According to Regime Change, the new Maggie Haberman and Jonathan Swan account of Trump’s second term, Blanche had a vivid forecast for what would happen if the DOJ stumbled into federal court to push a series of revenge indictments. Roger Parloff of Lawfare surfaced the passage:

Todd Blanche confided to colleagues that he was concerned about appointing someone a judge might regard as an illegitimate appointee to bring these high-profile cases. After the sugar high of “indicted” headlines wore off, he explained, they would be left with cases that were almost certainly going to be dismissed, and they would spend the next six months “getting skullfucked” by judges — Blanche’s colorful shorthand for long and detailed judicial objections.

Check out MAGA Cassandra over here.

Lindsey Halligan — a former insurance lawyer and Trump favorite — got installed as the not-really-genuine-US Attorney for the Eastern District of Virginia to pursue indictments against James Comey and Letitia James only to see both cases fail miserably. The skullfucking arrived precisely on schedule.

Haberman and Swan describe it as, “Blanche’s colorful shorthand for long and detailed judicial objections,” but cinephiles know it as one of Full Metal Jacket‘s most entertaining rants:

Which is the strange, almost encouraging part. As of a few months ago, Blanche apparently worried about losing court cases. Worrying about losing implies Blanche harbored a flickering residual commitment to honor court orders. That’s significant when he served in this administration with Emil Bove, who reportedly instructed DOJ lawyers to tell federal judges “fuck you” if the courts ruled against the administration. Now Bove is a federal judge… because we live in hell.

More from Haberman and Swan’s book:

But Trump was losing patience. On the morning of September 24, when the President arrived in the Oval Office, he declared to an aide that he thought he needed to fire both Bondi and Blanche. “They don’t do anything,” he said angrily, pointing out yet again that prosecutors from the other side had never seemed to have any trouble going after him.

Here’s where that encouraging feeling withers. Because Bondi is gone and Blanche is not only still around, but auditioning for the top job. Not to read too much into it, but that suggests Blanche has convinced Trump he’ll do his job as Trump’s personal lawyer and embrace whatever skullfucking may follow.

Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news.

The post Todd Blanche Feared Judges Giving It To DOJ In Extremely Uncomfortable Place appeared first on Above the Law.

One of the things I’ve noticed over the years is that every major technology shift eventually develops its own recognizable contract.

It doesn’t happen overnight. In the early days, every agreement feels different because vendors are still figuring out how to describe the technology and customers are still learning what questions to ask. The contracts look highly customized, negotiations take longer, and every deal feels like uncharted territory.

Eventually, though, something changes.

The market starts asking the same questions over and over again. Lawyers begin negotiating the same issues in nearly every transaction. What once seemed unique gradually becomes familiar, and before long a recognizable contractual framework begins to emerge.

We saw it happen with cloud computing. We saw it happen with SaaS. We saw it happen with open source software.

I think we’re beginning to see the same thing happen with artificial intelligence.

The Language Is Different, But The Questions Are The Same

At first glance, enterprise AI agreements still appear to be all over the map.

Every major vendor has its own terminology, its own structure, and its own way of describing AI functionality. If you compare contracts side by side, they rarely look identical, which makes it easy to conclude that the market is still too immature for any real standardization.

But after reviewing enough of these agreements, I’ve started to notice something else.

The wording changes from contract to contract, but the underlying questions are remarkably consistent.

Customers want to know whether their data will be used to train future models. They ask who owns prompts, inputs, and AI-generated outputs. They want assurances about confidentiality, security, intellectual property protection, transparency, and human oversight. They ask how vendors address hallucinations, inaccurate outputs, model updates, and regulatory changes. They also want to understand where responsibility shifts from the vendor to the customer when AI is put into production.

Those conversations are happening across the market, regardless of which AI platform is involved.

That tells me something important.

The Market Is Beginning To Build A Framework

I don’t think we’re moving toward a standard form agreement. Technology companies will always differentiate themselves through their contracts, and legal teams will continue negotiating language that reflects their own risk tolerance and business models.

What I do think is emerging is something more subtle.

The architecture is becoming familiar.

Most enterprise AI agreements now revolve around the same core subjects. They address data use, model training restrictions, ownership of AI-generated content, confidentiality, acceptable use, security, transparency, governance, audit rights, intellectual property, and liability. Individual clauses may differ, but the overall structure increasingly feels recognizable.

That is often how markets mature.

Long before contract language becomes standardized, the issues themselves become standardized. Lawyers know where to look because they already know the questions that need to be answered.

Contracts Tell Us Where the Market Is Going

One reason I find this interesting is that contracts often reveal where a market is headed before legislation does.

Every negotiated agreement represents a practical attempt to solve real business problems. Customers are asking for commitments they believe they need. Vendors are deciding which promises they are willing to make. Over time, those negotiations begin shaping expectations across the industry.

That’s exactly what happened with cloud computing.

Eventually, cloud agreements settled into a familiar rhythm. Customers expected uptime commitments, service level agreements, disaster recovery provisions, data security obligations, and clear allocation of responsibility. Those concepts became so common that they now feel like ordinary commercial terms.

AI appears to be following the same path.

The specific language will continue evolving, particularly as regulators introduce new requirements and the technology itself changes. But the overall framework is becoming easier to recognize with every negotiation.

Why This Matters For Lawyers

For legal departments, this changes the nature of AI contract review.

A few years ago, reviewing an AI agreement often felt like exploring completely new territory. Every deal raised unfamiliar questions, and there were very few market norms to rely upon. Today, many of those questions have become recurring themes, even if the answers remain heavily negotiated.

That doesn’t make the lawyer’s job easier.

If anything, it raises the stakes. Legal teams are no longer reviewing isolated clauses. They’re evaluating an interconnected governance framework that affects privacy, security, intellectual property, procurement, compliance, and the day-to-day operation of AI inside the business.

I’ve also noticed that these conversations are becoming much more collaborative. Product teams, security professionals, procurement, privacy counsel, compliance, and legal all have legitimate interests in the same contractual provisions because those provisions increasingly determine how AI can actually be deployed after the agreement is signed.

In many ways, the contract has become the blueprint for enterprise AI governance.

A Sign That The Market Is Growing Up

We often measure the maturity of a technology by looking at the products themselves. Are the models getting better? Are customers adopting them? Are businesses finding meaningful use cases?

Those are important indicators.

But I think there’s another measure that’s easier to overlook.

Markets mature when their contracts become recognizable.

That doesn’t mean every agreement looks the same. It means experienced lawyers begin to recognize the architecture because the same issues appear in deal after deal. The negotiations become more sophisticated, expectations become clearer, and both vendors and customers develop a shared understanding of what responsible contracting looks like.

Artificial intelligence seems to be entering that stage now.

The headlines still focus on the latest model, the latest funding round, or the latest regulatory proposal. Meanwhile, commercial lawyers are quietly building something just as important. Through thousands of negotiations taking place every day, they’re creating the contractual framework that will shape how enterprise AI is bought, sold, and governed for years to come.

That may not generate the same headlines as the latest AI breakthrough.

But it is one of the clearest signs that the market is beginning to grow up.

Olga V. Mack is the CEO of TermScout, where she builds legal systems that make contracts faster to understand, easier to operate, and more trustworthy in real business conditions. Her work focuses on how legal rules allocate power, manage risk, and shape decisions under uncertainty. A serial CEO and former General Counsel, Olga previously led a legal technology company through acquisition by LexisNexis. She teaches at Berkeley Law and is a Fellow at CodeX, the Stanford Center for Legal Informatics. She has authored several books on legal innovation and technology, delivered six TEDx talks, and her insights regularly appear in Forbes, Bloomberg Law, VentureBeat, TechCrunch, and Above the Law. Her work treats law as essential infrastructure, designed for how organizations actually operate.

The post Enterprise AI Is Quietly Creating Its Own Standard Contract appeared first on Above the Law.

A new report from Casepoint titled “From AI Hype to AI Accountability: What You Can Learn From Legal and FOIA Teams About AI Modernization,” built on practitioner interviews with legal, records, and FOIA teams, covers a broad array of AI topics. One that jumped out is the admonition that lawyers need to understand that this new technology may excite the legal community, but when it comes to eDiscovery, the old ways are still the best.

“Privilege review is one example,” the report explains. “The commercial appeal is obvious: privilege review is expensive, time-consuming, and often pattern-driven.” The problem is that mistakes can be devastating and users all too often trust AI’s confident, mansplaining-as-a-service output without performing needed diligence. “Al may still have a role in privilege review, responsiveness calls, redactions, FOIA exemptions, legal advice, and production decisions,” the report continues, but teams “should know how output will be checked, what human remains accountable, what documentation will be retained, and whether the process can be explained later.”

Conveniently, we already have these processes for existing technologies:

GenAl also lacks the mature validation history that helped technology-assisted review gain acceptance. TAR became defensible because practitioners developed ways to explain and test the process, including recall, precision, seed sets, sampling, and quality-control protocols. GenAl review has not yet reached the same level of accepted process maturity.

As a user at a large federal civilian agency explained, “There’s no point using AI in the manner I described if we get to a production, go to court, and they say: this is not verifiable, this is not TAR 2.0, and I don’t have any of the statistics and measures I need.”

We’ve built technology to perform these tasks that doesn’t have the hankering to take black box flights of hallucinatory fancy and spent years making sure it had the reliability and accountability to hold up in court. The Casepoint interviewees describe hallucinations not as a bug to be patched but as a permanent “workflow reality.” TAR had plenty of flaws. Confidently inventing a case that does not exist was not among them. Vendors are building generative AI into their systems to handle what it’s suited to handle, and metric tons of digital ink have been spilled describing the efforts legal tech providers have made to build out genuine audit processes, but part of adopting AI is understanding its limits.

It’s been 14 years since Magistrate Judge Andrew Peck issued the first judicial opinion approving predictive coding, in Da Silva Moore v. Publicis Groupe. The crux of the case wasn’t “computers are good,” but that defensibility lives in the process. TAR didn’t become boringly trustworthy because the technology was magic. We just poured years into forging a system we could test, measure, and explain to a judge. Part of selling the courts on the rules was using (for the most part) rules-based technology.

But the selling point of the “agentic” era is that it comes up with its own rules to match the goal. That doesn’t mean it’s always wrong — the rules it comes up with may well be defensible — but it’s a different task to explain that to a judge.

Though so far, the courts don’t seem to think so. In Schulte v. LinkedIn, decided in the Northern District of California this June, Judge Eumi Lee declined to invent a special legal framework for generative AI in discovery and simply analyzed it under the principles courts already apply to TAR. At a high enough level this is true — it has to consistently display recall, precision, and a testable process. But it also hamstrings the technology a bit. It’s not quite like demanding that cars be built as mechanical horses, but some of the old processes just won’t get the most out of the new technologies.

The practical advice is simple: do not treat Al governance as policy language alone. Build it into the workflow. Legal teams need to define what Al may do, what information it may access, what records it creates, how output is reviewed, and how decisions are documented. Without that structure, organizations may not discover the accountability gap until a matter, FOIA request, production, court challenge, or oversight review forces the issue.

None of this means generative AI won’t take on more and more discovery tasks or that courts won’t eventually develop rules optimized to the technology… but it also means everyone needs to be a little honest about what it’s not capable of doing.

Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news.

The post Lawyers Learning You Can’t Square Peg AI Into eDiscovery Round Holes appeared first on Above the Law.

[It was] pretty, pretty, pretty good. The obvious highlight was getting to see firsthand how true experts in their field approach comedy and storytelling.

Quite frankly, I thought I was simply going to be an extra. Fortunately, there were actual professionals on set to make sure I didn’t embarrass myself too badly, and it also turns out that good directors can cover up a lot.

— Andrew Clubok, the global chair of Latham & Watkins’ securities and mergers and acquisitions litigation practice, in comments given to the ABA Journal, concerning his recent appearance on client Larry David’s HBO series, Life, Larry and the Pursuit of Unhappiness: An Almost History of America. Clubok says he’s “very happy being a lawyer” and has no plans to act again. “But after this experience, I’ve learned never to say never,” he said.

Staci Zaretsky is the managing editor of Above the Law, where she’s worked since 2011. She’d love to hear from you, so please feel free to email her with any tips, questions, comments, or critiques. You can follow her on Bluesky, X/Twitter, and Threads, or connect with her on LinkedIn.

The post This Biglaw Partner Got A Crash Course In Comedy From Larry David appeared first on Above the Law.

An LLM is a brain, and almost every legal AI tool on the market today is built on top of a general model trained on the open web. AI has a long memory and an earnest desire to please. The moment it can’t find an answer in the material you’ve provided, the model performs exactly as trained, reaching back through everything else it knows in effort to help. The open web is full of material not rooted in fact, or in other words made up, and AI doesn’t know the difference.  

Context is always underestimated  

Most of what you can pull for free online is primary source law. Legislation. Rules. Official missives from agencies like the IRS. Primary source law is an essential component of effective legal AI, but taken alone it’s still like trying to assemble a 1,000 piece puzzle without being able to look at the image on the box for context. Secondary source material — the treatises, the analysis, the practitioner guidance that explains how a rule actually works, why it’s important and what it relates to — is that bigger picture.  

My go-to example is elections. Search “elections” in almost any state and you’ll get a mountain of provisions, and most of them have nothing to do about the rule that prompted your original search. Absent the larger context provided by authoritative secondary source content, a large language model has nothing to which it can anchor. The result is an answer unmoored from any verifiable body of information, such as a table of contents, drifting closer and closer towards potential hallucination.    

Guardrails don’t stay put  

A new round of testing should accompany every model update and feature release. Each iteration of an LLM reasons differently, and a guardrail that previously kept the system anchored to your vetted content may now have a loophole you never anticipated.  

Ask the system to build a contract and lay out the appropriate clauses. You can watch to see if the eager-to-please model will out-think your guardrails in an effort to deliver an answer.  

An automated test will more than likely pass a fault output so long as the citations provided appear genuine. A human reviewer needs to physically click the citation and verify that it does not answer the question in order to catch the problem.

Saying “I don’t know” is harder than it sounds  

Nobody likes admitting they don’t have all the answers. Teaching a model to say “I don’t know” is both a matter of training and a statement of purpose. While it’s true that a tool which sometimes comes back with “I don’t have a source for this” may seem less impactful in a sales demo, practicing attorneys will actually recognize it as solid ground.  

When I practiced law, my greatest fear was missing something critical. If there’s no reputable authority that can be cited, that would have been genuinely useful for me to know. It empowers me to begin taking steps in the right direction. On the other hand, a paragraph buoyed by false confidence and built from dubious sourcing takes much longer to untangle.  

Remember that transparency is a vital component of trust. Accuracy isn’t just about pinpointing the correct answer. It’s also about being able to acknowledge when there is no answer to be found. Establishing that level of visibility as table stakes is essential to building a strong future for legal AI. 

Nicole Stone is Director of AI & Agentic Solutions Product Management at Wolters Kluwer Legal & Regulatory U.S., where she leads product strategy and development for digital legal content and technology solutions. With over 22 years of experience in legal technology and a background as a practicing attorney, she focuses on integrating emerging technologies, including generative AI, into products that serve legal professionals.

The post The One Thing You Didn’t Think To Look For In Legal AI  appeared first on Above the Law.

The man best known for authoring vaguely legal-sounding memos turning the U.S. into a torture state went on Fox News yesterday to muse that New York Mayor Zohran Mamdani should be investigated by the Trump Justice Department. Yoo explained that he was “not trying to exaggerate” as he effortlessly exaggerated potential criminality based on Mamdani asking his own city lawyers if New York can do anything about the International Criminal Court’s arrest warrant for Benjamin Netanyahu when the Israeli prime minister comes to town for the U.N. General Assembly this fall.

Here’s Yoo, making the pitch to anchor Gillian Turner:

The mayor is violating the American Servicemen Protection Act, passed in 2002, and the Justice Department should be investigating him. And the president, under that law, is authorized to use all necessary means to prevent any American or ally from being handed over to that court. So if Mayor Mamdani wants to spend a lot of time dealing with FBI agents and Justice Department investigators, well, he just started that.

Of note, it’s the American Service-Members’ Protection Act — hyphen and all — so Yoo can’t even correctly identify the one statute he’s pegged his baseless legal threat upon. An extra rough look given that, moments earlier, he’d bragged to everyone that he helped pass it back when he was gleefully playing legal Mad Libs to build a case for presidents having unaccountable authority to crush prison testicles… for freedom!

And it is fair to call this a threat, because Yoo has taken on a new role in the administration gray zone, helping Joe DiGenova brainstorm criminal charges that the DOJ can bring against Donald Trump’s critics. It’s clear to everyone paying attention that Yoo has the ear of the DOJ, even from his current role — which has all the legal clarity of the CIA black sites Yoo believes presidents can whisk people off to.

Despite not even knowing the name of the law he’s talking about, Yoo isn’t entirely wrong.

The ASPA — which, for all the talk about its proper title, would much more accurately be called The War Crimes Protection Act — passed in 2002 to shield Americans from war crime prosecutions. Because, in 2002, the W. Bush administration really wanted to do some war criming. The statute does unambiguously bar federal, state, and local officials from cooperating with the International Criminal Court. The law is also sometimes called the “Hague Invasion Act,” because it ostensibly authorizes the president to use the military to spring any American or ally out of a Dutch jail cell. While incredibly stupid and probably a violation of our NATO obligations, the law exists.

Though we really don’t have to reach this law since Netanyahu, as a foreign head of government attending a UN meeting, would carry diplomatic immunity anyway.

In either event, Mamdani arresting Netanyahu would be illegal. But the thing is… Mamdani has not arrested Netanyahu.

The ASPA does not really lend itself to penalizing inchoate conduct. It’s a leash on government officials. And even if it somehow could support a conspiracy charge, “I asked my lawyers about my legal options” is not a predicate act in any theory of conspiracy or attempt an American court has ever recognized. Asking whether a thing is legal is the opposite of doing an illegal thing. In fact, if checking with lawyers counted as a predicate act to support a conspiracy, the incentive would be to never talk to lawyers, which is exactly what the law hopes to discourage.

Theoretically, Yoo understands that this is utter nonsense. But he’s a partisan thug whose current job is ginning up ways to use the Justice Department to exact retribution against Trump’s enemies, and he knows how to chill an administration critic.

It’s telling that Yoo declares, “if Mayor Mamdani wants to spend a lot of time dealing with FBI agents and Justice Department investigators, well, he’s just launched that process,” in the same register as “nice little city you got here, shame if anything happened to it.” And he’s sending the signal that this process is “launched” by merely pointing out to the public that Netanyahu is wanted by The Hague for war crimes. And, for what it’s worth, Yoo thinks the administration should be entirely unaccountable, adding an ominous twist to this strategy.

It’s not limited to Mamdani, because Yoo is telling everyone that if you mouth off about Trump and his friends, FBI agents will bang on your door, waste several days of your life, and leave you in limbo worried about charges that will almost certainly never come because they would be both frivolous and unconstitutional.

In the spirit of “not trying to exaggerate,” siccing the FBI on a guy for making a phone call to his own lawyers is Third World shit. Then again, so is torture, and we know that gets Yoo all lathered up too.

Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news.

The post John Yoo Wants DOJ To Investigate Mamdani For Crime He Literally Did Not Commit appeared first on Above the Law.

Enter CVN

Occasionally, when I go to legal conferences such as the one put on this week by AALL, I stumble upon a concept, an idea, or a product that actually moves the needle. Such was the case when I attended a talk by representatives of something called the Courtroom View Network (CVN).  Instead of relying on AI to solve any and all problems, CVN found a practical solution to the problem of a lack of trial skills training.

I got the sense that these guys have been around a while, but as its CEO, Lou Freitas, said at the session, they only recently began marketing their product as a trial training tool. It capitalizes on the use of video to teach lawyers how to try cases, much like you go to YouTube to watch videos on how to repair or install things.

What they have done is videoed literally thousands of state court trials in their entirety. Rather than essentially creating a video report of the trial and its highlights, CVN instead videos the complete trial without any commentary or analysis. As Arlin Crisco, the Sections Editor of CVN, put it, CVN is like the C-SPAN of trials.

So What?

So what, you say: what’s the good of a bunch of videos that no one has the time to watch? You can’t really have a lawyer spend chunks of valuable time watching a three-day trial all the way through. Why not just rely on AI and AI-driven simulations to promote the training of trial skills?

But here’s the beauty of what these guys have recently done: they recognized that if these trials, or portions of them, could be easily accessed and watched, they could serve as a valuable training tool at a time when there are so few trials and even opportunities to observe trials. So, they created a way to index and then access these trials to do just that.

Their search tools allow you to search for trials by type of case, by attorney, and by industry. Perhaps more importantly, you can search the library of trial videos by the function you want to observe such as opening or closing statements, cross-examination, direct examination, etc. They also created a playlist of subjects that are easily accessible.

They even have interviews of some of the lawyers who are in the videos, describing not only what they did but why and how they did it. Indeed, using their tools, you can slice and dice these trials any way you want. It essentially creates a best-in-practice library for inexperienced and, for that matter, experienced lawyers.

Real Life, Not Simulations

The concept is based on the notion that one of the best ways to learn how to try a case, short of actually trying a bunch of them, is to observe trials or at least parts of them. By seeing and hearing real lawyers in the courtroom and noting how they do things, lawyers can see what really happens and get a real sense of how things are done. How problems and roadblocks are overcome. How the lawyer moves around the courtroom and uses their voice and inflection to better persuade. These are things, as Crisco put it, you can’t learn these by reading a book.

I remember as a young lawyer, one of my biggest fears was how to practically do many of the little things. Like how to get documents properly admitted. Or how and when to make an objection. Or how to proffer proof. It would have been wonderful, not to mention comforting, to be able to watch a video of a seasoned trial lawyer doing these very things. One less worry when you’re trying your first case or even your tenth.  And more an opportunity to focus on how to persuade instead of worrying about being tripped up.

A Persuasive Example

And make no mistake, these videos are chock full of great and memorable techniques to persuade. Crisco actually gave a demonstration of how a simple video could teach powerful persuasive techniques. He showed a clip of an actual opening statement by a good plaintiff’s lawyer in a medical malpractice trial.

The plaintiff’s claim was based upon the failure of the doctor to send a biopsy of a cyst to a lab for analysis after extraction. The lawyer used a red piece of heavy clay to demonstrate the cyst. He removed the clay from a model of the plaintiff’s face and then walked over to the metal trash can and said, “Instead of sending it to the lab, the doctor just threw it away.”

He then dropped a piece of heavy red clay into the trash can where it landed with a resounding thud. I will remember that trash can moment. Which is exactly what the lawyer wanted the jury to do.

Start With the Problem, Not AI

A few weeks ago, Melissa Rogozinski and I wrote an article about the perils of letting AI take over and drive what we do instead of thinking through the ways to solve a problem. And using AI only when it actually enhances the solution. We used the Cite Checker tool created by Clearbrief as a prime example. That tool allows for the quick check of citations generated by an AI tool without using AI itself. The CVN tool is in the same class. It uses AI as a tool to access real life.

Why is this so valuable in the age of training by simulation? An AI simulator, as good as it may be (and there are some good ones), is still a simulation. I can tell you from experience that what happens in the courtroom is not predictable. It’s nuanced. What the lawyer does depends so much on what happens right in front of him. That’s why what persuades depends not only on technique but also on knowing and observing the audience, seeing them and picking up signals. A trial can’t be fully simulated.

Moreover, a simulation is still a simulation. It’s not real life, and those that go through and use the simulation tools know that. It’s just not the same as watching an actual trial lawyer in action and seeing what they do.

And finally, there’s another reason being able to access the real thing and see what a trial lawyer does is important. In the age of AI, where there is a paucity of effective ways to learn how to do things in the courtroom, many may turn to a chatbot, an LLM, to ask how to do things – one that a) may give wrong information and b) may not be practical advice.

The Next Best Thing

This is why the very best trial lawyers are often those who try a lot of cases, who become comfortable in the courtroom, who can read the courtroom and command it. This on-the-job training can’t be easily replicated when there are so few opportunities to try cases. So, the next best thing is being able to observe trial lawyers in action and what they do. Actually seeing a trial and hearing it is the next best thing to experiencing it.

It’s also a great example of using AI appropriately and in its proper place and combining it with non-AI tools. I can’t vouch for how good the CVN product is, but I can vouch for how good of an idea it is.

The ability and confidence to try cases is not a forgotten skill at a time when so few cases are tried. One thing I learned over the years is that the best way to get a case settled is to treat it from the beginning as a case that will go to trial. To know and act like you have that ability. What CVN has done, if it works as it claims, is to give lawyers that ability and confidence. And they didn’t start with AI to do that.  

Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes TechLaw Crossroads, a blog devoted to the examination of the tension between technology, the law, and the practice of law.

The post Courtroom View Network: Learning Trial Skills By Watching Real Lawyers In Real Trials appeared first on Above the Law.

The Magic Circle firm reported revenue of £2.5 billion ($3.3 billion) for fiscal 2026, a 6.8% increase over the previous year. Pre-tax profit climbed 11.6% to £1.2 billion ($1.6 billion), while profit per equity partner rose 11.4% to £2.5 million ($3.34 million).

The firm’s strongest performances came from the very markets it’s been working hardest to expand.

Profits in the U.S. jumped 39%, while Asia posted 36% growth, easily outpacing Europe (10%) and the UK (7%). For firmwide managing partner Paul Lewis, there was no question which region stole the show. As he explained to Bloomberg Law:

“Asia has been the story of the year for me. Our Asia performance, after a number of years where the entire region’s market was a bit slower, came roaring back and we were strong across the board.”

The U.S. remains another major focus. Lewis said the firm is continuing to build its brand in the American market as it expands its cross-border practice. “We are building a brand in the US rather than having a brand in the US,” Lewis said. “And as things currently stand, we know that it’s building on trajectory.”

That strategy has included lateral hiring, with the firm continuing to add prominent lawyers to strengthen its U.S. platform. One thing Lewis says isn’t part of the plan, however, is a merger with an American firm.

His comments come just days after a Semafor report claimed Linklaters unsuccessfully approached Cleary Gottlieb about a merger earlier this year — a report both firms publicly denied. Lewis flatly rejected the idea: “We have not approached anyone for a merger. We’re happy pursuing the strategy we’re pursuing.”

If these financial results are any indication, Linklaters doesn’t see a merger as necessary. Its strategy of growing its U.S. practice, capitalizing on cross-border work, and taking advantage of Asia’s rebound appears to be doing exactly what the firm hoped it would.

Linklaters Revenue Creeps Up as US, Asia Return Soaring Profits [Bloomberg Law]

Earlier: Another Rumored Biglaw Merger Falls Apart Before It Begins

Staci Zaretsky is the managing editor of Above the Law, where she’s worked since 2011. She’d love to hear from you, so please feel free to email her with any tips, questions, comments, or critiques. You can follow her on Bluesky, X/Twitter, and Threads, or connect with her on LinkedIn.

The post No Merger Needed: Linklaters Posts Strong Financial Results appeared first on Above the Law.

Auth. note: Though it was certainly tempting, no artificial intelligence technology was used to create this publication. All appearing em dashes were added by the author — just as nature intended. 

Welcome to the sequel to the Legal Tech-to-English Dictionary (2022). This edition updates a vestige of the pre-AI world to share all of the AI and practice management terminology you need to be funner at parties. 

Here, we’re pleased to share Chapter 5. Feel free to download the full eBook using the form at the bottom of this post.

Pieces of Flair – Working With AI in a Professional Environment

While the modes of working with artificial intelligence in a professional environment might seem still to be very much like the Wild West, smart law firm operators are acquiring competent governance over AI usage and output.  

In this series of definitions, we’ll explore both what it looks like for AI to run unchecked in a business, and also how managers can control how this new technology is optimized for their teams and clients.  

Of course, it’s not like AI itself is running unchecked in most businesses; it’s more like people are messing everything up — like always.  

Just as all politics is local, all business management issues are people problems. The best news is that controlled usage of artificial intelligence in a law firm can be a boon for everyone involved.

AI slop n. 1. Demonstrably fake, low-quality content, generated by AI, and often used for engagement farming. (*Merriam Webster’s ‘Word of the Year’ for 2025. Hey, a real dictionary!*)

Etymology

The term ‘slop’ originated around the year 1400, essentially being defined as a slimy mud puddle, and effectively emanating from a word for cow dung. Over time, the meaning shifted to describe food waste. Before its application to AI, slop had been used to describe cheaply made media content without artistic value — like ragebait or reality TV or the ‘Sharknado’ movies.

Examples

Shrimp Jesus, Gordon Ramsay’s infinite spaghetti, the Service Kangaroo, Dinosaur at the Window

retrieval augmented generation (RAG) n. 1. the accessing of additional, specialized datasets by LLMs to buttress output results; often used in specialized industries, like the legal sector. 

Synonyms

flesh out, enrich, enhance, augment, supplement, refine, synthesize, expound, reinforce

Related Words

chunking, dense retrieval, sparse retrieval, re-ranking, grounding, retrieval-augmented language model, agentic RAG

semantic search n. 1. information surfacing technique that utilizes the context around the seeker’s intent, including locations, prior attempts, and relationships between words used, all to deliver optimized results.

Related Words

natural language processing, synonyms, conceptual clusters, intent modifiers, entity relations, query analysis, vector search, knowledge integration

Antonym

lexical (keyword) search

Example Sentence

Celestine: Look at this, I just tried this search in Google, and it considered stores near me, based on my location.

Agripinna: Oh, that’s just semantics.

Celestine: Exactly!

AI governance n. 1. the full collection of documentation (including policies, procedures, programs, and materials) that guide the appropriate use of artificial intelligence within a business.

Related Words

AI usage policy, written information security program, policies and procedures manual, compliance, regulation, maintenance, vetting

Antonyms

turn a blind eye, denial isn’t just a river in Egypt, sweep under the rug, look the other way, ostrichism, overlook, reject, scorn

Example Sentence

Manuel: I don’t have an AI usage policy for my law firm.

Manuela: Then, everybody’s using it.

Manuela: You’ve got a big shadow AI problem . . .

shadow AI n. 1. unauthorized use of artificial intelligence within a business organization

Related Words

shadow IT, shadow IoT, shadow data, shadow processes, shadow boxing, Storm Shadow

Antonyms

governance, control, controls, authorization, oversight, management, modeling, security

Example Sentence

Alanna: As I’m sure you know, we can’t just input our firm data into any old free AI product.

Jameson: Oh, for sure.

Jameson: . . . .

Jameson: Like, I mean, I would never enter our best client’s most sensitive data into Grok to make a word cloud with unicorn illustrations.

Jameson: Like, ever.

Alanna: Are you kidding me right now?

CosmoLex is proud to sponsor this edition of the Legal Tech-to-English Dictionary. As an end-to-end practice management platform built specifically for small to midsize law firms, CosmoLex integrates AI directly into the workflows attorneys rely on every day — summarizing documents, filtering matters and invoices in plain language, automating intake, and streamlining firm workflows. No separate tool to learn, no AI layer dropped onto a legacy system. Practical functionality built into the platform where the work already happens, with the compliance guardrails law firms actually need.

Download the Legal Tech-to-English Dictionary — Version 2.0

Jared Correia, a consultant and legal technology expert, is the host of “Adventures in Legal Tech,” the featured podcast of Above the Law’s Legal Tech Center.

The post The Legal Tech-To-English Dictionary (2.0): Working With AI In A Professional Environment appeared first on Above the Law.

The proposal, delivered last week in a dissent from the Ninth Circuit’s denial of rehearing en banc in Knife Rights, Inc. v. Bonta, is that the Supreme Court should start summarily reversing his own court’s Second Amendment decisions on sight. Not because a summary reversal is doctrinally warranted in any particular case, but because his colleagues would find it humiliating, and humiliation is the only currency VanDyke believes they still accept.

“To put it more colloquially, it’s time for some benchslaps,” he writes. “Nothing less will give this court any pause before ultimately blessing every arms restriction it reviews.”

Nothing less indeed.

Some context on the man making the ask. This is the judge who, back in March, opened a dissent in a trans discrimination case with the sentence “This is a case about swinging dicks,” then spent the rest of it calling his colleagues “woke judges” who were “complicit” in harming women and girls.

And those colleagues responded. Judge McKeown, writing for herself and 26 colleagues including the Chief Judge, filed a statement observing that the American legal system has long been a place to resolve disputes with some dignity and is “not a place for vulgar barroom talk,” and that language like his “makes us sound like juveniles, not judges.” Judge Owens and one other colleague filed twelve words of their own: “Regarding the dissenting opinion of Judge VanDyke: We are better than this.”

And if you’re wondering if that, ahem, benchslap landed, well, it was received, catalogued, and filed away for later use. VanDyke cites the McKeown statement twice in Knife Rights.

Here’s the second one, teeing up the benchslap pitch:

I’ve been accused of using “vulgar barroom talk” to get my point across, so let me assuage my colleagues’ selective Victorian predilections up front. Calm down. In making my recommendation today, I’ve chosen a word that, while evocative, would never be the punchline to a coarse joke at your local bar, and even has an official entry in the lawyer’s dictionary of choice.

VanDyke is the epitome of Walter Sobchak in the bowling alley, waving a firearm around and insisting he’s the calmer man in the room. Complete with a citation to Black’s Law Dictionary, offered as proof that this time he’s brought a nice clean word to the party.

The word benchslap, as longtime readers know, was coined by Above the Law’s founder, David Lat, in 2004 and got into Black’s (notably over objections that it was too vulgar for a dictionary). It is a perfectly appropriate term for a legal gossip blog, but that doesn’t mean it earns the greenlight for judicial opinions where decorum and professionalism are supposed to reign.

Here’s the thing: VanDyke’s childish rebuke is completely superfluous! Judge Tung already wrote a seventeen page dissent — joined by, inter alia, VanDyke. Agree with it or don’t; it’s an argument, pitched in the register arguments get pitched in, by a bench full of conservatives who managed to get through the exercise without a single hissy hit. The 27 pages VanDyke wrote add nothing to the legal argument — the separate writing exists only to hold the pen.

And that pen is particularly vicious. Ninth Circuit Second Amendment litigation is “a lot like watching a Harlem Globetrotters game.” The court is “like a clever, spoiled child.” A summary reversal is “the same monosyllabic correction that a frazzled parent gives her whining toddler,” which is “kind of embarrassing (to the toddler).” He drafts the Supreme Court’s imagined reply for it — “Your decision sucks…. And it sucks so badly, it isn’t even worth a closer look” — and suggests that forcing his colleagues to write a pro-gun opinion “will operate as a form of exposure therapy.” Anyone impressed by the court’s one post-Bruen ruling for the Second Amendment is, “to quote our Chief Justice, a chump.” The whole enterprise, he says at last, is a “cover-up,” and his colleague’s response to him is nothing but “manipulative calls for faux collegiality.”

All of which would be entertaining to read if the man writing it were, say, a columnist on Temu-Above the Law. He is instead a life-tenured federal appellate judge whom the ABA rated not qualified — “arrogant, lazy, an ideologue, and lacking in knowledge of the day-to-day practice including procedural rules.” He turned up on the first shortlist for the next Supreme Court vacancy. Five months ago he was inventing a fictional jurisdiction called the “Circuit of Wackadoo” and scripting his colleagues as Oprah handing out stays of removal.

Which is what takes the fun out of it. The insults exist primarily to get the attention of the White House, should a SCOTUS spot open up. And if the country is ever cursed with a Justice VanDyke… well, we know exactly what he’ll be doing.

Read the full decision below.

Earlier: ‘We Are Better Than This,’ Say Ninth Circuit Judges Despite All Evidence To The Contrary

Judge Pens MAGA-Friendly Dissent That Sure Reads Like A Supreme Court Audition

We Need To Talk About What The VanDyke Video Dissent Gets Right

Ninth Circuit Judges Sick And Tired Of Unqualified Trump Judge’s Spamming The Record With Irrelevant Screeds

Kathryn Rubino is a Senior Editor at Above the Law, host of The Jabot podcast, and co-host of Thinking Like A Lawyer. AtL tipsters are the best, so please connect with her. Feel free to email her with any tips, questions, or comments and follow her on Twitter @Kathryn1 or Bluesky @Kathryn1