familiar weakness: a group can be anonymous without being private.

If all 50 users in a cohort share the same sensitive characteristic, anonymity does little good. For certain query categories—a rare medical condition, for example, or a politically sensitive term within a small linguistic minority—the Commission’s k=50 threshold may still permit effective disclosure of the sensitive attribute across the entire cohort.

The Commission’s implicit answer seems to be that sensitive records can be filtered through entity and length thresholds. But that misses the point. Composition attacks are orthogonal to query rarity or query length.

The regime does not address active influence attacks 

The regime also appears vulnerable to active influence attacks by recipients themselves, a point Craddock and Leiser both emphasize.

Eligible recipients have lawful access to Google Search. They—or their employees, contractors, or paid panel users—can deliberately run searches designed to push particular entities across the 50-user threshold. Once a recipient has reason to believe a seeded query will appear in the Search Dataset, locating the resulting record becomes much easier.

None of this requires hacking, illicit databases, or breach of contract. It involves lawful conduct. That matters because Breyer asks whether identification methods are “reasonably likely to be used.” Lawful conduct that recipients can undertake unilaterally almost certainly qualifies.

The Commission’s contractual prohibition on “re-identification” or “sessionisation” does not solve this problem. (See paragraph 38(b).) Those restrictions matter only if the Commission later discovers the behavior and successfully enforces the rules after the fact.

The auxiliary-data problem is massive 

Finally, the linkage surface with auxiliary data is enormous. As the Chamber of Progress and others note in their consultation submissions, the dataset includes URLs, approximate timestamps, language, location, device information, access-point data, and increasingly detailed click and dwell-time signals.

Eligible recipients—by definition, rival search engines—already possess their own logs of user activity against many of the same URLs, during roughly the same time periods, and often involving overlapping users. Publishers and advertisers possess additional complementary data.

In other words, the DMA-protected dataset is intentionally designed to complement recipients’ own information. That is the entire competitive rationale behind Article 6(11).

Anonymization that might withstand attack by a recipient with no auxiliary data becomes far weaker when the attacker possesses structurally aligned datasets by design. Recital 26 of the GDPR explicitly directs regulators to consider identification means “reasonably likely to be used … by another person.” The Commission has not grappled seriously enough with that reality.

In short, the technical pipeline is more sophisticated than Google’s initial implementation. But it remains brittle along several dimensions the Preliminary Measures barely address. The Commission’s answer is to shift the remaining burden to contractual restrictions. That is where the second concern begins.

Trust Us, We Audited It

The Commission’s contractual measures, set out in Section 3.2 of the Preliminary Measures, are extensive on paper.

Recipients are prohibited from re-identification, data augmentation, record-level linkage with auxiliary datasets, sessionization beyond “mini-sessions,” and onward sharing. The regime also imposes purpose limits: recipients may use the Search Dataset only to optimize online search engine (OSE) services. (Paragraph 40.) There is a 13-month retention cap. (Paragraph 41.) The measures further require encryption at rest and in transit, least-privilege access controls, phishing-resistant multifactor authentication, restrictions on local workstation copying, and logging obligations with one-year retention periods. (Paragraphs 45-49.)

The compliance architecture culminates in a two-tier independent audit cycle under ISAE 3000 or an equivalent standard. Level 1 audits assess the design and suitability of controls; Level 2 audits assess operating effectiveness. (Paragraphs 52-70.)

At first glance, this looks like a serious compliance package. The problem is that the audit system is weaker than it appears. Three points matter in particular.

The audits mostly check boxes

The Level 1 and Level 2 reports focus on whether the recipient has documented controls and whether those controls appear to function as designed. (Paragraphs 56-57.)

Alphabet’s role is remarkably limited. It need only confirm that the report exists, is signed by a qualified practitioner, addresses the required assurance objectives, and satisfies the required format. (Paragraph 111.) The Preliminary Measures explicitly state that Alphabet “shall not reassess the substance or scope of the assurance conclusions.”

Nor does the Commission appear to conduct substantive review of the audits, beyond receiving notification that they occurred. The regime largely treats the existence of an audit as evidence of compliance, without meaningfully interrogating the quality of the audit or the rigor of its conclusions.

‘Independent’ auditors are not necessarily independent

The independence requirement is narrower than it sounds. The Preliminary Measures define auditor independence by reference to ISAE 3000 ethics standards. Those standards address the auditor’s independence from the audited entity itself. They do not address independence from outside commercial or strategic interests that may align with the audited entity.

That distinction matters. A formally qualified ISAE 3000 practitioner who is commercially eager for business, operating in a weak supervisory environment, technically outmatched by the auditee, or simply disinclined to push too hard against a paying client can still produce a clean report.

And that is before the harder cases: auditors who fully understand the weaknesses in the system, but nonetheless issue a reasonable-assurance opinion at the outer edge of what the standard tolerates.

The Commission’s framework assumes that “independent auditor” is a meaningful substitute for adversarial verification. Often, it is not.

The data move faster than the audit cycle

The timing mismatch is the final problem. After the initial Level 1 assessment, the audit cycle becomes annual. (Paragraphs 56-57.) The Search Dataset, by contrast, flows daily.

That means the system relies heavily on retrospective compliance review while disclosure occurs continuously and at scale. Even within an audit period, problematic behavior could persist for months before detection—assuming it is detected at all.

Put differently, the regime is built around after-the-fact accountability, not ex ante technical impossibility.

Those weaknesses become more serious once we stress-test the framework against the kind of recipient the Commission’s first assurance objective is supposed to screen out.

The Regime Is Built for Compliant Recipients

The Commission’s first assurance objective, set out in paragraph 67, comes closest to a substantive eligibility screen.

It excludes recipients that are directly or indirectly:

• sanctioned under EU restrictive measures;
• covered by sanctions that otherwise prohibit provision of the Search Dataset; 
• subject to action under the EU foreign direct investment screening regime  (Regulation 2019/452);
• designated as a “high-risk supplier” under Union law; or 
• identified by Union or member-state authorities as cybersecurity, public-security, or public-order threats. 

If a recipient is not on a sanctions list, has not been screened out under foreign-investment rules, and has not been formally designated a high-risk supplier, it clears the first assurance objective.

That sounds reassuring—until one asks what kinds of actors the framework actually excludes.

Consider a third-party “online search engine” incorporated in an EU member state, but beneficially owned through layered structures by actors aligned with an adversarial state. A Chinese or Russian intelligence-linked commercial proxy is the obvious example, although the same logic applies to organized criminal groups or sanctions-evasion networks.

Given the Commission’s expansive definition of “online search engine”—which now reaches AI chatbots with search functionality—setting up such a service is increasingly easy. The recipient applies, signs the license agreement, hires a formally qualified but commercially marginal ISAE 3000 auditor, obtains a Level 1 report, and begins receiving daily Search Dataset exports through the API.

What protects users—and anyone else named in search queries—in this scenario?

Very little.

The auditor verifies that, at the time of the report, the recipient is not formally sanctioned and has not been subject to an EU investment-screening decision. But the auditor need not pierce the corporate veil to identify ultimate beneficial ownership.

Nor is the auditor conducting an intelligence investigation. The recipient’s “credible and documented plans” for search-engine development, required under paragraph 68, are evaluated largely on the basis of the recipient’s own documents. Serious bad actors can produce credible paperwork.

The “change of control” procedure is similarly weak. (Paragraphs 117-120.) Recipients must notify Alphabet within 10 days of a public announcement, or 30 days before the change takes effect, whichever comes earlier. But the ownership transitions that matter most are precisely the ones designed not to become public.

Alphabet’s role remains largely ministerial. Under paragraph 111, it may verify only the formal completeness of the audit materials. It may not reassess the auditor’s conclusions. The Commission receives notice under paragraph 112, but there is no required forward-looking substantive review.

The contractual restrictions are equally fragile under real-world conditions. Yes, the recipient contract prohibits re-identification, augmentation, onward sharing, and related conduct. But contractual enforcement becomes largely aspirational once data has been exfiltrated, especially across borders.

Article 7 of China’s National Intelligence Law, for example, empowers Chinese entities to be compelled to cooperate with state intelligence work. Comparable authorities exist in other adversarial jurisdictions. If a state-linked recipient—or a state-linked subsidiary designed to receive the data—obtains the Search Dataset, the data will move. Any eventual contractual or regulatory penalties will fall on the EU shell entity long after the data has crossed the relevant border.

The expedited-termination mechanism does little to solve this. Paragraph 138 permits suspension only where there is “an urgent risk of serious and irreparable damage to the anonymisation of end users’ personal data.” That is a high evidentiary threshold. By the time the necessary evidence exists, the damage is likely done.

And termination does not rewind disclosure. The contract may require deletion of previously received data. (Paragraph 134.) A bad-faith recipient will simply ignore that obligation, especially where the EU lacks meaningful enforcement leverage.

The contrast with the U.S. District Court for the District of Columbia’s “Qualified Competitor” framework in U.S. v. Google is instructive. As I noted last September, the decree defines a Qualified Competitor as:

[A] Competitor who meets the Plaintiffs’ approved data security standards as recommended by the Technical Committee and agrees to regular data security and privacy audits by the Technical Committee, who makes a sufficient showing to the Plaintiffs, in consultation with the Technical Committee, of a plan to invest and compete in or with the GSE and/or Search Text Ads markets, and who does not pose a risk to the national security of the United States.

Several differences matter here.

First, the audits are conducted by the Technical Committee, not by auditors hired and paid by the recipient. That largely neutralizes the incentive problem on which the Commission’s framework depends.

Second, eligibility is substantively approved by the plaintiffs in consultation with the Technical Committee. It is not granted automatically based on the recipient’s own documents and a formally compliant audit report.

Third, recipient suitability is treated as a continuing supervisory question, overseen by a standing body throughout the life of the decree—not as a one-time compliance screen followed by annual reporting.

The national-security overlay is only the most obvious distinction.

To be clear, most actual recipients will not be hostile actors. Most will be legitimate search engines and AI assistants trying to build competing products. But that is not the relevant design question.

When the dataset may contain highly sensitive personal information, the regime cannot merely be safe for the median recipient. It must be resilient against the worst-case recipient, because the worst-case recipient is the one capable of imposing catastrophic privacy harms on EU users.

The Commission’s own first assurance objective implicitly recognizes this. Otherwise, there would be no sanctions-screening mechanism at all.

The problem is that the Commission has designed the screen around the wrong variable: formal designation status. The harder and more consequential questions—beneficial ownership, jurisdictional control, downstream state-compulsion risk, and operational alignment with adversarial actors—are left almost entirely to the recipient’s own auditor.

Anonymous by Trust Alone

The Commission has plainly moved beyond the crude frequency-thresholding regime that characterized Google’s early implementation, and the Preliminary Measures reflect a genuine effort to take privacy seriously.

But the regime still has two important weaknesses.

First, the technical layer leaves residual identification risks that, by the Commission’s own account, can only be reduced to an “insignificant” level through contractual restrictions. Second, the contractual layer depends heavily on private enforcement and an audit cycle that is formally rigorous but substantively thin. It also lacks anything resembling the continuing recipient-qualification framework the district court imposed in U.S. v. Google.

Together, the two layers produce an uncomfortable result: the legal status of the Search Dataset as “anonymous” depends on a chain of trust. That chain includes the recipient’s good faith, the auditor’s diligence, and Alphabet’s ability to police conduct it cannot directly observe. That is a fragile foundation for compliance with a statutory anonymization requirement.

Two changes would substantially improve the final implementing act.

First, the Commission should strengthen the technical layer, even at some cost to utility. The Preliminary Measures already contemplate multiple “samples” (A, B, and C), while the broader literature—particularly Leiser’s submission—offers a more graduated toolkit: aggregate-only access, controlled API access, clean-room environments, and regulator-supervised escrow arrangements. Differential-privacy budgets for aggregate signals, along with stronger detector-quality assurance backed by published false-negative metrics, are all feasible.

The Commission is right that Article 6(11) data should not be rendered useless. But Recital 61’s instruction that utility not be “substantially degraded” operates within the anonymization requirement, not above it. The recital cannot override the operative command that personal data “shall be anonymised,” and the final implementing act should say so clearly.

Second, the recipient-eligibility framework needs to grapple seriously with the risks the current proposal largely sidesteps—especially the incentive problem created when auditees choose and pay their own auditors, and the inability of formal-designation screening to detect undisclosed hostile ownership or jurisdictional control.

The Commission’s current framework does not do that. The precise mechanics of a more substantive merits-based review can be refined through the specification process. But the direction is already clear from the U.S. v. Google comparison.

At bottom, the Commission is attempting something difficult: forcing broad data sharing while insisting the shared data is no longer personal data. That is a legally and technically precarious balancing act.

And if the system works only so long as everyone behaves, it is probably not an anonymization regime. It is a trust regime.

The post The European Commission’s Search-Data Trust Fall appeared first on Truth on the Market.

Armen Alchian’s 1950 article, “Uncertainty, Evolution, and Economic Theory,” offers a powerful reply. Rather than defend the standard assumption, Alchian acknowledged the critics and suggested economists should stop describing firms as profit maximizers altogether. Under genuine uncertainty, firms cannot calculate profit-maximizing strategies in the way traditional models assume. Continuing to rely on that assumption, he argued, unnecessarily exposes economic theory to criticism.

Yet Alchian did not reject the predictive power of economic reasoning. Instead, he explained why profit-maximizing outcomes can emerge even when firms do not consciously maximize profits. Markets generate those outcomes through a process of variation, adaptation, and survival.

As Alchian famously observed, economists should focus less on firms’ intentions and more on the consequences of competition:

The economic counterparts of natural selection are imitation and innovation, the counterparts of heredity are the transmission of successful business practices.

In competitive markets, firms need not solve optimization problems for profit-maximizing behavior to emerge. It is enough that firms experiment—and that firms whose practices fail to generate sufficient profits eventually disappear.

Uncertainty and the Limits of Profit Maximization

Alchian emphasized a basic feature of economic life: uncertainty. Most economic models assume firms face risk—that is, situations in which the probabilities of various outcomes can be estimated and incorporated into decision-making. Under such conditions, firms could, in principle, calculate expected profits and choose strategies that maximize those expectations.

Real-world economic decisions rarely resemble this stylized environment. Firms introducing new products, entering unfamiliar markets, or responding to technological change often cannot reliably estimate the probabilities of future events. They do not know whether consumer preferences will shift, whether competitors will adopt new technologies, or whether regulatory environments will change.

Under genuine uncertainty, the information required to calculate profit-maximizing decisions simply does not exist. Firms cannot solve the optimization problems that economic models often attribute to them.

For Alchian, this observation carried an important methodological implication. If firms cannot calculate profit-maximizing strategies under uncertainty, then continuing to describe them as profit maximizers invites unnecessary criticism. Economists should instead focus on a different mechanism—one that does not require deliberate optimization. That mechanism is survival.

Survival of the Profitable

Markets operate as selection environments. In any competitive market, firms adopt a wide range of strategies. Some innovate; others imitate competitors. Some invest heavily in new technologies; others rely on established practices. Some pricing strategies succeed; others fail. Variation is inevitable.

Not all firms survive. Firms that consistently fail to generate sufficient profits eventually lose access to capital, fall behind competitors, or exit the market entirely. Firms that generate stronger profits persist and expand. Over time, this process produces a population of surviving firms whose behavior appears consistent with profit maximization.

That appearance, however, is misleading. Profit-maximizing behavior does not emerge because firms deliberately calculate optimal strategies. It emerges because firms that fail to approximate such behavior disappear.

This perspective also highlights something traditional models tend to obscure: widespread business failure. Many firms do not survive. They misjudge consumer demand, adopt inefficient production methods, or fail to adapt to changing competitive conditions. In a world characterized by uncertainty, such mistakes are inevitable.

Standard profit-maximization models offer little explanation for why so many firms fail. Alchian’s framework does. Firms operate with imperfect information and limited foresight. They make decisions based on guesses about the future, and some of those guesses inevitably prove wrong. Survival therefore depends not on perfect planning, but on the ability to adapt when predictions fail.

The firms we observe at any given moment are therefore not a random sample of all firms that attempted to compete. They are the survivors of a competitive filtering process. As Alchian put it, “Those who realize positive profits are the survivors; those who suffer losses disappear.” The appearance of profit maximization emerges from this process of elimination.

Business failures, however unfortunate, play an important role in this evolutionary process. When firms fail, their strategies provide information to competitors, investors, and entrepreneurs about what does not work under particular circumstances. Competitors observe those outcomes and adjust their own practices accordingly. In this way, markets learn collectively through the successes and failures of individual firms. The process is not merely eliminative. It is informational.

Adaptation Beats Perfect Foresight

If survival explains why profit-oriented behavior dominates at the population level, what does economic “rationality” look like inside firms? It does not look like solving equations. Instead, it looks like hard work, educated guesswork, and constant adjustment.

Alchian did not suggest business owners are indifferent to profits. On the contrary, the pursuit of profit remains the central objective of commercial activity. Business owners plainly want their firms to earn as much as possible. The difficulty is that, under conditions of uncertainty, no one knows in advance which decisions will produce the highest profits.

Alchian emphasized that economic success under uncertainty rarely results from perfect foresight. Firms operate by making conjectures about an unknowable future and adjusting when those conjectures prove mistaken. As he explained, “The greater the uncertainty of the environment, the greater the probability that actions will be taken which are not optimal.”

Managers and entrepreneurs therefore rely on intuition, experience, and experimentation. They try new pricing strategies, adjust production methods, explore new markets, and respond to feedback from customers and competitors. Some of these experiments succeed; others fail.

Consider a restaurant owner experimenting with menu items, pricing, and kitchen organization. She does not calculate the optimal menu mathematically. Instead, she observes which dishes sell, adjusts prices, revises recipes, and drops unpopular items. Over time, unsuccessful experiments disappear, while successful practices persist.

The same dynamic operates in large corporations. Firms test marketing campaigns, revise executive-compensation structures, restructure operations, and reallocate capital across divisions. Managers continually revise decisions in response to new information.

Alchian acknowledged that chance also plays an important role in this process. Some firms benefit from favorable circumstances, while others encounter unexpected setbacks. But chance does not imply pure randomness. What matters is how firms respond to these events. Firms that adapt to good or bad fortune are more likely to survive than firms that remain rigid in the face of changing conditions.

Profitability therefore emerges from a process of continual adjustment. Firms that repeatedly experiment, learn from mistakes, and revise unsuccessful strategies tend to move closer to profitable outcomes. Firms that fail to adapt eventually disappear.

Corporate Governance by Natural Selection

Alchian’s framework also sheds light on corporate governance. Governance debates often focus on whether boards are sufficiently independent, whether executive compensation properly aligns managers’ incentives with shareholder interests, or whether shareholder-voting mechanisms adequately discipline management. These discussions often assume governance arrangements must be carefully designed to achieve optimal results. But governance structures themselves evolve under competitive pressure.

Capital markets provide one important mechanism of selection. Firms that consistently underperform face higher costs of capital as investors demand greater returns to compensate for risk or shift investments toward more successful firms. Over time, this reallocates resources toward firms whose governance structures support profitable operations.

The “market for corporate control” provides an even more direct form of discipline. As Henry Manne famously argued, declining stock prices often signal managerial inefficiency. Those declines invite takeover bids by investors who believe they can operate the firm more effectively. When takeovers succeed, new owners frequently replace management, restructure operations, and adopt governance practices designed to improve performance. In this way, the possibility of acquisition encourages boards and executives to respond to competitive pressures before performance deteriorates too far.

Henry Butler and other scholars later emphasized the broader role financial markets play in this disciplinary process. Investors, analysts, and lenders continually evaluate corporate performance and governance practices. Firms that fail to meet expectations face declining share prices, restricted access to capital, and increased vulnerability to takeovers or activist intervention.

Executive labor markets reinforce these pressures. Managers associated with successful firms gain reputational advantages and expanded opportunities, while managers associated with persistent underperformance are less likely to retain leadership positions or secure future executive roles. Managerial practices that support firm survival therefore tend to spread through professional networks and imitation.

Corporate-governance arrangements such as independent directors, performance-based compensation, and enhanced disclosure requirements did not emerge all at once through centralized design. They developed gradually as firms, investors, and regulators observed which arrangements produced more reliable performance.

ESG and the Risks of Convergence

Alchian’s evolutionary perspective also offers a useful framework for thinking about contemporary debates over environmental, social, and governance (ESG) practices. Supporters argue ESG initiatives improve long-term corporate performance by strengthening reputation, retaining employees, or reducing regulatory risk. Critics worry such initiatives may distract firms from their core objective of generating profits.

In principle, Alchian’s framework suggests markets should eventually sort out these competing claims. Firms experiment with different strategies under conditions of uncertainty. Investors observe performance outcomes. Capital flows toward firms that demonstrate resilience and profitability, while less successful approaches gradually disappear. Over time, practices that support survival become more widespread.

This evolutionary process, however, depends on decentralized experimentation and effective selection mechanisms. When variation is suppressed or selection pressures weaken, the market’s ability to distinguish successful practices from unsuccessful ones may diminish.

Recent developments surrounding ESG illustrate this challenge. Large institutional investors, which collectively hold substantial stakes in many major corporations, have actively encouraged firms to adopt ESG policies. At the same time, regulatory initiatives increasingly require ESG disclosures and, in some cases, specific reporting frameworks. When governance practices spread primarily through coordinated pressure from large investors or through regulatory mandates, the process may begin to resemble centralized adoption rather than decentralized experimentation.

Under such conditions, the evolutionary mechanism Alchian described may operate less effectively. If many firms adopt similar practices simultaneously because of investor or regulatory pressure, markets have fewer opportunities to observe alternative approaches and compare their outcomes. Likewise, when regulatory frameworks constrain corporate responses, firms may have less freedom to adapt quickly to new information.

This does not mean ESG practices are necessarily inefficient. But it does complicate the evolutionary testing process that Alchian believed allowed markets to discover which business practices work best under conditions of uncertainty.

The Evolutionary Foundations of Law & Economics

Alchian’s evolutionary insight resonates with broader themes in law & economics. His framework complements Ronald Coase’s explanation of how transaction costs shape institutional structures and Harold Demsetz’s account of how property rights evolve in response to economic incentives. It also parallels Friedrich Hayek’s description of spontaneous order. Hayek emphasized that market institutions coordinate dispersed knowledge without centralized planning. Alchian showed that competitive selection produces profit-oriented behavior without requiring conscious optimization. In both accounts, order emerges from decentralized processes rather than deliberate design.

Alchian’s contribution therefore occupies an important place in the intellectual foundations of law & economics. He explained how competitive selection connects individual decision-making—often guided by guesswork, intuition, and imperfect information—to systematic market outcomes.

Why Alchian Still Matters

Seventy-five years after its publication, “Uncertainty, Evolution, and Economic Theory” remains a powerful reminder that markets do not require perfect rationality to function effectively.

Economic models often assume profit maximization, but Alchian showed such behavior can emerge even when decision-makers lack the information required to calculate optimal strategies. Firms survive not because they solve optimization problems, but because they adapt—and because firms that fail to adapt eventually disappear.

In a world characterized by uncertainty, prediction is difficult and mistakes are inevitable. Yet markets possess a remarkable capacity to learn from those mistakes. Firms experiment, adjust, and compete. Successful practices spread; unsuccessful ones fade away.

Profit maximization, in this sense, is not the product of perfect planning or flawless calculation. It is the outcome of competition, adaptation, and survival.

Markets do not reward omniscience. They reward firms that learn fast enough to stay alive.

Further Reading

• Armen A. Alchian, “Uncertainty, Evolution, and Economic Theory,” Journal of Political Economy, Vol. 58, No. 3 (1950)
• Friedrich A. Hayek, “The Use of Knowledge in Society,” American Economic Review, Vol. 35, No. 4 (1945)
• Ronald H. Coase, “The Nature of the Firm,” Economica, Vol. 4, No. 16 (1937)
• Milton Friedman, “The Methodology of Positive Economics,” in “Essays in Positive Economics,” University of Chicago Press (1953)
• Henry G. Manne, “Mergers and the Market for Corporate Control,” Journal of Political Economy, Vol. 73, No. 2 (1965)
• Armen A. Alchian and Harold Demsetz, “Production, Information Costs, and Economic Organization,” American Economic Review, Vol. 62, No. 5 (1972)
• Harold Demsetz, “Toward a Theory of Property Rights,” American Economic Review, Vol. 57, No. 2 (1967)
• Henry N. Butler and Larry E. Ribstein, “Opting Out of Fiduciary Duties: A Response to the Anti-Contractarians,” Washington Law Review, Vol. 65, No. 1 (1990)

The post ‘Uncertainty, Evolution, and Economic Theory,’ by Armen Alchian appeared first on Truth on the Market.

A little context. In 2006, Boeing and Lockheed Martin combined their launch divisions into a joint venture called United Launch Alliance (ULA). That’s what I mean when I say they “combined.” It wasn’t technically a merger, and the distinction matters once you get deep into antitrust doctrine. For simplicity, though, think of it as a merger to monopoly. That sounds especially bad in an industry responsible for launching national-security payloads.

By every static measure, this is the textbook nightmare.

Readers know I think textbook metrics are often a bad starting point. So I was intrigued by a recent paper from Ruibing Su, Chenyu Yang, and Andrew Sweeting arguing that the deal was probably a good idea after all.

Their key insight—missing from the standard model—is that every launch teaches engineers something. The team flying its 50th mission knows things the team flying its first mission does not. Before the “merger,” that knowledge accumulation was duplicated across two separate programs, with separate engineering teams, supply chains, and production systems. That creates a clear possibility for efficiencies in a very specific sense—not just cost cutting, but faster learning.

The question is whether those efficiencies were large enough to outweigh the standard monopoly problem.

Su, Yang, and Sweeting argue yes. Using a structural model packed with all the usual modern industrial-organization bells and whistles, they analyze the space-launch industry from 1985 to 2024. Their conclusion: the learning synergies were real and large enough to offset the harms from increased market power.

They also find that when the government committed to multiyear block buys—instead of shopping for launches one mission at a time—costs fell dramatically. Forward-looking procurement gave the supplier stronger incentives to invest in improving future performance, even without a direct competitor applying pressure.

That’s a serious empirical result for one industry. The harder question is what regulators were supposed to do with that possibility at the time.

‘Too Speculative’ Usually Means ‘Too Hard’

Every merger in a learning-intensive industry raises the same questions. Will the merged firm keep innovating, or grow complacent? What happens to entry when the incumbent has a decade of accumulated know-how and the challenger starts from scratch? Will the dominant firm use its position to lock up critical inputs and foreclose rivals?

In practice, regulators mostly punt. Merger review tends to focus on static effects—prices, concentration levels, market shares—while treating “dynamic efficiencies” as something vague and speculative floating out in the ether. That bias consistently cuts in one direction: against mergers in the industries where dynamic effects may matter most.

To be fair, studying dynamic efficiency is hard.

Economists can build full dynamic models that try to capture learning, investment, and long-run competition. But even the people building those models openly acknowledge the tradeoffs. Ariel Pakes and Ulrich Doraszelski, in their handbook chapter on “Applied Dynamic Analysis in IO,” caution that the framework “delivers very little in the way of analytic results of applied interest.” Steven Berry and Giovanni Compiani, surveying the same literature, are not much more optimistic: “[T]he attempt to add dynamics may create enough compromises that the result is not better than the static model.”

So, yes, if the only way to incorporate dynamics into merger review is through a five-year structural-modeling project, regulators will mostly keep ignoring them.

You Can’t Buy Experience

Let’s work through some alternatives. Instead of going full structural industrial organization, let’s think in simple price-theory terms. Strip away the details for a moment. There’s a relatively straightforward way to capture many of the important dynamics here.

The key idea is that output today affects costs tomorrow. A firm that produces more accumulates something valuable over time.

That’s a reduced-form way to capture several different phenomena. The rocket example relies on learning by doing, but it’s hardly unique. Retailers benefit from denser distribution networks. Airlines build advantages through route density. Manufacturers refine tooling and supplier relationships through sustained production volumes in ways smaller rivals struggle to match. In platform markets, the equivalent is an installed base.

The engineering details differ. Don’t let that distract from the economics. In each case, output today builds a productive stock that lowers future costs.

Notice that this differs from standard capital accumulation. There, investment and production are separate decisions that compete for scarce resources. Cash spent on a new machine is cash not spent making products. Time spent building the next semiconductor fabrication plant is time not spent operating the current one.

The productive stock here works differently. You can’t simply write a check for a year of launch experience or a denser airline route map. The only way to build those assets is by producing. Output and investment therefore do not compete with each other. They are the same decision. Producing more today is investing more today.

To circle back to mergers specifically, this framework does not explain every dynamic-efficiency claim. Some dynamics involve patent races, product repositioning, entry timing, demand-side network effects, or strategic investments chosen separately from output. Those may require different tools.

Most merger-efficiency claims, though, are more mundane. They involve scale, experience, density, know-how, or installed bases. Take T-Mobile and Sprint in 2019. The companies argued that combining their spectrum holdings and cell sites would allow them to build a higher-quality 5G network than either could alone. That’s fundamentally a network-density claim. The “stock” is network capacity, and producing more output helps build it.

Competition Falls. Production Might Rise.

A merger involving this kind of productive stock pulls in two directions at once.

Start with the familiar concern: less competition. After a merger, the combined firm no longer worries as much about losing customers to its closest rival because, in a sense, it owns the rival too. Before common control, if Boeing’s launch division raised prices, some customers would switch to Lockheed Martin. After common control, many of those “lost” sales stay inside the same organization. The merged firm recaptures business it used to lose.

That weakens the incentive to fight for the marginal customer. The predictable result is higher markups and less output.

You can think of this graphically. Common control rotates the marginal-benefit curve inward. Before the merger, winning a launch contract from your rival is a real gain. Afterward, winning business from another division of your own company is partly just stealing from yourself. The merged firm internalizes cross-product diversion, so the marginal benefit of expanding output falls. Quantity falls with it.

That’s the standard merger story.

But this framework introduces a competing force. If producing today builds productive capability for tomorrow, then every additional launch also generates experience, operational knowledge, and learning by doing. Those gains lower the effective marginal cost of future production.

Push that effect far enough, and the supply curve can actually slope downward. The firm still pays the immediate cost of the launch, but it is also effectively purchasing future cost reductions.

Under that logic, consolidation can increase the incentive to produce. A combined firm captures more of the returns from building productive capability, so producing today becomes more valuable. Larger production volumes accelerate learning. Greater internal coordination makes it more likely the resulting efficiencies will actually materialize, instead of being duplicated across separate organizations and supply chains.

Where the Textbook Graph Starts Misbehaving

To see both forces in a single picture, we need to think about the marginal cost of building this productive stock. How does learning change the firm’s marginal-cost curve? What’s the “price” of experience?

Physical capital has a rental rate—the amount you would pay each period to use it. Dale Jorgenson called this the “user cost of capital”: the implicit rental value of an asset the firm owns rather than leases. Experience and operating capability have a similar structure, even if nobody literally sends the firm a bill.

What’s the value of owning that stock? Each unit of output today adds to it and lowers future costs. The present value of those future savings is effectively the rental value of the stock. Producing one additional launch is therefore cheaper than the accounting cost suggests, because part of the expenditure is really purchasing future productivity improvements.

Call the normal accounting cost per launch the static marginal cost. Call the static cost minus this implicit “rental rebate” the dynamic marginal cost, net of rental. Static marginal cost is what shows up on the invoice. Dynamic marginal cost is what actually drives the firm’s production decision once it accounts for learning effects.

Pre-merger, in the top panel, the firm faces residual demand for its own product and a standard single-product marginal-revenue curve. Static marginal cost sits at 3. Dynamic marginal cost, net of the rental rebate, sits closer to 2. The firm produces where marginal revenue intersects dynamic marginal cost, and we can read off price and quantity from there.

Post-merger, two things change simultaneously.

First, marginal revenue rotates inward to “owned-product marginal revenue.” This is the standard diversion effect in picture form. If that were the only change, the merger would mechanically reduce output and raise prices.

Second, dynamic marginal cost rotates further downward to “post-merger dynamic marginal cost.” The combined firm captures more of the value from producing today. There is less duplication across engineering organizations. Higher combined output pushes the firm farther up the learning curve—or, equivalently, farther down the cost curve. More of the future cost reduction stays inside the firm instead of leaking to a competitor. The rental rebate grows larger, so the firm’s perceived marginal cost falls.

In the figure, the dynamic-marginal-cost shift dominates.

The picture also clarifies when the result flips. If the learning curve is relatively flat, dynamic marginal cost barely moves, and the marginal-revenue rotation wins. Quantity falls, prices rise…cats and dogs living together, mass hysteria.

In one sense, this is all almost tautological. If the force pushing quantity upward exceeds the force pushing it downward, output rises. Not exactly a profound insight. The value of the framework is that it helps us think systematically about when each force dominates.

That means thinking more carefully about how learning actually works. If experience spills over heavily to competitors, a merger does little to change how much of the rental rebate the firm captures. Dynamic marginal cost barely moves. The industry may already exhibit a downward-sloping cost structure even if no individual firm fully internalizes it. The outcome then depends on how steep the learning curve is, how private the learning remains, and how durable the resulting advantage proves to be. The theory helps organize the investigation.

It also pushes us to think about alternatives. A merger is one way to increase the returns from producing today, but it is hardly the only one.

Su, Yang, and Sweeting find that costs fell sharply when the government shifted from buying launches individually to committing to multiyear block purchases. That policy effectively guaranteed suppliers a stream of future orders. The mechanism is the same one illustrated in the figure. A larger committed order book increases the value of investing in learning today because the firm expects to produce the future launches that benefit from those improvements. The rental rebate gets larger. Dynamic marginal cost rotates downward.

Not Every Market Gets a SpaceX

The figure shows when a merger is most likely to generate genuine cost savings: when learning curves are steep, experience remains private, and the productive stock is durable.

We should be careful, though. Those same conditions also make entry harder. A new entrant starts with zero accumulated experience, while the merged firm sits on years of operational knowledge. If learning is steep, the cost gap will be large. If learning is private, entrants cannot easily catch up by poaching engineers. If the stock is durable, the advantage persists.

It is tempting to treat those entry barriers as the offsetting “cost” of the efficiency and simply net the two effects against each other. I’ve argued before that this gets the analysis backwards. Achieving scale is not an antitrust harm. Preventing rivals from achieving scale through better products or lower prices is not an antitrust harm either. That is what competition on the merits looks like.

The same logic applies to accumulated learning. If ULA’s experience made it harder for entrants to win contracts, that productive stock was doing exactly what productive stocks are supposed to do. The merger accelerated the accumulation of that stock by combining output. It did not do so by sabotaging competitors.

The actual antitrust harm must come from some separate exclusionary mechanism. The restraint is the problem, not the stock itself. Foreclosing key inputs. Locking up distribution. Raising rivals’ costs through means unrelated to the merged firm’s own productivity.

As it happens, ex post, entry turned out to be possible. SpaceX arrived and detonated the market structure so completely that much of this now feels almost quaint. Not every market gets a SpaceX, though.

Still, we have tools for thinking carefully about these problems. This kind of simplified framework does not replace a full structural model, but it gives you a way to reason through the question before building one—and a way to understand what the model is actually doing once you have.

Most importantly, it forces specificity. Which curves are shifting? What mechanism is moving them? Which effects are observable ex ante, and which only become visible after the fact?

If we can answer those questions clearly, we are already a long way toward understanding the market.

The post In Space, No One Can See Your HHI appeared first on Truth on the Market.

The carrier’s disappearance has transformed the JetBlue-Spirit merger litigation from an ordinary postmortem into a test case for how antitrust law should treat distressed challengers in concentrated network industries.

Protecting Competition, Minus the Competitor

The starting point is Judge William Young’s January 2024 decision enjoining JetBlue’s acquisition of Spirit under Section 7 of the Clayton Act after the U.S. Justice Department (DOJ) challenged the deal as anticompetitive. (The DOJ has jurisdiction over antitrust review of domestic airline mergers.)

In a Federalist Society discussion of the failed merger, I argued the case reflected a continuing skepticism in merger enforcement toward efficiencies, particularly claims that a merger could strengthen competition against larger incumbents. Events since then have sharpened a related critique: the court appears to have underweighted “out-of-market” efficiencies—benefits that would have accrued across a broader set of city-pair markets, even if travelers on certain overlapping routes lost the specific option of flying Spirit.

The critique is not that the route-level harms were imaginary. The government’s case followed a familiar Section 7 framework. Spirit was an ultra-low-cost carrier whose market entry often pushed fares downward. JetBlue, by contrast, operated as a higher-cost, higher-quality airline and planned to retrofit Spirit aircraft into the JetBlue product. In United States v. JetBlue Airways Corp., Judge Young concluded that, on certain routes, Spirit customers likely would face higher prices or fewer low-fare options if Spirit disappeared into JetBlue.

The objection, instead, is that the opinion treated those route-specific losses as dispositive, even though the merger’s broader competitive effect may have been to create a larger, more relevant JetBlue capable of challenging American, Delta, United, and Southwest—the “big four” U.S. airlines by market share. The court itself acknowledged that a stronger JetBlue would place competitive pressure on larger carriers and extend JetBlue’s higher-quality service to more customers. It nonetheless concluded those broader benefits could not justify the transaction if the merger substantially lessened competition in any relevant market.

What Happens When the Maverick Crashes?

Spirit’s subsequent collapse has made that reasoning look far less secure. After the ruling, Spirit entered bankruptcy, struggled to restructure, and ultimately ceased operations. The Associated Press reported that Spirit shut down after 34 years, ending the carrier most associated with rock-bottom base fares. Other accounts highlighted the fallout for passengers and the loss of roughly 17,000 jobs.

Christopher Gowen argued in a Bloomberg Law essay that Spirit’s demise should serve as a cautionary tale for merger review. In his telling, the government succeeded in preserving an independent Spirit in legal theory, but not in market reality.

That is an uncomfortable lesson for merger law. A court can block a merger to preserve a maverick competitor—a disruptive firm that constrains rivals through aggressive pricing or innovation. But if the maverick is too weak to survive on its own, the injunction may leave consumers with neither the original firm nor the stronger challenger the merger might have created.

The legal implications extend beyond airlines. Merger analysis depends on a realistic “counterfactual”—a prediction of what competition would look like if the deal does not occur. A static, route-by-route snapshot can overstate the competitive significance of a distressed carrier that is shrinking, financially unstable, or unable to secure aircraft, labor, capital, or airport access on workable terms.

To be sure, courts are understandably cautious here. The failing-firm defense—which permits an otherwise anticompetitive merger when the target company is effectively doomed—is notoriously difficult to establish. Judges do not want firms casually invoking financial distress to justify consolidation.

Still, there is middle ground between a formal failing-firm defense and treating a distressed company as though it were a durable, independent competitive force. In a network industry like air travel, a weakened-competitor analysis should ask whether the allegedly preserved competition is likely to survive long enough, and at sufficient scale, to constrain incumbents. JetBlue-Spirit may become the clearest recent example of what happens when courts answer that question too optimistically.

The Airlines Deregulated. The Airports Didn’t.

The post-Spirit U.S. airline industry presents a paradox. On paper, the market still includes plenty of names: American, Delta, United, Southwest, JetBlue, Alaska-Hawaiian, Frontier, Allegiant-Sun Country, Breeze, Avelo, and others. In practice, network effects, airport access, frequent-flyer programs, corporate contracts, and hub dominance give the largest carriers durable advantages.

The DOJ’s own recent account notes that American, Delta, Southwest, and United control roughly three-quarters of U.S. domestic markets, and that the largest airlines have absorbed dozens of rivals since deregulation. In a November 2025 speech, Deputy Assistant Attorney General Dina Kallay described airline competition as a core consumer issue and linked current concerns to the industry’s long history of consolidation.

Hub-and-spoke competition sits at the center of that market power. American, Delta, and United operate dense hub systems that let them offer business travelers frequent departures, extensive connections, lounges, loyalty perks, and corporate-contracting arrangements that point-to-point or leisure-focused airlines struggle to match. Those hubs also facilitate price discrimination. Competition may be fierce on major trunk routes, while passengers traveling to or from fortress hubs face fewer meaningful choices.

Southwest historically constrained that system through point-to-point service. More recently, though, Southwest has moved closer to the legacy-carrier model, relying more heavily on sophisticated revenue management, loyalty-program monetization, and ancillary-fee strategies. Spirit imposed a different kind of constraint. It was not a network-based rival to the legacy carriers, but an ultra-low-cost fare anchor. The so-called “Spirit effect” often disciplined prices even where Spirit’s market share remained modest.

Blocking the JetBlue-Spirit merger may have strengthened the Big Three’s position in two ways. First, it eliminated JetBlue’s most plausible path to rapid scale. Organic growth is slow in an industry constrained by aircraft shortages, engine problems, pilot supply, and limited airport access. Spirit’s Airbus fleet offered JetBlue a ready-made path to greater national relevance. Without Spirit, JetBlue remains an important but subscale competitor—strong in parts of the East Coast, but lacking the network depth of American, Delta, and United.

Second, Spirit’s liquidation fragmented its competitive assets. Aircraft, gates, slots, employees, and routes can all be redistributed. But the ultra-low-cost-carrier business model, network structure, brand identity, and pricing discipline do not automatically survive asset sales. The court’s hoped-for replacement through new market entry must now occur piecemeal, through selective route expansion and asset transfers, rather than through a functioning integrated airline.

A related point deserves emphasis: some of the most durable barriers to airline competition arise not from airline conduct alone, but from legal and regulatory limits on airport capacity. At the most congested airports, the Federal Aviation Administration (FAA) uses runway slots to ration scheduled operations. FAA slot-administration materials identify John F. Kennedy International Airport, LaGuardia Airport, and Reagan National Airport as slot-controlled airports, while describing schedule-facilitation processes at O’Hare International Airport, Los Angeles International Airport, Newark Liberty International Airport, and San Francisco International Airport.

Those restrictions may be justified by congestion, runway limits, air-traffic-control constraints, and delay externalities. They also turn airport access into a scarce asset. Scarcity predictably favors incumbents that already control slots, gates, terminal space, and commercially viable operating positions. New entry, therefore, is not simply a matter of an airline deciding to add service. It often requires access to a scarce bundle of slots, gates, takeoff and landing rights, ground facilities, and commercially usable flight times.

The deeper problem is that expanding airport capacity is legally and politically difficult. Major airport expansions, new runways, terminal projects, and new airport construction all trigger environmental review under the National Environmental Policy Act (NEPA) and related environmental statutes. FAA guidance states that environmental review must be completed before covered airport projects can begin. Local zoning rules, land-use controls, noise restrictions, community opposition, airport curfews, emissions review, surface-transportation bottlenecks, and funding constraints can further delay or block expansion.

The Government Accountability Office (GAO) has long recognized that airport-access barriers include physical constraints such as slots, gates, and noise restrictions. In key metropolitan areas, the competitive question is therefore not merely whether Frontier, Breeze, Allegiant-Sun Country, JetBlue, or Alaska-Hawaiian want to enter a market. The real question is whether they can obtain the airport access necessary to discipline incumbent hub carriers on commercially meaningful terms.

If You Can’t Merge, Collaborate

Recent industry developments illustrate both mitigation and risk. Rivals are already moving into some former Spirit markets. The Wall Street Journal reported that low-cost competitors are carving up Spirit’s former routes and airport slots, with Breeze, Frontier, and Allegiant seeking growth opportunities while larger airlines pursue valuable airport assets.

That redeployment will soften Spirit’s disappearance in some leisure markets and smaller airports. But selective entry is not the same thing as systemwide replacement. An airline that adds one or two former Spirit routes does not necessarily recreate Spirit’s low-cost network, pricing discipline, or willingness to stimulate demand through ultra-low base fares.

The same dynamic appears in new consolidation among smaller carriers. Earlier this month, Allegiant announced it had completed its acquisition of Sun Country, creating a larger leisure-focused airline serving nearly 175 cities with a combined fleet of 195 aircraft. The deal reduces the number of independent low-cost and leisure-focused airlines. At the same time, it may create a stronger scaled competitor in the very segment best positioned to backfill some of Spirit’s abandoned markets.

That is the recurring airline-antitrust tradeoff. A merger among smaller carriers may eliminate competition on certain routes while increasing the merged firm’s ability to challenge larger network airlines across many others. Whether courts and agencies should block or permit that tradeoff depends on the counterfactual—and on whether the acquired capacity would otherwise remain a viable competitive force.

The Alaska Airlines-Hawaiian Airlines integration, announced last month, reflects a more permissive regulatory approach. The airlines described their shared passenger-service system as “a significant integration milestone that provides guests flying with Alaska Airlines and Hawaiian Airlines a more seamless and consistent travel experience from booking to boarding across a growing global network.” The U.S. Department of Transportation (DOT) allowed the transaction to proceed after securing binding public-interest commitments. According to the department’s announcement, those commitments covered rewards programs, essential flight service, rural access, Honolulu hub access, family seating, and military benefits.

The contrast with JetBlue-Spirit is revealing. Alaska-Hawaiian involved fewer direct overlaps and less elimination of an ultra-low-cost fare constraint. It also rested on a clearer network-complementarity rationale—one regulators could condition and supervise without effectively destroying the transaction. Future airline mergers will likely be framed in similar terms: limited overlap, complementary networks, service preservation, loyalty-program protections, and concrete commitments designed to make claimed benefits verifiable.

Joint ventures and looser collaborations may become even more important than outright mergers. JetBlue and United’s “Blue Sky” collaboration, for example, links loyalty programs and customer benefits while emphasizing that the airlines will continue independently managing and pricing their routes, frequencies, and promotions.

Blue Sky, Red Flags

The BlueSky collaboration’s careful drafting is no accident. The American Airlines-JetBlue Northeast Alliance was enjoined because the DOJ and the courts viewed it as a de facto merger in the Northeast—a coordination arrangement that reduced the airlines’ incentives to compete independently on routes, schedules, and capacity. In her recent airline-competition speech, the Deputy Assistant Attorney General Kallay described the Northeast Alliance as a cautionary example and suggested that domestic “metal-neutral” arrangements—agreements that make airlines financially indifferent about which carrier actually transports a passenger—will face intense scrutiny.

The political and enforcement environment reinforces that caution. In July 2025, Sen. Richard Blumenthal (D-Conn.) sent a letter to the chief executives of United and JetBlue questioning whether the Blue Sky partnership resembles the failed Northeast Alliance. Blumenthal pointed to reciprocal loyalty-program features, airport-infrastructure implications, and potential effects on corporate accounts.

The letter itself carries no legal significance. Still, it reflects the concerns likely to shape future government oversight. Limited loyalty reciprocity, interline cooperation, or customer-facing benefits that stop short of coordinating capacity, schedules, or pricing may survive antitrust review. Deeper domestic joint ventures that pool revenue, allocate markets, coordinate frequencies, or create metal-neutrality are far more likely to draw DOJ challenge. The more a collaboration resembles a merger without Hart-Scott-Rodino review, the more skeptical courts are likely to become.

The DOJ and DOT have also made clear that airline competition remains an active enforcement priority. In October 2024, the agencies launched a broad public inquiry into the state of competition in air travel. The inquiry sought information on consolidation, exclusionary conduct, airport access, aircraft manufacturing, sales channels, pricing, rewards programs, and labor issues.

Dina Kallay’s 2025 remarks likewise emphasized airline competition, but with a notable twist: the need to address not only private restraints, but also public barriers to entry. That distinction matters. A serious airline-competition agenda cannot consist solely of blocking private mergers. It must also examine whether government-created scarcity—including slot rules, airport-access limits, infrastructure bottlenecks, and regulatory delays—helps entrench incumbent market power.

The Next Airline Deals Will Look Different

What comes next? The most likely near-term transactions involve acquisitions of Spirit assets: slots, gates, aircraft, airport facilities, route authorities, and personnel. Where those assets are scarce, the DOJ and DOT should generally prefer distributing them to non-dominant carriers. Sales of Spirit assets at LaGuardia, Newark, Fort Lauderdale-Hollywood International Airport, Orlando International Airport, or other constrained airports to American, Delta, or United would raise greater concerns than sales to Frontier, Allegiant-Sun Country, Breeze, JetBlue, or Alaska-Hawaiian.

The competitive value of those assets lies in enabling entry and expansion by challengers. Asset purchases are usually easier to clear than mergers. That changes, however, when the transaction strengthens fortress hubs or removes scarce airport access from smaller rivals.

A Frontier-led acquisition of Spirit remnants—or further consolidation among ultra-low-cost and leisure carriers—would present a more complicated case. Before Spirit’s collapse, a Frontier-Spirit merger would have been a straightforward horizontal merger between two closely competing ultra-low-cost carriers. In the wake of Spirit’s bankruptcy, a transaction involving scattered assets rather than an operating airline becomes easier to defend. Regulators would likely ask whether the buyer is preserving capacity that otherwise would disappear, whether meaningful overlaps remain, and whether the deal restores some degree of low-fare discipline. Courts may also view those arguments more sympathetically after JetBlue-Spirit, because the liquidation counterfactual is no longer hypothetical.

A future JetBlue merger would face skepticism, but not automatic condemnation. JetBlue’s strongest argument remains scale: without more aircraft, slots, and network breadth, it cannot seriously challenge the Big Three. Its biggest problem is its recent antitrust history. The Northeast Alliance was condemned as anticompetitive, while the Spirit merger was blocked because JetBlue planned to replace Spirit’s ultra-low-cost model with a higher-cost, higher-fare product.

Any future JetBlue deal would therefore need to look very different. JetBlue would likely need to show limited route overlap, preservation or expansion of low-fare capacity, no elimination of a maverick business model, and perhaps divestitures to low-cost entrants. It would also need to present a more concrete and measurable out-of-market efficiencies case than it did in the Spirit litigation.

A merger involving one of the Big Three legacy airlines would be the hardest case of all. Transportation Secretary Sean Duffy has suggested there may be “room for some mergers” in aviation, while emphasizing that deals would be reviewed case by case and larger airlines might need to divest assets to avoid excessive concentration. That reflects a plausible policy middle ground: openness to consolidation that creates stronger challengers, combined with skepticism toward deals that deepen hub dominance or transfer scarce airport access to incumbents.

A Big Three acquisition of a meaningful low-cost or regional competitor would almost certainly trigger DOJ opposition absent a strong failing-firm or failing-division justification, substantial divestitures, and enforceable service commitments.

The most likely equilibrium is selective consolidation paired with strict limits on coordination. The DOJ and DOT may tolerate mergers among smaller or complementary carriers when the result plausibly creates a stronger challenger, particularly if the parties accept conditions relating to routes, slots, loyalty programs, and service commitments. By contrast, regulators will likely resist transactions that hand scarce airport access to dominant incumbents or create metal-neutral domestic coordination.

Courts, meanwhile, will continue applying Section 7 market by market. But merging parties are likely to press harder on failing-firm, weakened-competitor, and out-of-market efficiencies arguments. The unresolved question is whether courts will adapt merger doctrine to account for an uncomfortable possibility: blocking a merger may sometimes accelerate the disappearance of the very competition antitrust law seeks to preserve.

When Antitrust Wins and Competition Loses

Spirit’s disappearance should not be read as proof that all airline mergers are beneficial. It should instead underscore how unusually dependent airline antitrust is on counterfactual assumptions. A static snapshot can mislead in an industry where aircraft shortages, fuel-price shocks, airport access constraints, loyalty ecosystems, and bankruptcy risk can rapidly reshape competition.

If the post-Spirit market ultimately leaves the Big Three more secure, the antitrust failure was not simply that JetBlue and Spirit were kept apart. It was that the legal system protected route-level competition without preserving the only carrier supplying much of that competitive pressure in the first place.

Over the longer term, antitrust enforcement alone cannot solve a scarcity problem created partly by public regulation. A more durable procompetitive agenda would pair merger and conduct enforcement with regulatory and legislative reforms that expand airport capacity, accelerate review of capacity-enhancing projects, rationalize slot-allocation rules, discourage slot hoarding, make divested slots and gates available to genuine entrants, and reduce unnecessary legal barriers to new airports and runways.

Future airline antitrust should therefore ask not only whether a deal eliminates a competitor, but whether blocking the deal realistically preserves competition—and whether public policy has made entry so difficult that incumbents remain protected even when merger enforcement formally “wins.”

In the end, Spirit’s collapse may become the uncomfortable reminder that antitrust can preserve competition on paper while watching it disappear at the gate.

The post The Case of the Vanishing Competitor appeared first on Truth on the Market.

Gov. Gavin Newsom signed Executive Order N-6-26 today, setting California’s workforce agencies in motion on directives involving research reviews, revisions to the state’s Worker Adjustment and Retraining Notification (WARN) Act, studies of new safety-net programs, a review of collective bargaining frameworks, an employment dashboard, and—near the end—a study of programs that would redirect artificial intelligence (AI) company revenues toward state-selected applications.

The animating concern is AI-driven labor disruption. Newsom’s order treats that disruption as sufficiently imminent to justify building new regulatory infrastructure across California’s workforce apparatus.

The empirical literature suggests that concern is running well ahead of the evidence. That includes a literature review published by the International Center for Law & Economics (ICLE) that I co-authored with Kristian Stout.

The Labor Panic Is Outrunning the Data

Generative AI has spread faster than any comparable technology. By late 2024, nearly 40% of U.S. adults ages 18-64 reported using AI tools—a pace that exceeded personal computers and the internet at comparable stages of adoption, according to a 2025 National Bureau of Economic Research (NBER) working paper by Alexander Bick, Adam Blandin, and David Deming.

The productivity gains are real and remarkably consistent across settings. Controlled studies show GitHub Copilot users complete coding tasks 55.8% faster. A randomized experiment by Shakked Noy and Whitney Zhang found ChatGPT reduced professional-writing completion times by 40% while improving quality scores by 18%. A Fortune 500 customer-support deployment produced a 15% average productivity gain, including a 36% gain for workers in the bottom skill quintile.

Those gains have not produced aggregate job destruction. The Budget Lab at Yale found no clear correlation between AI exposure and unemployment through August 2025. Jonathan Hartley, Filip Jolevski, Vitor Melo, and Brendan Moore found that, by December 2025, 35.9% of U.S. workers were using generative AI, with small positive wage effects and no statistically significant declines in job openings or aggregate employment in exposed occupations. Anders Humlum and Emilie Vestergaard linked survey-reported ChatGPT adoption to Danish administrative records across 11 occupations and found essentially no effects on earnings or hours through 2024.

Employment effects do appear, but in a narrower and more specific place: entry-level positions. Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen found that workers ages 22-25 in highly AI-exposed occupations experienced employment declines of roughly 16% relative to trend following ChatGPT’s release, while employment among senior workers held steady. Bouke Klein Teeselink found a similar pattern in UK data, with losses concentrated among junior roles, advertised salaries in exposed occupations declining, and average firm-level compensation rising as firms shed lower-paid entry positions.

Across these studies, the pattern is task reallocation and adjustment at the start of workers’ careers—not mass displacement. That distinction matters. WARN Act expansions, new safety-net programs, and revenue levies on AI companies are calibrated for broad labor-market disruption. The available evidence describes something much narrower.

Before New Mandates, Try Using the Old Ones

One of the executive order’s “whereas” clauses makes an admission that sits awkwardly with what follows. The order states that “California already has robust worker protection laws that apply to firms adopting emerging technologies,” and that existing programs like Work Sharing “are currently underutilized.”

Work Sharing allows employers to reduce employees’ hours instead of conducting layoffs, while unemployment insurance partially compensates workers for lost income. The program is designed for exactly the sort of gradual, partial labor adjustment that AI appears to be producing. If employers are not using it, the more likely explanation is lack of awareness and administrative friction—not a gap in the law.

Directive 3(b) sensibly responds to that diagnosis by calling for a plan to expand awareness of and enrollment in Work Sharing. That provision, though, sits alongside 13 others that would build new regulatory mechanisms atop programs the order itself acknowledges are already underused.

Before California imposes new mandates, it should answer a more basic question: Why are existing tools underutilized, and could better outreach address the problem without expanding the regulatory state?

Regulating AI Like a Factory Shutdown

Directive 2 instructs the Labor and Workforce Development Agency to recommend revisions to California’s WARN Act to make it “responsive to, and effectively provide[] early warning data on, emerging industry trends.”

California’s WARN Act addresses discrete mass-layoff events. If a firm closes a facility or lays off 50 or more workers, the law requires 60 days’ advance notice so employees have time to prepare.

The disruption described in the executive order is different in kind. The concern is not factory closures or sudden mass layoffs, but individual workers in AI-exposed occupations gradually losing hours or positions as firms reorganize tasks over time. Extending WARN-style notification requirements to that sort of reorganization could mean requiring advance notice whenever a firm reallocates work from employees to AI tools.

The compliance burden from such a regime would fall especially hard on smaller California employers. More fundamentally, the evidentiary case for expansion remains thin. The available data shows entry-level hiring reductions and task reallocation—not the plant closures and mass layoffs WARN was designed to address. Expanding the law to cover AI-driven labor adjustment would require rewriting the statute’s underlying logic.

That matters because California is home to 33 of the world’s top 50 private AI companies. The compliance costs from such a rewrite would fall directly on the industry the executive order’s own preamble identifies as a strategic asset.

The State’s New Idea: Tax AI, Then Pick Winners

Directive 9 asks the Government Operations Agency to recommend options, including “voluntary or mandatory programs that direct a portion of revenue generated by AI companies to support beneficial deployments of AI that otherwise would not be pursued based solely on market incentives.”

Economically, that amounts to a proposal to study a levy on AI-company revenues, with the proceeds directed toward state-selected applications. The proposal rests on two claims: first, that AI markets systematically underprovide socially beneficial uses; and second, that California officials can identify and fund those uses more effectively than private actors can.

The first claim requires identifying a concrete market failure—a situation in which the social benefits of an AI application genuinely exceed what any private buyer would pay for it. The executive order asserts such a failure without establishing one. If AI tools that improve government services, expand health care access, or address climate challenges are genuinely valuable, there are—and will be—actual buyers for them.

The Organisation for Economic Co-operation and Development’s (OECD) 2025 cross-country survey found that 31% of small and medium-sized enterprises had already adopted generative AI by 2024 without state direction, with most citing improved performance and reduced workloads. Government agencies are themselves buyers. The executive order’s own recitals note that California has already signed memoranda of understanding with NVIDIA, Adobe, Google, IBM, and Microsoft for AI-literacy programs. Those are voluntary market arrangements, not revenue levies.

The second claim runs into a more fundamental problem: information about which AI applications create the most value is dispersed across millions of firms, consumers, and developers. A mandatory revenue diversion would replace that distributed information with the preferences of the agencies allocating the funds. There is little reason to assume those preferences will reliably track actual social value.

The proposal also raises a straightforward competitive concern. If California requires AI companies to redirect revenues toward state-directed purposes, it creates incentives for those firms to incorporate or expand operations elsewhere. The executive order itself describes California’s AI dominance as a strategic asset. Strategic assets, though, are portable. Policies that structurally disadvantage AI firms operating in the state should clear a high evidentiary bar before adoption. This proposal does not.

Some Parts of the Order Actually Fit the Evidence

Several provisions in the executive order are better matched to the available evidence.

Directive 7 calls for an employment dashboard using unemployment-insurance data to track AI’s effects across sectors in real time. That is exactly the kind of infrastructure policymakers need before committing to larger interventions. One genuine constraint in the current debate is the lack of reliable, real-time labor-market data on AI’s effects. Building monitoring capacity now would allow future policy decisions to respond to actual disruption, rather than speculative forecasts.

Directive 5’s workforce-training review also addresses a documented productivity complement. Controlled experiments consistently find that AI’s productivity gains are largest when workers know how to evaluate AI outputs critically—when to rely on them and when to override them. Fabrizio Dell’Acqua and colleagues found that Boston Consulting Group consultants working within AI’s capability boundaries improved performance substantially, while consultants using AI for tasks just beyond those boundaries performed worse because they overrelied on plausible but incorrect outputs. Training workers to recognize that distinction has real value. California’s community college system, which serves more than 2.1 million people each year, along with university extension programs, is well-positioned to provide that training at scale.

Directive 11, which instructs the Governor’s Office of Business and Economic Development (GO-Biz) and the California Office of the Small Business Advocate (CalOSBA) to support small-business AI adoption, likewise targets a concrete barrier. The OECD found that 50% of non-adopting small businesses cited insufficient internal expertise as a reason for not adopting AI tools. Technical assistance and outreach can reduce that barrier at relatively low cost, without imposing additional compliance burdens on the firms receiving help.

Regulate the AI Economy You Have, Not the One You Fear

Executive Order N-6-26 contains provisions calibrated to two very different visions of AI’s labor-market effects.

The data-gathering and training provisions align with the adjustments the evidence actually documents: entry-level disruption, task reallocation, and a growing need for workers to build AI literacy. By contrast, the proposed WARN Act revisions and revenue-levy study are calibrated to a mass-displacement scenario the evidence does not yet support.

That distinction matters because the two categories carry very different cost profiles. Building an employment dashboard and expanding community-college AI training impose relatively low costs while generating useful information and human capital. WARN Act expansions and mandatory revenue diversions, by contrast, would impose real compliance costs on California employers today based on a disruption that may never arrive in the form the order anticipates.

The labor adjustment currently underway is narrower and more specific. AI appears to be compressing career ladders in some occupations, reducing demand for the discrete tasks that traditionally served as professional entry points, while leaving senior employment largely intact. That is a real problem for younger workers trying to build skills, experience, and credentials. Better on-ramps to midlevel work, expanded AI-literacy training, and redesigned career pathways address that problem directly. Mandatory notice requirements and revenue levies do not.

California also has a concrete interest in getting the calibration right. The state’s AI sector—which includes 33 of the world’s top 50 private AI companies—is producing the productivity gains the executive order seeks to distribute more broadly. Regulatory costs that push those firms toward lower-cost jurisdictions would not redistribute AI’s benefits to California workers. They would relocate them.

The provisions worth enacting now are the ones that build information and workforce capacity: the employment dashboard, the workforce-training review, and small-business outreach. The provisions worth deferring are those premised on speculative mass displacement. If disruption at that scale eventually materializes, the data infrastructure created by the order will reveal it, and California can respond with evidence in hand, rather than panic in search of a justification.

The post The AI Jobs Panic Comes to Sacramento appeared first on Truth on the Market.

Most of the current action has to do with consumer protection. That’s par for the course, really. Apart from issuing the ill-fated noncompete rule, since vacated—comments here if anyone wants a recap—FTC rulemaking has been almost wholly on the consumer-protection side for as long as there’s been FTC rulemaking and a consumer-protection side to the agency. “Unfair or deceptive acts and practices,” or the FTC’s UDAP authority, were added to Section 5 in 1938 via the Wheeler-Lea Amendments, so it’s been a while.

For one thing, there’s no controversy over whether Congress has granted the commission substantive rulemaking authority over consumer-protection matters. It has—both a general authority under Section 18 of the FTC Act to prescribe rules against specific acts or practices that violate Section 5’s UDAP prong, and authority under various statutes that charge the FTC with adopting and enforcing particular restrictions addressing specific issues or practices. These include, among others, the Fairness to Contact Lens Consumers Act, the Children’s Online Privacy Protection Act, and the Fair Credit Reporting Act, later amended by the Dodd-Frank Wall Street Reform and Consumer Protection Act, under which most—but not all—regulatory authority shifted to the Consumer Financial Protection Bureau.

But some competition-adjacent rulemaking remains under the Hart-Scott-Rodino Antitrust Improvements Act of 1976 (HSR Act).

Rulemaking under the HSR Act is not exactly competition rulemaking. It doesn’t directly regulate which mergers are lawful and which are not. Still, it regulates the process by which mergers are screened and imposes affirmative obligations on firms contemplating mergers and acquisitions—at least for transactions above the filing threshold. Hence, these are competition-adjacent rules, and HSR rulemaking is competition-ish.

HSR Rulemaking Gets a Do-Over

That brings us to a joint FTC/Department of Justice (DOJ) request for information (RFI) “on the effectiveness of the Hart-Scott-Rodino Antitrust Improvements Act (‘HSR Act’)’s premerger reporting requirements.” That’s about as preliminary as rulemaking gets, as it’s neither a notice of proposed rulemaking (NPRM) nor even an advance notice of proposed rulemaking (ANPRM)—a more preliminary stage required by statute for the FTC’s consumer-protection rulemaking under Section 18 of the FTC Act, but not a stage exclusive to Section 18 rulemaking or the FTC.

Well, it is and it isn’t all that preliminary. As the RFI recounts, the commission published an NPRM on revisions to the HSR reporting requirements and form in June 2023, followed by a final rule in November 2024, both under the energetic—if not so process-oriented, or consumer-welfare-oriented—leadership of former Chair Lina M. Khan. Side note: yes, it’s an FTC/DOJ rule, but the FTC takes the lead on HSR rule revisions, in consultation with DOJ. HSR NPRMs and final rules are published in the Federal Register by the FTC “with the concurrence of the Assistant Attorney General, Antitrust Division, Department of Justice.” As per Congress, this is the way.`

As I pointed out in another post, there was nothing odd, in the abstract, about updating the HSR reporting requirements—something the FTC, with DOJ concurrence, had done many times before. Indeed, there might have been a rough consensus across antitrust law and economics that some updates to the form and process were warranted.

At the same time, the 2023 NPRM shot for the moon, and then some. Some of the proposals were downright ludicrous. As I said at the time: 

In a nutshell, the proposed revisions are controversial because they promise to make pre-merger filing more cumbersome and, not incidentally, more costly, and it’s not at all clear what the payoff is likely to be. 

The International Center for Law & Economics’ (ICLE) formal comments on the NPRM’s overreaching motley provide more detail, and ICLE was hardly alone in its critique (see also Gus Hurwitz, Alden Abbott, the Global Antitrust Institute, Bilal Sayyed on behalf of TechFreedom, the Information Technology and Innovation Foundation, and the U.S. Chamber of Commerce.)

The final rule was at least somewhat responsive to comments and, no doubt, to haggling among the commissioners. The latter was noted in concurring statements by then-commissioners Melissa Holyoak and Andrew Ferguson, with Holyoak providing a useful summary table of deletions and other adjustments made to the initial proposal. To be fair, the final rule was considerably better than the NPRM.

Questions remained, however. Among them was a quick-and-dirty—and dubious—cost-benefit analysis. The U.S. Chamber of Commerce, among others, brought suit successfully, and on Feb. 12, 2026, Judge Jeremy D. Kernodle of the U.S. District Court for the Eastern District of Texas issued an order of vacatur. Kernodle found that the rule exceeded the FTC’s statutory authority because the commission had failed to show that “the Rule’s claimed benefits will ‘reasonably outweigh’ its significant and widespread costs,” and that the rule was arbitrary and capricious for “much the same reason.” The 5th U.S. Circuit Court of Appeals subsequently denied the FTC’s motion for a stay pending appeal. 

So now we’re back to the prior version of the rule, including the pre-2025 form and, not incidentally, the RFI. Inquiry seems an appropriate first step, and the RFI reasonably proposes various topics of interest and poses various reasonable questions. Both Ferguson and Holyoak had remarked on lingering doubts about certain provisions in the now-vacated final rule. Taking a beat and asking additional questions makes more sense than rushing to another NPRM.

The RFI notes that the updated HSR form was in use for more than a year—from the final rule’s publication to the order of vacatur. Further:

During that period, practitioners prepared, and the Agencies’ staff reviewed, over 3,000 filings for a wide range of transactions. Commission staff have also answered many questions from practitioners about complying with the Updated Form. Based on this experience, the Agencies’ staff and practitioners have obtained insights about what aspects of the Updated Form may be improved. 

My First Very Modest Proposal: Agencies share what they learned during that period. That might include, at least, informal observations from enforcement staff about which information was more—or less—useful. As I noted in November 2024:

…new requirements regarding submission of information on “non-horizontal” or supply relationships still seem excessive. Not because such information couldn’t be pertinent to a merger investigation, but because it is not typically useful to a preliminary screen and can be obtained in those cases where it’s more likely to be pertinent—likewise for required submissions about, e.g., private-equity acquisitions, “roll-up” strategies, and interlocking directorates.

What new information—if any—made a material difference, either in generating second requests or avoiding them? In decisions to file complaints? How often?

In addition, how does the agencies’ research staff—notably, staff in the FTC’s Bureau of Economics—analyze information gathered during that year, and previously? That information is not necessarily confined to matters at the enforcement margin. It could help show what new information is not just possibly useful, but likely to be most useful to the screening process.

The RFI sensibly asks about the time, labor, and financial costs associated with HSR notification requirements. That is no small matter, not least because the rule’s failure to show that its “claimed benefits will ‘reasonably outweigh’ its significant and widespread costs,” as required by the HSR Act itself, proved fatal to the rule. Gathering better information on compliance costs is a start, even if an open RFI has limits as a means of data gathering. Still, it’s hard to balance benefits and costs without a clear sense of the benefits. More focused reflection on the agencies’ learning up front might well lead to a more focused and productive RFI.

Getting this right may take a bit of time, but it will produce a more useful rule, and a more durable one, to the benefit of both competition and enforcement.

But mostly, it’s been about consumer-protection regulation.

A Premature Delivery

There’s also the recent ANPRM on unfair and deceptive fee practices in online food and grocery-delivery services. For the very, very short version, ICLE’s recently submitted comments recognize the potential for Section 5 violations and the need for vigorous enforcement of the FTC Act.

We also recognize that an ANPRM is a preliminary stage in rulemaking—one that doesn’t even require the FTC to propose specific regulatory requirements. At the same time, it is a distinct statutory requirement, and one of several process restrictions expressly imposed on such rulemaking in Section 18 through the Magnuson-Moss Warranty Act and related amendments to the FTC Act.

Based on those restrictions, among other things, we argue that the ANPRM is premature. The FTC has relatively limited enforcement experience in this area. There also remains room for continued federal and state law enforcement against harmful fee practices, new legislative initiatives at the state level, and wide variation among the practices and businesses that might be subject to such regulations.

No Room for a Sectorwide Rule

Similarly, Eric Fruits and I submitted comments on behalf of ICLE in response to the FTC’s ANPRM on “Unfair or Deceptive Rental Housing Fee Practices.” There, too, we noted real consumer-protection concerns, the potential for Section 5 violations, and enforcement efforts by the FTC and the states.

But we raised substantive concerns about the breadth of the inquiry and, again, both substantive and process concerns about the misfit between the agency’s enforcement experience and the requirements for Section 18 rulemaking. We concluded, in brief, that “the record does not justify a sector-wide Magnuson-Moss rule.”

At least one question seemed at odds with prior FTC policy positions on price and cost disclosures. For example, in comments addressing a New York State bill that would have regulated contractual relationships between health plans and pharmacy benefit managers, FTC staff noted that certain disclosure requirements were “analogous to requirements that firms reveal aspects of their cost structures to customers.” Staff recognized the importance of consumer access to truthful and non-misleading price information. At the same time, the comments noted that “[t]here is no theoretical or empirical reason to assume that consumers require sellers’ underlying cost information for markets to achieve competitive outcomes.”

Further, some cost disclosures might tend to undermine price competition, to consumers’ detriment. Indeed, the RFI itself cautions against submitting “competitively sensitive information, such as costs ….”

Rules mandating cost-reflective pricing do not merely regulate how firms present price information; they govern the relationship between firms’ costs and prices. In that regard, they can function as cost-of-service ratemaking—a form of price regulation historically applied to public utilities under distinct statutory authority.

There was an added wrinkle in the RFI’s consideration of “unfair” practices. Section 5(n) of the FTC Act expressly links the competition and consumer-protection missions charged to the FTC. Under Section 5(n), nothing is unfair under the FTC’s UDAP authority:

…unless the act or practice causes or is likely to cause substantial injury to consumers which is not reasonably avoidable by consumers themselves and not outweighed by countervailing benefits to consumers or to competition.

By statute, that restriction applies equally to Section 18 rulemaking that is supposed to address such unfair practices. In effect, the substantive and procedural restrictions of Sections 5(n) and 18 are bundled when it comes to unfairness regulation.

Click to Cancel Gets Cancelled

Building an appropriate record, and paying attention to process more generally, are no small matters. That should be clear from the order vacating the HSR amendments and, not incidentally, the commission’s 2025 loss in the 8th U.S. Circuit Court of Appeals in Custom Communications Inc. v. FTC. There, the 8th Circuit vacated the FTC’s 2024 Negative Option Rule, or “Click-to-Cancel” Rule, holding that the commission had failed to meet the procedural requirements for rulemaking under Section 22 of the FTC Act. Specifically, the FTC had failed to conduct the preliminary regulatory analysis the act requires for Section 18 rulemaking.

The 2024 rule had been adopted on a 3-2 party-line vote, with Andrew Ferguson and Melissa Holyoak voting no. As it happens, I joined a group of former enforcers in a TechFreedom letter on the rule, but I’d especially recommend dissents by sitting commissioners—Holyoak’s dissenting statement on the rule’s adoption and Christine Wilson’s 2023 dissenting statement on the preceding NPRM. As Wilson pointedly put it:

I might have supported a tailored rule to address the negative option marketing abuses prevalent in our law enforcement experience that consolidated various legal requirements. This proposal instead attempts an end-run around the Supreme Court’s decision in AMG to confer de novo redress and civil penalty authority on the Commission for Section 5 violations unrelated to deceptive or unfair negative option practices.

Holyoak would echo and expand on Wilson’s complaints a full year and a half later: 

I respectfully dissent, for three reasons. First, this rulemaking did not follow the FTC Act’s Section 18 requirements for rulemaking because: (1) the Rule is much broader than the “area of inquiry” proposed by the advance notice of proposed rulemaking (“ANPR”); (2) the Rule fails to define with specificity acts or practices that are unfair or deceptive, improperly generalizing from narrow industry-specific complaints and evidence to the entire American economy; and (3) the Rule fails to demonstrate that the unfair or deceptive acts or practices related to negative option billing are “prevalent.” Second, the Rule’s breadth incentivizes companies to avoid negative option features that honest businesses and consumers find valuable. Third, the Rule represents a missed opportunity to make useful amendments to the preexisting negative option rule within the scope of the Commission’s authority.

In brief, neither Wilson nor Holyoak suggested, as Gertrude Stein did of 1935 Oakland, that “[t]here is no there there.” Rather, they raised both substantive and procedural objections to the Khan majority’s rush to issue overbroad regulations.

Notice a pattern? Of course—it’s not as if I’ve been subtle in citing the comments my colleagues and I submitted in response to the food-delivery fee NPRM and, not incidentally, orders of vacatur issued by the U.S. District Court for the Eastern District of Texas for the HSR notice amendments, the 8th Circuit for Click-to-Cancel, and the U.S. District Court for the Northern District of Texas for the noncompete rule.

But Wait, There’s More

Fear not—or not so much—I’ve skipped over the 5th U.S. Circuit Court of Appeals’ January 2025 order vacating the CARS Rule; the 2025 final rule on “unfair or deceptive fees,” narrowed to cover fees in live-event ticketing and short-term lodging; various regulatory activities surrounding multilevel marketing; ongoing rule review—much of it business as usual—and more.

What we’ve covered is, in any case, rather a lot of rulemaking for an agency that has traditionally considered itself more of an enforcement agency than a regulator.  

Another Modest Proposal: Stop Cutting Corners

I’m hardly the first to note that scrupulous attention to process was not exactly the Khan commission’s forte. Neither was regulatory humility. Perhaps it’s no surprise that the FTC’s more recent regulatory setbacks in federal court can be traced to those shortcomings.

Back in March 2025, I had a post called “What Changes Might, and Should, a New FTC Majority Bring?” All of it was hopeful; some of it was right; and some of it was premature, if not just wrong. This March, I was fortunate to moderate an ICLE-sponsored panel discussion with Noah Phillips—a former FTC commissioner—and Chris Mufarrige—director of the FTC’s Bureau of Consumer Protection—on “The Competition and Consumer Protection Year in Review: A Panel on Enforcement Policy at the FTC.”

There, Phillips observed several salutary changes at the FTC in its first year under new leadership. One was the suspension of general hostility to mergers, which should not be confused with a diminution of enforcement vigor. Another was attention to process. A third, on the consumer-protection side, was a renewed focus on consumer welfare and, in enforcement, on anti-fraud efforts aimed at real consumer harm. All three have application to the current regulatory program. That brings us to..

My Second Modest Regulatory Proposal: The commission should rescind the revisions to the Section 18 rulemaking process that it adopted in July 2021. These were changes to internal process rules. if reminiscent of substantive rulemaking by the agency under the Biden administration in that they were adopted on a strict party-line vote. Commissioners Christine Wilson and Noah Phillips issued a joint dissenting statement.

As I wrote at the time, their dissenting statement is instructive. Also:

I suppose I also recommend the commission’s (majority) statement on the rule change, in which it said that the changes would “modernize the way it issues Trade Regulations rules under Section 18 of the FTC Act.” Wilson and Phillips pointed out that the changes diminished both the transparency of the rulemaking process and the opportunity for independent input, and eliminated the requirement of an expert staff report (and for Bureau of Economics review of a preliminary staff report), as we noted in ICLE’s comments on the commission’s Advance Notice of Proposed Rulemaking on Commercial Surveillance and Data Security.

The majority statement provides a bit of selective history:

In 1975, Congress passed the Magnuson-Moss Warranty—Federal Trade Commission Improvement Act laying out specific procedures for the promulgation of “Trade Regulation Rules” to protect consumers in a dynamic and changing economic landscape. Indeed, the Commission rightfully responded to this grant of authority by initiating more than a dozen rulemakings in the few months and years after its passage.

Strike the word “rightfully,” and it’s true enough, if incomplete. Congress did enact Magnuson-Moss in 1975. The act’s amendments to the FTC Act granted the FTC consumer-protection rulemaking authority and set out procedures for such rulemaking—that is, what is now Section 18 of the FTC Act. And the FTC did engage, enthusiastically, in rulemaking.

But as Wilson and Phillips pointed out, Congress “imposed significant procedural obligations on the Commission to cabin its discretion” in adopting such rules. And for good reason: the rulemaking flurry in the wake of Magnuson-Moss followed hot on the heels of what was widely regarded—not least by Congress—as excessive FTC rulemaking during the 1960s and early 1970s.

Tim Muris, a former FTC chairman, and Howard Beales, former director of the Bureau of Consumer Protection, have reflected on “the disastrous decade of the 1970s.” Maureen Ohlhausen, a former FTC commissioner and acting chair, observed that “[t]he backlash against this sweeping regulatory agenda was fierce. Ultimately, even the Washington Post criticized the Commission for being a ‘National Nanny.’” And here’s Muris again, for good measure.

It is well known that the flurry of late-1970s Magnuson-Moss rulemaking cheered by Khan, Rebecca Kelly Slaughter, and Alvaro Bedoya in amending the agency’s rules of process was widely condemned. Wilson and Phillips trod no new ground in observing that “[b]acklash from the agency’s sweeping regulatory efforts culminated in the Federal Trade Commission Improvements Act of 1980, which imposed additional procedural obligations on Section 18 rulemaking efforts.” Ohlhausen recalls that Congress even “demonstrated its disapproval of the FTC’s overreach by refusing to fund the agency, causing the Commission to close its doors for a brief time.” Here’s the Washington Post’s April 30, 1980, coverage of the shutdown.

As former FTC Chairman William Kovacic testified before Congress while a commissioner, there are good reasons to be cautious in rulemaking under Section 18, given the breadth of issues that may be swept under the FTC’s UDAP authority and its sectoral range, “reaching broadly across the economy, except for specific carve-outs.” These are also reasons to prefer acting on congressional mandates that identify “specific consumer protection issues” that call for regulation.

The lack of a more focused mandate and direction from Congress, reflected in legislation with relatively narrow tailoring, could result in the FTC undertaking initiatives that ultimately arouse Congressional ire and lead to damaging legislative intervention in the FTC’s work. This is precisely what occurred toward the end of the Carter administration. Ongoing Commission initiatives led Congress to turn against the Commission in 1979 and 1980, enacting significant legislative constraints (while individual members proposed even more significant cutbacks in Commission authority).

In brief, finally, the regulatory streamlining adopted under Lina Khan’s leadership was ill-advised. And overhasty attempts to promulgate rules as workarounds to the Supreme Court’s decision in AMG Capital are liable to founder in the courts, at least. As they have.

Rescinding those revisions would be consistent with the greater focus on process seen under current FTC leadership. And it need not require anything so involved as issuing new substantive—or “legislative”—regulations. 

The post Antitrust at the Agencies: National Nanny Hangover Edition appeared first on Truth on the Market.

As I argued recently, “durable market power and demonstrable competitive harm remain elusive.” The market has not simply “tipped” toward Google, Meta, Amazon, Apple, or Microsoft. If anything, the most visible consumer-AI leader is OpenAI, which is reportedly preparing for an initial public offering.

Anthropic appears to have an edge in enterprise AI services, while Google and Microsoft benefit from distribution and infrastructure tied to their legacy businesses. Those advantages make them serious contenders, but hardly inevitable winners. On closer inspection, the AI ecosystem looks less like a market already captured by “Big Tech” than one defined by entry, rivalry, experimentation, and rapid technological change.

Despite those developments, a more pessimistic narrative continues to push for aggressive antitrust enforcement—and even direct state intervention—to “shape” AI markets. Advocates warn that AI could otherwise “supercharge ‘digital feudalism.’” That framing now runs through much of contemporary AI policy. The goal is no longer merely to police markets after anticompetitive conduct occurs, but to intervene ex ante—before the fact—to prevent a feared future of “private control” and “rent extraction.”

In his 2024 paper “The Case Against Preemptive Antitrust in the Generative Artificial Intelligence Ecosystem,” Jonathan Barnett argues that this turn toward enforcement in AI markets reflects a broader “preemptive approach” to antitrust. Under that approach, regulators presume certain practices by large technology firms are anticompetitive and place the burden on those firms to prove otherwise. Barnett’s warning is straightforward: in emerging markets, premature intervention risks suppressing “innocuous or efficient business practices” before regulators have enough evidence to assess their competitive effects.

That concern is especially acute in early-stage markets, where uncertainty is high and business practices that initially appear exclusionary may turn out to be competitively neutral—or affirmatively procompetitive. In AI markets, that includes product integration, minority investments, partnerships, licensing arrangements, and cloud-computing agreements that help firms assemble the complementary assets needed to compete.

Together with Dirk Auer, I have similarly warned that:

… overenforcement in the field of generative AI could engender the very harms that policymakers currently seek to avert. Indeed, preventing so-called “big tech” firms from competing in these markets (for example, by threatening competition intervention as soon as they build strategic relationships with AI startups) may thwart an important source of competition needed to keep today’s leading generative-AI firms in check.

This post examines three recent examples—from the European Union, Italy, and Brazil—that illustrate the common logic of this preemptive approach and suggest the warning is becoming increasingly urgent.

Emergency Remedies for Hypothetical Harms

One of the clearest examples is the European and Italian scrutiny of Meta’s integration of AI into WhatsApp.

The Italian Competition Authority (ICA) first opened an investigation in July 2025 into Meta’s decision to integrate Meta AI into WhatsApp. The ICA focused on the fact that Meta AI appeared prominently in the WhatsApp interface and was integrated into the search bar, while users allegedly had limited ability to remove or hide the feature.

The ICA later broadened the investigation to examine WhatsApp’s Business Solution Terms, which excluded Meta AI’s competitors from the platform beginning Oct. 15, 2025. Specifically, the terms barred rival AI-chatbot services from operating on WhatsApp. The ICA also opened proceedings to consider interim measures—that is, temporary remedies imposed before the underlying investigation has concluded.

The European Commission opened a parallel investigation, but moved faster and escalated further. After Meta updated its WhatsApp Business Solution Terms in October 2025 to exclude third-party general-purpose AI assistants effective Jan. 15, the Commission launched a formal Article 102 investigation into Meta’s restrictions on third-party AI assistants accessing WhatsApp. Article 102 of the Treaty on the Functioning of the European Union (TFEU) prohibits dominant firms from abusing their market position.

On Feb. 8, the Commission issued a Statement of Objections, a formal document setting out its preliminary view that Meta’s exclusion violated European Union antitrust rules. The Commission framed the conduct as an abuse of dominance in the “consumer communication applications market.”

Meta responded by restoring access for third-party AI assistants, albeit subject to a fee. The Commission remained unsatisfied. On April 14, it issued a Supplementary Statement of Objections arguing that the fee-based approach was “in effect equivalent to the previous access ban.” The Commission also notified Meta of its intention to impose interim measures requiring the company to restore third-party access under the pre-October 2025 terms, pending the outcome of the broader investigation.

Why call this preemptive? One could reasonably object that both the ICA and the Commission acted only after Meta restricted access to WhatsApp. The deeper problem is that both authorities appear to rely primarily on a theory of harm: that vertical integration—combining a platform with complementary services—creates incentives to disadvantage rivals.

But abuse-of-dominance cases require more than a plausible theory. They require evidence of dominance and evidence of harm to competition.

Even at this preliminary stage, the market definition looks implausibly narrow. As Dirk Auer has explained, consumers reach and use communications apps and AI assistants through a wide range of channels, often switching fluidly across platforms. Once those competitive dynamics are taken seriously, the relevant market likely becomes much broader, making it harder to sustain the claim that WhatsApp holds a dominant position.

The evidence of competitive harm also remains speculative. As Giuseppe Colangelo has warned:

It is uncertain that integrating Meta AI into WhatsApp would materially harm competition in AI assistants, especially given the success of rivals such as ChatGPT. ChatGPT achieved rapid adoption through cross-platform integration and partnerships enabling users to shop, book services, and perform other tasks within a single interface. By contrast, Meta AI’s market share remained minimal—about 0.2% during April 2024–March 202567 and below 1% in January 2026—and developer adoption was low.

The Commission’s choice of remedy makes the case look even more premature. Interim measures are an extraordinary tool in EU competition law. They are rarely used and traditionally require evidence of serious and irreparable harm before the main investigation concludes.

Yet the Commission seeks emergency intervention in a market where rival AI assistants—including ChatGPT, Claude, Gemini, Copilot, Perplexity, and Grok—have expanded rapidly through their own apps, browser integrations, operating-system partnerships, and enterprise channels. The Commission has not shown that exclusion from WhatsApp has actually foreclosed rivals from entering or expanding in the market.

Imposing emergency remedies based on asserted rather than demonstrated harms risks freezing the competitive process itself. It could even sideline Meta by stripping away one of the few advantages it plausibly holds in AI competition: distribution.

Brazil’s ‘Just in Case’ Merger Control

Brazil offers another example of this same preemptive impulse, this time in merger control.

In August 2024, Brazil’s competition authority, the Administrative Council for Economic Defense (CADE), opened proceedings into three major AI partnerships: Amazon/Anthropic, Microsoft/Mistral, and Google/Character AI. CADE said the goal was to “understand if these acquisitions, which would not require mandatory notifications, are to be investigated due to potential competitive harm.”

CADE’s own announcement acknowledged that opening proceedings did not necessarily mean the transactions were notifiable or raised competition concerns. The authority identified three possible outcomes: dismissal of the case, approval of the transaction, or initiation of formal merger-review proceedings to assess possible competitive harms.

Under Brazil’s merger rules, to be notifiable, one party to the transaction must exceed BRL 750 million (roughly $148 million) in turnover, while the other must exceed BRL 75 million (roughly $14.8 million). According to CADE’s March 31 technical notes for Amazon/Anthropic, Microsoft/Mistral, and Google/Character AI, the large technology firms each cleared the higher threshold. The AI startups did not.

In any mature merger regime, that should end the inquiry.

Instead, CADE’s General Superintendency referred each case to the Administrative Tribunal under the principle of in dubio pro societate. The authority argued broadly that “digital ecosystems present challenges for competition authorities” and that “the possibility of configuring associative contracts justifies, at least, the precautionary notification of certain contracts as acts of concentration.”

That reasoning stretches the principle beyond recognition. In dubio pro societate—roughly, “when in doubt, favor society”—is a controversial principle borrowed from criminal law. Whatever its proper scope there, it presupposes at least some meaningful doubt about harm. There should be conflicting evidence, or at least concrete indications of risk. In these cases, there appears to be neither.

To be sure, most merger-control systems include “call-in” powers, sometimes called “residual jurisdiction.” These mechanisms allow agencies to review transactions that fall below mandatory-notification thresholds. But such authority is meant to be exceptional, not routine.

The International Competition Network’s Recommended Practices for Merger Notification and Review Procedures emphasize precisely that point. They stress the limited nature of residual jurisdiction and note that, “[w]hen a jurisdiction maintains residual jurisdiction, it should take steps to address the desire of the parties to the transaction for certainty.”

That concern matters because merger thresholds exist for a reason: they provide legal certainty to firms contemplating transactions. Transplanting the controversial criminal-law principle of in dubio pro societate into merger control risks turning AI partnerships into inherently suspect arrangements under Brazilian competition law.

As Dirk Auer and I have argued elsewhere, the opposite is often true. Many of these partnerships may be essential mechanisms for AI challengers to scale. Developing advanced AI models is extraordinarily expensive. It requires computing power, specialized chips, engineering talent, distribution channels, and capital. Partnerships help firms assemble those complementary assets and compete against larger incumbents.

When Ex Ante Rules Age Overnight

The third example is the European Commission’s Digital Markets Act specification procedure concerning Alphabet’s data-sharing obligations under Article 6(11) of the Digital Markets Act (DMA). This is not traditional antitrust enforcement. It is explicit ex ante regulation.

The DMA imposes upfront duties on large digital “gatekeepers,” rather than waiting for a case-by-case finding of anticompetitive conduct. One could therefore argue that regulation is inherently preemptive and that this sort of intervention poses no special problem. But the case illustrates the broader concern running through this post: a pessimistic view of competition can produce flawed diagnoses and premature interventions in regulation as well as antitrust.

Even regulatory intervention requires threshold conditions. It is far from clear that this is an appropriate case for applying the DMA.

According to the Commission, Article 6(11) requires Alphabet to provide certain Google Search data to third-party online search engines on fair, reasonable, and non-discriminatory (FRAND) terms. In practice, FRAND rules are supposed to ensure access on terms that are not exclusionary, discriminatory, or opportunistically expensive. The Commission’s proposed specification would define how that obligation operates, including rules governing data access, anonymization, pricing, and eligibility.

Critically, the Commission’s preliminary measures would extend eligibility to “AI chatbots with online search engine functionalities,” even when search is merely one feature within a broader service.

My skepticism does not stem solely from the fact that Google would be compelled to share data. Data can represent a legitimate competitive advantage, although mandated access may be justified in some circumstances. The deeper concern is that data-sharing obligations can become tools for engineering competitive outcomes, rather than preserving the competitive process.

As Geoffrey Manne, Dirk Auer, and I argued in the International Center for Law & Economics’ (ICLE) recent submission to the consultation, the proposed measures “risk shifting Article 6(11) from a data-access obligation to a tool for delivering competitor success.” The effectiveness of the measures should be understood as creating opportunities to compete, not guaranteeing market-share gains for particular competitors.

More fundamentally, including AI chatbots within the scope of these obligations presupposes that Google can leverage dominance in search into dominance in AI-chatbot markets. That assumption looks increasingly detached from market realities.

The evidence points in the opposite direction: AI tools increasingly threaten Google’s position in search, rather than the reverse.

Google’s global search share fell below 90% for the first time since 2015 during the final quarter of 2024, a decline widely attributed to users shifting toward AI-native alternatives. Early in 2024, Gartner projected that traditional search-engine volume would decline 25% by 2026 because of AI chatbots—a forecast that, so far, appears remarkably prescient.

Meanwhile, Perplexity, which operates without privileged access to Google Search data, processed 780 million queries in May 2025, up 239% year over year, and reached a $20 billion valuation. ChatGPT now reportedly handles roughly 12% of Google’s daily search volume and has reached 900 million weekly users—again, without access to Google’s search data.

Perhaps most tellingly, during the remedies phase of United States v. Google, Apple executive Eddy Cue testified under oath that Google search volume on Safari had declined for the first time in more than two decades. Cue attributed the shift directly to users turning to AI tools for information discovery.

The U.S. court that conducted the most rigorous evidentiary review of Google’s search position reached much the same conclusion. Judge Amit Mehta found that “tens of millions of people use GenAI chatbots, like ChatGPT, Perplexity, and Claude, to gather information that they previously sought through internet search,” and that generative AI represents “a nascent competitive threat” to Google’s dominance in search.

That assessment mattered. It informed Mehta’s decision not to impose structural remedies on Google, with the court noting that competition in AI “is plentiful.” After a full adversarial proceeding examining precisely these market dynamics, the court declined to treat Google’s search dominance as the key to AI competition. If anything, the evidence suggested the opposite: AI competition is emerging as a threat to Google.

The Commission’s specification procedure moves in the other direction. It extends search-data-sharing obligations into a market where Google appears to be under competitive pressure, not successfully leveraging dominance.

That is the broader problem with ex ante AI regulation—and, more generally, with regulatory regimes like the DMA. They can quickly become moving targets. The DMA was designed to avoid the perceived slowness of traditional antitrust enforcement. But AI markets expose the limits of that ambition. When technology and consumer behavior evolve rapidly, ex ante rules risk hardening assumptions that may already be obsolete.