The Atlantic

The Hardest Things to Say to One Another

2026-05-30T11:08:00-04:00

This is an edition of The Wonder Reader, a newsletter in which our editors recommend a set of stories to spark your curiosity and fill you with delight. Sign up here to get it every Saturday morning.

Recently, Russell Shaw realized that he had texted his kids the same two words—Too loud—133 times since 2020. “The backstory to each, I’m sure, was relatively consistent,” he writes. “I was in bed, thinking about my schedule for the next day—a board meeting, a difficult conversation I needed to have—when from downstairs came the noise. Shrieks of laughter. Trash talk escalating over a video game ... Or perhaps it was someone deciding at 11 p.m. that they would absolutely die without a McFlurry, kicking off a negotiation over who should place the DoorDash order.”

The texts Shaw had sent weren’t just instances of minor annoyance: They have become a record of the precious time, he writes, when his kids and their friends were always around and the house was full. “My children knew, I think, that the Too loud texts were not quite what they appeared to be—that, yes, I was saying Keep it down, but what I meant was closer to I know you’re there; I’m glad you’re here.”

Shaw wishes he had known then what he was really trying to tell them; but that’s how it happens, he writes. “You don’t know you’re in the good years until you’re standing in the quiet they left behind.” Today’s newsletter explores what we share with our families, and what we find the hardest to say.

On Family

The Phrase I Texted My Kids 133 Times

By Russell Shaw

And all the things it didn’t say

Read the article.

Mister Rogers Had a Simple Set of Rules for Talking to Children

By Maxwell King

The TV legend possessed an extraordinary understanding of how kids make sense of language.

Read the article.

The Questions We Don’t Ask Our Families but Should

By Elizabeth Keating

Many people don’t know very much about their older relatives. But if we don’t ask, we risk never knowing our own history.

Read the article.

Still Curious?

Why we speak more weirdly at home: When people share a space, their collective experience can sprout its own vocabulary, known as a familect, Kathryn Hymes wrote in 2021.
Want to understand Socrates and Sartre? Talk with your kid. Children might have a natural aptitude for grappling with our deepest philosophical questions, Elissa Strauss wrote in 2022.

Other Diversions

Courtesy of Bliss G.

I recently asked readers to share a photo of something that sparks their sense of awe in the world. “On my morning walk today, where there was only a low wall a few days ago, peonies bloomed in a color combination I’d never seen before,” Bliss G. writes. “When I stopped, startled by the beauty of the contrasting dark and light pinks, the blossoms reminded me that not every day is exactly the same.”

I’ll continue to feature your responses in the coming weeks.

— Isabel

Trump’s Approach to Global Leadership

2026-05-30T10:58:17-04:00

Editor’s Note: Washington Week With The Atlantic is a partnership between NewsHour Productions, WETA, and The Atlantic airing every Friday on PBS stations nationwide. Check your local listings, watch full episodes here, or listen to the weekly podcast here.

On a special edition of Washington Week With The Atlantic, David Ignatius, a foreign-affairs columnist at The Washington Post, joined the editor in chief of The Atlantic, Jeffrey Goldberg, to discuss the state of negotiations with Iran, what the future could hold for the war between Russia and Ukraine, and how Trump may view the U.S.’s relationship with Cuba.

A fear is spreading that America’s commitment to upholding the promises that underlie NATO and other global alliances is faltering, Ignatius argued last night. “The NATO umbrella is getting pretty tattered, and I don’t think Americans appreciate just how dangerous that is,” he said. “Other countries will go their own way.”

Watch the full episode here.

The Trump Administration Is Done With Social Science

2026-05-30T09:10:05-04:00

In the summer of 1945, four days after Japan’s official surrender and a few weeks into the Atomic Age, President Harry Truman began floating the idea of an agency guided by “the free intelligence of the scientist” that would fund investigations into how the world works. As of 2024, the agency that Truman had envisioned, the National Science Foundation, supplied about one in every 10 federal research dollars going to U.S. universities. Its Social, Behavioral, and Economic Sciences division funds roughly 63 percent of academic research in the psychological and social sciences, according to the NSF.

The Trump administration now seems determined to shrink the NSF—and to quash its ability to fund social sciences. The Trump administration has proposed cutting the agency’s budget in half and eliminating the SBE division altogether in the next fiscal year. Congress would need to approve those changes, and it may not: Last year, when President Trump also requested drastic cuts to the NSF, Congress rejected them and warned the White House against cutting federal research dollars for any one division by more than 5 percent. But already, change has come. On April 24, Trump fired all 22 members of the NSF’s board, which must approve any major changes at the agency. They have not been replaced. (The NSF did not respond to requests for comment.)

The outlook is especially grim for social sciences. In response to a detailed list of questions, the White House spokesperson Kush Desai told me in an email that the administration “is committed to cementing America’s dominance in cutting-edge technologies of the future—innovation that is being driven by advancements in hard sciences, not in ideologically-driven ‘social sciences.’” In an all-staff meeting last month, a group of NSF leaders said that the SBE division would shut down. According to two current agency employees (who, like other government employees I spoke with for this article, requested anonymity out of fear of retribution), the leaders also announced at the meeting that experts who review grant proposals related to the social sciences would be reassigned to different departments within the agency. Some staff have already moved, the two employees said.

Staff at the NSF learned last month, too, that the SBE division’s research budget for the current fiscal year is two-thirds smaller than last year’s, several current employees told me—and last year, funding was already at historic lows. Even that money seems not to have been passed on to researchers. By late May in a normal year, the NSF would give out about 250 social-science awards. This year, it has distributed five, according to Grant Witness, an effort that tracks federal research spending.

Social scientists have been alarmed about the National Science Foundation for months. Beginning last year, the agency ended its support for doctoral-dissertation research in archaeology, linguistics, geography, and anthropology. That funding had been a lifeline in some fields. “Now it’s much less clear how independent research by early-career anthropologists can be supported,” a current NSF employee told me. Last month, 160 behavioral and cognitive scientists attended a Zoom meeting to discuss how they might save the SBE division. (Among the academics whose fields are under threat, behavioral and cognitive researchers may have the least reason to worry: Trump’s budget request calls for sparing some funding for this type of research, possibly because of its utility in developing AI.) National organizations that represent NSF-funded academics communicated the “devastating implications” of Trump’s proposed changes to the agency’s board—which soon would cease to exist. “This is their only place in the federal government to get support,” Antoinette WinklerPrins, a geographer and senior official at the SBE division until last April, told me. “If that money is gone, that is just devastating to those sciences.”

If the National Science Foundation does stop funding social scientists, experts told me, our 5,000-foot-view of American life will get foggier. The NSF, through the SBE division, is the primary funder of the “big three” social-science surveys, which have enabled the work of several generations of academics, economists, and policy wonks. The surveys are run out of university centers, but the agency helps offset the massive cost of executing them, including by funding database upkeep, compensation for thousands of participants, and the surveys’ ground game. (All three still conduct face-to-face interviews.) At times, support for these surveys has accounted for roughly one-sixth of the NSF’s entire social-science budget; according to two NSF employees I spoke with, there has been no indication that the surveys would be insulated from larger cuts.

One of these projects is the world’s longest-running survey of families, which allows for the study of economic mobility and the long-term effects of child poverty; at least nine federal agencies rely on its data. Another is the General Social Survey, which asks about virtually every aspect of domestic life, including respondents’ pets, cultural values, credit history, and general satisfaction. Without federal money supporting this data collection and the SBE division’s other research, the American Political Science Association said in a statement last month, academics’ ability to understand shifts in American attitudes would “undoubtedly weaken.” The last survey is the American National Election Studies database, which has tracked American voting behavior since 1948. In the 1990s, it showed a growing mistrust of government and animosity between parties, which helped birth the study of political polarization. (Polarization is one of many “DEI” words on a list that Senator Ted Cruz’s office compiled last year to flag NSF proposals for extra scrutiny, two former agency employees told me.)

The NSF also has already effectively blocked grant making in at least one area that the foundation has historically supported: science and technology studies, an interdisciplinary field that examines issues in how research is done and the societal ramifications of new technologies. The field tends to engage with thorny social and political questions using theoretical frameworks such as feminism and structural inequality—both words on Cruz’s DEI list. Usually, outside advisers review grant proposals at an annual spring meeting, but those advisers have been told that this year’s meeting for science and technology studies has been canceled, according to an email from NSF leadership that I reviewed and that contains no explanation for the decision. “There’s no indication another meeting will be held,” Martha Kenney, one of the field’s reviewers, told me. “That was the thing that was the most alarming.” Without that meeting, grants in the field seem to have no pathway for being funded.

The social sciences can get a bum rap for lacking rigor, and for being frivolous. Accurately measuring emotions and attitudes is notoriously difficult, and many studies in these fields cannot be reliably replicated. Accordingly, debates have raged for decades about whether the federal government has any business funding such squishy lines of inquiry. The Truman White House initially was in favor of the NSF funding the social sciences, but backed down after conservatives objected. In 1975, William Proxmire, a Democratic senator with a zeal for accounting, gave out his inaugural Golden Fleece award—recognizing the most useless government-supported research—to the NSF for funding a study on why people fall in love. Several years later, President Ronald Reagan proposed cutting social-science funding at the agency by 75 percent.

But Reagan’s position softened. Jolted by the proposed cuts, social scientists made the case to Congress that understanding how society functions is good for the economy (which was then in a recession). Lobbyists used social-science research to advance anti-tax policies. Ultimately, government funding for social sciences doubled during the Reagan years.

Some analysts at conservative think tanks are making similar arguments in favor of the social sciences today, despite a general furor among Republicans for mocking ridiculous-sounding research. Michael Strain, the director of economic-policy studies at the American Enterprise Institute, recently wrote on X that “if you care about giving businesses and policymakers the information they need to understand the world,” you should care about the Social, Behavioral, and Economic Sciences division staying. Joshua Katz, a senior fellow at the institute who has helped review NSF grant proposals in linguistics, thinks the federal government should continue to fund good social-science research even if it lacks immediate practical benefits. “There are actually quite a lot of sociologists I like, which is not the sort of thing that you would perhaps expect to hear from somebody like me,” Katz, who has critiqued “hyper-woke ” excesses in academia, told me. “A civilized society,” in his view, should earmark at least some money to understanding the human past through fields such as archaeology and anthropology.

Plenty of NSF-funded social science has turned out to have concrete benefits. In the 1990s, NSF funded economics research that was then used to create a far more efficient national kidney-donor-matching system. U.S. families now save more for retirement because an NSF-funded study by a tax researcher discovered the right tactics to nudge them. Even the work of the much-ridiculed love researchers has remained relevant. Twenty years after being awarded the Golden Fleece, one came up with the idea of emotional contagion, a theory for how exposure to others’ feelings influences our own. (They never received NSF funding again; one of the researchers credits the loss of her dog, marriage, car, and shot at an early retirement to the Fleece.) That concept has since been used to study the impact of social media on mental health—something that the Trump administration has said requires urgent attention.

A coalition of about 40 organizations representing hard-science disciplines, led by the Computing Research Association, recently told Congress, too, that losing the SBE division could lead to “long-lasting, potentially permanent” damage to national research as a whole. The stickiest barriers to progress in areas that the administration wants to prioritize—AI, biotechnology—are “fundamentally human,” the organizations argued. The SBE division often funds tacking a social scientist onto interdisciplinary research projects, Sara Kiesler, a former SBE-division head whose own research has focused on how the adoption of email changed the workplace, told me. And because other NSF divisions would still need to understand how humans interact with technical systems, they might essentially create their own in-house SBE divisions anyway. Whether this administration recognizes it or not, rigorously studying how our society works is possible, and helpful.

The Pope’s Admirers Are Missing Something

2026-05-30T08:00:00-04:00

Magnifica Humanitas, Pope Leo XIV’s new encyclical “on safeguarding the human person in the time of artificial intelligence,” has received widespread praise. This isn’t surprising. A popular and learned world leader with a significant degree of moral authority is pointing out the dangers of a deeply unpopular technology created by deeply unpopular people.

The laudatory coverage of the encyclical is justified, but it has obscured perhaps Leo’s most important insight. Magnifica Humanitas (“Magnificent Humanity”) is not only—or even mostly—about AI itself. And the pope is challenging much more than Big Tech.

To be sure, Leo devotes considerable space to addressing the technology and its purveyors. In one of the document’s most-quoted passages, he declares that AI “must be disarmed” and prevented “from dominating humanity.” Elsewhere, he argues for the need “to establish adequate regulatory tools capable of upholding justice and curbing the distorting effects of technological power” on human relationships, working conditions, public discourse, international affairs, and much else. Given AI’s “energy-intensive infrastructure,” the pope warns, “it is essential to develop more sustainable technological solutions that reduce environmental impact and help protect our common home.” He also points to the impunity bestowed on AI developers by the extraordinary resources they have at their disposal.

[Elizabeth Bruenig: The pope grasps the limits of AI]

But Leo doesn’t stop there. “Technology promises emancipation” for the stable and secure, he writes, which in turn “produces new forms of global subordination” for those in precarious situations. No doubt, many encyclical readers fall, like me, into the first group. He’s implicating us, too.

“Nothing in the world of AI is immaterial or magical,” Leo challenges us to keep in mind. “Every seemingly immediate and flawless response is the result of a long chain of mediation, involving vast networks of natural resources, energy infrastructure and, above all, people.” These include people “working under demanding conditions for minimal wages,” Leo writes. “In some regions of the world, children and adolescents work in dangerous conditions, crushing the materials from which rare earth elements are extracted. The bodies of these people are scarred, injured and worn down so that computational flow may continue uninterruptedly.”

Leo rightly calls this a form of slavery—and the ones who benefit from it include much of his audience. We are the ones who demand and expect endless, frictionless data so that we can be more efficient and adept at work and school, more entertained and eased at home, more liberated from the slow, hard effort of reading and understanding. And instead of reckoning with our own complicity, we blame AI and Big Tech. Now, thanks to Magnifica Humanitas, we can selectively quote an authority no less than the pope to heighten that blame even more.

AI hasn’t created this situation. Rather, it has dramatically accelerated a preexisting one in which human affairs were already governed by paradigms that place the ultimate good in technology, economics, and unconstrained individualism. “We live at a time of significant spiritual and cultural blindness,” Leo writes, thanks in part to “a disconcerting loss of historical memory.” Warmongers prolong conflict “as a source of power and income,” while “globalization has provoked fundamentalist, identity-based and nationalistic reactions.”

Humanity, not technology, is responsible for this polycrisis. Accordingly, Leo says that we’re responsible for remedying it. This will require building “a civilization of love” based on “the sum total of small and steadfast acts of fidelity that serve as a bulwark against dehumanization,” the pope writes. He grounds this account in biblical precedents, invoking the humble, gradual, communal rebuilding of Jerusalem recounted in the Book of Nehemiah, contrasting it with the story of the proud, striving, failed effort to build the Tower of Babel. Throughout the encyclical, and especially near its conclusion, he also invokes Christianity’s core proposition, which accounts for the title of the encyclical.

Humanity is magnificent, Leo says, because God made us in his own likeness; because a first-century Jewish teenage girl from Nazareth gave birth to the son of God; because this indwelling of the eternal divine in mortal human form endows humanity itself with the highest possible meaning, purpose, and value.

[Read: The pope doubles down on the beautiful struggle]

To nonbelievers, the Jesus-and-Mary stuff may be unconvincing, even a little ridiculous. But Leo proposes that the incarnation is the firmest possible ground for the conviction that “no computational system, however sophisticated, can create a heart that gives itself, or a conscience that discerns good from evil.”

Leo has offered an answer to the signal question of the digital age. This question isn’t what to do about AI—that’s secondary. The question is: What does it mean, what does it take, to be human? You don’t need to agree with the pope about why humanity is more magnificent than AI, but his encyclical reveals what’s happening now that fewer and fewer people believe that it is.

A Surprising Spin on the World War II Drama

2026-05-30T08:00:00-04:00

The World War II drama has been a hearty staple of the film industry’s diet for more than 80 years—even as Hollywood has turned away from the kind of meat-and-potatoes offering that the genre represents. And after so many decades, directors somehow still keep finding new narrative nooks and crannies to explore. Take Anthony Maras’s latest movie, Pressure, which asks a question that had never occurred to me: Just how stressful was it to be the person tasked with picking the opportune moment for the D-Day landings? Would anyone be shocked to learn that it was, in fact, incredibly stressful?

Of course they wouldn’t. Indeed, the title works two ways—there’s air pressure, and then there’s office pressure, and this movie has heavy helpings of both. If that feels a little on the nose, then Pressure may not be for you. But it is the kind of straightforward bit of dad-bait I am always happy to see in a theater; it somehow manages to invest real tension in a story that has been told many times on the big screen. Although everyone watching knows that World War II is going to go the way of the Allies, the film makes that feel less like a guarantee.

An adaptation of the playwright David Haig’s 2014 stage play, Pressure dramatizes the story of the Royal Air Force meteorologist James Stagg, who was vital to planning the final stages of D-Day. As with any biopic, it nips and tucks some details, excising a historical figure or two. But the tighter focus is both simple and effective. The film zooms in on the few final days before the Allies landed their troops in Normandy in 1944, as Stagg tried to predict whether their boats would be washed away by a huge storm.

[Read: Nothing prepares you for visiting Omaha Beach]

Stagg is played by Andrew Scott, a versatile actor capable of portraying preening villains, swoon-worthy love interests, and emotional wrecks. Here, his remit is clear-cut: possess the stiffest upper lip imaginable. Stagg is taciturn, data-driven, and hardly a people person, shuffling between bustling rooms in Southwick House (the headquarters for the D-Day planners) with nary a word to his compatriots. He’s worried about his pregnant wife at home. But just as stressful, as he consults his weather maps, is the legion of military minds breathing down his neck—chief among them Supreme Allied Commander Dwight D. Eisenhower (played by Brendan Fraser).

I confess that my image of Eisenhower, perhaps tinged by his later presidency, is of a slightly reedier and more thoughtful person than the one Fraser embodies. In Pressure, the commander is a seething powder keg, carrying years of experience on his shoulders as he readies for the last big push to win the war. Fraser keeps things mostly nervy rather than fully screamy, yet his Eisenhower is still a big, angry bulldog—he’s intent on getting the answer he wants and befuddled by the stoic, mysterious Brit telling him that the day he’s picked for the Normandy invasion will result in all Allied ships being battered by a disastrous storm.

Although, going into Pressure, I wasn’t very familiar with Stagg, I could not believe how invested I was in the specifics of an event whose conclusion was foregone. The story explores the margins of decision making, through the lens of how a difference of 24 hours ended up being vital to Allied success. Maras, who directed as well as co-wrote with Haig, is wise to confine most of the action to Southwick House; the stakes of World War II are already obvious. Every fraught conversation thus feels life-and-death; Stagg is trying to convince a group of baffled Americans, including a hotdogging meteorologist named Irving P. Krick (Chris Messina), that the Northern European climate can viciously change on a dime—no matter how sunny the skies might look.

[Read: Revisiting the grim heroism of Saving Private Ryan]

As a former longtime resident of London, I nodded emphatically at Stagg’s warnings, tickled by this ultimate manifestation of the Brit who endlessly discusses the varying nastiness of the forecast. That character detail speaks to the mix of styles at the highest levels of Allied leadership—among them Eisenhower’s ruthlessness, Krick’s pugnaciousness, and the wise and moderating affect of Eisenhower’s secretary, Kay Summersby (an excellent Kerry Condon). Pressure, however, is first and foremost a celebration of a particular kind of frosty British calm: a rigid insistence on structure even when one faces the total chaos and terror of war. As a chamber piece set within the grander opera of battle, it’s a comforting watch. And as a fresh meat-and-potatoes dish offered by a type of cinema that will hopefully never expire, it’s satisfying too.

America Has a Pangram Problem

2026-05-30T07:30:00-04:00

Basically every recent, high-profile accusation of someone passing off AI-generated writing as their own has started in the same way: with a tool called Pangram. In March, when a horror novel from a major publishing house was pulled just days before its scheduled U.S. release date, it was in part because Pangram, an AI-detection program, had identified the text as AI-generated. Other people have fed text into Pangram to suggest that chatbots have been used to write articles in major newspapers including The New York Times, multiple short stories awarded a prestigious literary prize, and most recently, significant chunks of Pope Leo XIV’s encyclical warning about the dangers of AI. The tool is also used by universities to vet student work and scientific associations to scan research papers. As panic builds over AI-generated writing, Pangram is at the foundation.

Just a few years ago, it seemed like it might never be possible to instantly and reliably determine whether a piece of text was written by a bot or a person. In 2023, one detection tool, ZeroGPT, declared the U.S. Constitution to be AI-written; the same year, OpenAI abandoned its AI detector altogether owing to a “low rate of accuracy.” And that was when the quality of ChatGPT’s writing was markedly worse than it is today. But detection tools have gotten much better of late—and Pangram, in particular, has emerged as the gold standard: Paste a chunk of text into Pangram, and the model appraises what portions were “AI Generated,” “AI Assisted,” or “Human Written.”

Yet an AI detector that is mostly reliable might in some ways be more dangerous than a broken one. While Pangram is accumulating the power to end reputations and careers, the tool does make mistakes, perhaps to a greater extent than is currently understood. In turn, AI accusations could very quickly spiral into a witch hunt.

[Read: AI-writing scandals are getting very confusing]

Pangram says its algorithm is so accurate that it incorrectly identifies text as an AI output only about one in every 10,000 times. “There is a great responsibility, a huge weight” in saying something is AI-generated, Max Spero, Pangram’s CEO, told me. “The only reason we do so is because we’re extremely confident.” Several independent analyses have also confirmed that it is quite good. One paper, from the University of Chicago, found that Pangram had almost no false positives on some 3,000 sample texts of roughly 500 to 1,000 words.

But Pangram’s ability to guarantee something was written by a human is shakier. Spero pointed me to a test showing that Pangram’s false-negative rate, or how frequently the model incorrectly labels text as human, is closer to one-in-70 (although some other assessments say it is more accurate than that).

Part of the problem is that Pangram is in an arms race with the major AI labs, which have an interest in making the writing of ChatGPT and Claude sound as natural and human as possible. And at the same time, Pangram has to deal with AI “humanizers”—programs designed explicitly to disguise AI text as your own. Reddit users rave about a humanizer called Walter Writes AI, which I decided to test out for myself. I had ChatGPT and Claude write brief articles, then pasted them into Walter Writes AI. The program, like other humanizer tools, does some anodyne rewording, swaps one clunky transition clause for another, and introduces grammatical oddities. For instance, ChatGPT’s “The numbers are no longer small enough to ignore” became “The sheer size of these usage figures can no longer be ignored.” When I pasted any output from Walter Writes AI into Pangram, it invariably told me that the twice-baked AI article was human-written. (It’s worth mentioning that The Atlantic forbids using AI-generated text unless labeled as such, and that I do not use AI for research.)

Pangram, in other words, can only provide so much insight. A teacher at a public high school in New York City told me that he has “run some of my students’ papers through Pangram, and it shows up as 100 percent human. And I don’t think it is.” He knows what his kids are capable of and, especially for those with a history of cheating with AI, ample reason to doubt Pangram. (I agreed not to identify the teacher by name so that he could speak freely about how he suspects his students are using AI.) But on the flip side, accusing a student of getting undisclosed help from a chatbot with circumstantial evidence is high stakes: The student will either fail or, if exonerated, be bitter and resentful. “The stakes are so high,” the teacher said, “but our way of assessing what is AI-generated is still so unformed.”

Further complicating matters are the opaque ways in which Pangram and similar tools are designed. The model was trained by feeding it mountains of examples written by a human and by a bot—a book review in an actual magazine, then a review about the same book in the style of the same magazine, but produced by ChatGPT—until it can tell the two apart. This is akin to feeding millions of photos of cats and dogs into an image-recognition algorithm until it learns to spot the differences. Pangram cannot point to much specific evidence or patterns in diction, phrasing, or punctuation to support why it deems something AI or human. (I do not, for instance, understand why “these usage figures” was more human than “the numbers.”) Moreover, while Pangram distinguishes between “lightly” and “moderately AI-assisted,” these broad categories can mean just about anything short of copy-pasting from Claude—using AI for research, coming up with counterarguments, as a thesaurus, for a grammar check. The algorithm’s inner workings are “pretty uninterpretable,” Spero said, and although he wants to make Pangram’s “AI-assisted” label more granular, he is also “still not sure how possible it is.” Amid concerns of overreliance on AI chatbots, we risk simply layering on dependence on yet another black-box algorithm.

[Read: The people outsourcing their thinking to AI]

Spero told me that Pangram should “never be the ending arbiter” but instead a starting point for a more thorough investigation, and that the company looks into every reported error its model makes. He also noted that all sorts of detection technology we rely on—smoke detectors, TSA scanners—have base error rates too. On some level, in all these cases the biggest problems lie not in the technologies themselves but in what they’re trying to detect. It’s a problem that buildings catch on fire. It’s a problem that AI is seeping haphazardly into every facet of written communication.

As AI-writing accusations continue to escalate, though, there will only be greater reliance on Pangram—or whatever AI detector can dethrone it—to convict or exonerate. Consider that Pangram can connect to Canvas, the popular education platform, allowing teachers to use it to scan student submissions. There are more than 10 million high schoolers in the United States and some 20 million undergraduates, each of whom likely submits many dozens of written assignments every year. At that scale, Pangram would produce plenty of false accusations even with a one-in-10,000 error rate.

Nor is it guaranteed that Pangram will improve or even maintain its current ability to spot AI prose. As chatbots and AI humanizers adjust, AI detection “will wax and wane in its effectiveness for reasons we can’t predict, at times we can’t predict,” Tim Requarth, a neuroscientist who teaches science writing at NYU and has written extensively about AI detection, told me. Even as schools, publishers, scientific institutions, and the like come to rely more on AI detection, any third-party assessments of Pangram’s accuracy will be from weeks, if not many months, in the past—which in the accelerating world of AI renders them all but obsolete. Basing any AI rules or norms on the reliability of AI detection is like building a sandcastle at low tide.

All of this seems like a disaster in the making. The murkiness and ambiguity of AI detection creates room to launch or deny accusations of nearly any sort. Earlier this month, the technology journalist Taylor Lorenz was accused on X of using AI to write a story for Vanity Fair, which she vehemently denied. Spero investigated and, as he detailed on X, found that Pangram had erred. “Thank god for edit history,” Lorenz told me. The experience heightened Lorenz’s concerns about such allegations: “I’m so paranoid,” she said.

“AI-generated” and “AI-assisted” can be easily confused, by accident or in bad faith. James Taranto, an editor at The Wall Street Journal, recently called Pangram a “defamation machine” and claimed it had falsely flagged three op-eds in his newspaper as AI-generated; two of the implicated authors admitted to using AI to revise some of their work, which Taranto wrote is “inaccurate and unfair to characterize” as “AI-generated.” One of the people who first used Pangram to analyze Pope Leo’s encyclical noted that, because only some sections seemed AI-generated or AI-assisted, perhaps it was not the pope himself but some senior Vatican officials who had used AI while drafting portions of the text. That didn’t stop headlines such as “Did the Pope Use AI to Write About the Dangers of AI?” (The Vatican did not respond to a request for comment, although a writer who covers the Vatican said on X that the AI allegations are “100 percent false” and that Leo actually drafted the encyclical with pen and paper.)

All of this recalls another recent moral outrage over alleged writerly misconduct: The plagiarism wars of 2023 and ’24, when right-wing activists such as Christopher Rufo mobilized to accuse high-profile academics and university leaders of plagiarism—most notably leading to the resignation of then–Harvard president Claudine Gay. Many of these accusations were spurious and likely based on the assessments of plagiarism-detection algorithms that, as my colleague Ian Bogost judged at the time, were fairly useless. The AI-detection wars to come may be even more contentious.

Pangram, to be clear, is not useless. But this is exactly the problem: It’s too easy to twist and contest Pangram’s conclusions, especially when nobody really agrees on which uses of AI are or aren’t ethical. Just like chatbots, AI-detection tools have become effective enough for widespread use, but not reliable enough to fully trust. In this way, Pangram and other detectors are mirror images of the AI products they are hunting for.

The Feeling of Control Slipping Away

2026-05-30T07:00:00-04:00

Back in the web-traffic-obsessed days of 2018, at a time of dawning awareness of how easily audiences online could be manipulated and spoofed by bots, the writer Max Read argued that the internet had crossed a threshold known as “the Inversion.” Not only had bots proliferated across the internet; they had come to constitute it. In outnumbering humans, bots were also loosening everyone’s grasp on the very reality of online experience. “What’s gone from the internet, after all, isn’t ‘truth,’ but trust: the sense that the people and things we encounter are what they represent themselves to be,” Read wrote.

Today, “the Inversion” feels almost quaint. Autonomous AI agents roam the internet, answering emails, sending texts, and occasionally deleting the code repositories of entire companies. An endless library of chatbot-speak crowds out human-written words in every Google search. Bots are spinning up music and videos, conjuring bad poetry and prose, building websites, doing research, making transactions, writing plodding memos to your boss, solving geometry conjectures. Those AI outputs then ride the rails of an internet controlled by black-box algorithms. Computers talk to computers, producing information to train computers to sound more like humans or to better engage them. Humans type into the box, scroll, and wait.

AI is driving people insane in all kinds of ways. Its overwhelming speed and existential stakes have given rise to generalized malaise and hostility directed at the industry, to say nothing of actual cases of AI psychosis. But a lot of this is subtler—a deepening of the bewildering, corrosive feeling Read previously described. Culturally, the flood of slop, AI influencers, fake accounts, and AI tools is blurring the lines of an already post-truth age. A specific paranoia is in the air, an abiding concern about being manipulated, suckered, influenced. Stealth marketing campaigns, mercenary armies of bots, and paid clippers have led anyone or anything that appears dubious to be deemed a potential “psyop.” Cheap imitations of expressions of human creativity are easier than ever to fake. Sentiment, perhaps even popularity, is easier to manipulate. On top of all this is the push into agentic AI—a future we’re told will consist of an internet crammed with bots performing human tasks.

People who don’t feel empowered by all of this are unmoored. Across so many levels of culture, there’s a feeling of control slipping ever so slightly away. You, me, all of us, whether or not we enjoy or use these tools, are living through a crisis of agency. The agita and paranoia, even the excitement—over AI’s encroachment on work, education, art, and culture—is the by-product of a cultural and technological moment in which humans are sliding into a more passive role in many activities. One way to look at the generative-AI boom is as a massive societal experiment foisted on us by Silicon Valley, the animating question of which is: What is a human for?

When you start looking, you see the anxiety over agency everywhere. You see it in the reactions to the mass layoffs at places such as Meta in preparation for an AI transformation, in the coverage of venture-capital-funded, bulk-content-creation bot-army start-ups that proudly claim, “Never pay a human again.” You can sense it among the software developers who feel that their reliance on coding tools is eroding their skill set, in the executives who confess that they don’t know whether their AI spend is justifiable. Or when you read reports that medical journals are filling with made-up citations, or a study that suggests that chatbot use is degrading our thinking, or an announcement from Google that it will offer an alternative to its link-based search results: AI agents that can scan the web on your behalf and either bring back a canonical answer or send you personalized alerts.

The discomfort is playing out in real time. Last week, after the literary magazine Granta published the Commonwealth Short Story Prize–winning story “The Serpent in the Grove,” suspicious readers began to point out what they believed to be evidence of chatbot text in the story. Soon, two other Commonwealth Prize winners came under similar scrutiny, as people began running passages through AI detectors. (The Commonwealth Foundation first said in a statement that none of its prize winners had used AI, but then it issued a second statement suggesting that it is taking another look.) AI boosters celebrated the news as an example of the sophistication of current language models; skeptics viewed it as something of a slop tipping point. In a recent essay, the writer Sam Kriss described the experience of scrolling through websites right now: “The more I clicked around, the more I started to panic. There was nothing, no human voices anywhere, just thousands of versions of the same cheery demon. Am I alone out here? Something’s happened to the world; it’s all gone flimsy.”

The glut of AI writing, the detection arms race, and the debates over what constitutes appropriate use are part of the bigger questions posed by this technology. What does it mean to outsource our creativity? How long will we have the ability to discern whether something we like is human or not? Does our taste matter? If it doesn’t, then what are we even doing here?

It’s unsurprising, perhaps, that at the same time when Silicon Valley is building and breathlessly promoting these tools—self-directed agents that can accomplish complex tasks without human supervision—many of its loudest voices have grown obsessed with the idea of their own agency. In builder circles, people deemed “high-agency” sit atop the hierarchy. They are individualistic, ambitious, focused. They just do things. They are especially adept at marshaling the use of people and machines alike. It is implied that those with high agency are, for now, insulated from becoming replaceable or irrelevant in a time of great precarity—not yet doomed to be part of “the permanent underclass,” another Bay Area coinage for the late adopters who will be left behind. How could a person hear such language and not feel at least a little paranoid?

AI companies use the term human-in-the-loop to describe the relationship between humans and AI tools in everything including chatbots and warfare. The humans perform managerial tasks: They prompt, evaluate, approve, monitor, correct. Being in the loop is meant to sound active, but the truth is, beyond the prompt, what humans are so often doing is reacting to an interaction of multiple machines. As chatbot- and AI-assisted search has outsourced web exploration to language models, this dynamic has become a primary way that humans interact with the internet. Sifting through a mix of AI- and human-generated videos, images, and text on your social-media feeds? You are passively consuming an interaction between an algorithm and things made by a computer. This is why being glued to our phone and feeds can feel so extractive and joyless, even numbing.

Recently, the online-culture researcher Aidan Walker memorably described the user experience across much of the internet as akin to the user being cuckolded by the endless scroll. People have long theorized that reliance on algorithms and the flood of bots and fake content have led to an internet that’s effectively dead. Walker’s theory is that the internet is not “dead” or “fake” but that models and algorithms have the bulk of agency online. Humans in the loop sit and watch, voting on short-form videos and giving feedback to the machines with every swipe. It’s a bleak vision. It’s also difficult to argue with.

In this system, it’s not hard to see how mistrust develops, how one might suspect ulterior motives behind every piece of information they get served. Are these really the best wireless speakers, or am I falling for SEO slop? Is this band I don’t like really popular? What is popularity anymore? Reality starts to blur. Everything goes flimsy.

The AI companies and boosters speak the language of empowerment. They’re not wrong in the sense that the tools are powerful and, in many cases, quite useful. But it’s tough to overstate how much these tools represent a reversal of the early promise of the weird and wild internet; of user-generated content; of stumbling upon information, people, and communities; of the crackling-static feeling of real people on the other end of the modem. In 2011, the writer Paul Ford described the internet as a “customer service medium,” arguing that “humans have a fundamental need to be consulted, engaged, to exercise their knowledge (and thus power), and no other medium that came before has been able to tap into that as effectively.” Ford was describing the impulse behind the creation of user-generated-content sites such as Reddit and Wikipedia but also the rise of comment sections, likes and thumbs-up reactions to posts and videos, and the very instincts that led to the rise of social media, before it became quite so algorithmically mediated.

Revisiting that essay now, in the middle of our current agency crisis, I’m struck by the fact that Silicon Valley may have taken Ford’s words to heart, but in service of the opposite outcome. It has built arguably the most impressive customer-service medium in the history of our species—an infinitely scalable, highly personalized answer machine that flatters our insecurities and mimics our idiosyncrasies. But this system automates, and even negates, the human need to be consulted. Now we consult the chatbots and they provide canonical answers. Generative AI’s much-touted efficiencies also often erase the necessity of collaboration. They do the problem-solving, the heavy lifting. It is no surprise, then, that the backlash to AI, particularly the opposition to data centers, has come in the form of protests and public comments at town and city-council meetings across the country. People are taking up their agency in one of the few places they can: the physical world.

The AI companies seem to be missing all of this. To them, only the information—not the humanity—has value for their models. This is not right, but so many of us, lost in a swirl of ceaseless information, will be lulled into thinking that it is. In the new paradigm, we are not so much consulted as tasked with feeding data into the machinery. We perform our humanity, and the machine learns to mimic us. Its goal is to be better at whatever we’re doing than we are.

The AI industry has ushered in an era of offshoring that’s as irresistible to some people as it is disorienting to others. What’s at stake is not just truth or trust, but also a sense of direction, orientation, and purpose for all of us. Do your own research became shorthand for the problem of the internet leading people to ivermectin hawkers and Pizzagate conspiracy theorists. But it could just as easily describe the thrill of autonomy promised by the open internet. We may miss it when we fully reckon with what comes next: a black-box machine of obscure corporate motivation that simply tells us, Here is the answer.

The Kennedy Center Enters the Unknown

2026-05-29T21:13:00-04:00

For months, the dwindling ranks of staffers at the Kennedy Center have been bracing for July, when the Washington, D.C., arts complex had been slated to shut down. How the bruised institution would bounce back after a two-year closure ordered by the president of the United States—and what it would look like once it did—were major questions. This week brought an even bigger one: How could it possibly stay open?

In a pair of rulings today, a federal judge dealt two blows to Trump’s stewardship of the Kennedy Center, which he took over last year: U.S. District Judge Christopher Cooper ordered the removal within two weeks of Trump’s name from the institution, which Congress established as a living memorial to President John F. Kennedy in 1964, and he partially granted a preliminary injunction, saying that the center had to halt plans to close. “There is no evidence that the Board took account of its full range of statutory obligations in determining that a wholesale shuttering of the Kennedy Center was appropriate,” Cooper wrote in a 94-page opinion in a lawsuit filed by a member of Congress. Trump announced the Kennedy Center’s two-year renovation in February, following a year in which the politicized center had seen audiences plummet and prominent artists cancel appearances.

And now? The president wrote today on Truth Social that he wants to offload responsibility for the Kennedy Center to Congress: “Unless I am free to do what I do better than anyone else, bring this Institution back, physically, financially, and artistically, I have no interest in continuing what could only be a hopeless journey into ‘NEVER NEVER LAND.’”

In December, not long after Trump himself hosted the Kennedy Center Honors, a board whose general trustees were all appointed by Trump voted to rename the Kennedy Center to include the president. But Cooper said that the law made “crystal clear” that the building was to be named for Kennedy alone. “Congress gave the Kennedy Center its name, and only Congress can change it,” Cooper wrote. He left open the possibility of the board closing the center for renovations after “independently balancing its multiple obligations to the Center in a prudent fashion.“ (In a separate lawsuit, filed by a coalition of historic preservationists and architects, Cooper denied a similar request for an injunction, because the plaintiffs had not shown that the renovations were subject to certain federal-review processes.)

The Kennedy Center wrote to me in a statement that it would review the judge’s order to keep the institution open and added that it would pursue every legal option to carry out the planned renovation work. It was more direct about Trump’s name. “We are confident that on appeal the court will uphold the Board’s will to recognize President Trump’s historic contributions to our nation’s cultural center,” Roma Daravi, the center’s vice president of public relations, said.

But the center’s comments reveal some dissonance with Trump, whose Truth Social post demonstrates that he is willing to walk away from the cultural institution entirely.

That leaves unclear what will happen next, save that the plaintiff, Representative Joyce Beatty, and the Department of Justice lawyers representing the Kennedy Center will continue to battle it out in court. Trump had insisted that his renovation would restore the creaking building. But staffers I spoke with today worried that he’d already permanently broken the institution that lives there.

The center has felt like a ghost ship in recent months, they told me. With internal communications scarce, programming thin, and departments gutted or entirely shuttered, the national cultural center seemed to be entering hibernation. The employees, who spoke with me on the condition of anonymity for fear of reprisal, described the center as a shell of itself. As their duties have all but dried up, they’ve found themselves with little to do.

“We’ve already shot ourselves in the foot,” one person said. “It would be a herculean effort to try and salvage the absolute mess this has become.”

[Read: What I saw inside the Kennedy Center]

Even before Trump’s February announcement of the shutdown plans, the center was in a state of upheaval as performers, donors, and patrons fled the institution in defiance of the president’s takeover. Beginning in March, the center slashed its workforce with a series of layoffs.

Meanwhile, Broadway tours to the Kennedy Center have been canceled. The Washington National Opera dropped its affiliation to become nomadic. The center’s remaining anchor, the National Symphony Orchestra, has begun to make plans to spend two seasons performing elsewhere. And although the center has not made any recent disclosures about its finances, its resources are likely strained from diminished ticket sales and donations. For longtime supporters of the center, there may at this point be no good outcome.

In lieu of its own programming, many of the events at the Kennedy Center lately have been booked as campus rentals by Trump allies and organizations supportive of the name change, a staffer said, adding: “which makes me feel like even that could dry up when his name thankfully comes down.”

The Kennedy Center Executive Director Matt Floca similarly claimed earlier this week in a court filing that fundraising could be jeopardized by the removal of Trump’s name—a curious declaration to close observers who recall reports of sharp donation drops because of the president’s affiliation. Floca also offered a surprising, and perhaps accidental, window into the center’s health: Despite earlier claims that fundraising had surged to $130 million last year under Trump, Floca told Cooper that the center has raised only tens of millions of dollars in that period.

The filing was the Kennedy Center’s final entreaty to Cooper following several efforts in recent months to preserve Trump’s plan.

For much of this year, leaders at the Kennedy Center have been making the case that their workplace ought to shut down. In March, Trump swapped out Richard Grenell, the pugnacious loyalist he’d tapped last year to lead the institution, for the lower-key Floca, then the facilities head. After a year of negative headlines and artist cancellations, the center’s vibe shifted from spiky political operation to a pending construction site.

Step one was the court of public opinion.

On a sunny midweek morning last month, Floca took a group of journalists, including me, into the bowels of the Kennedy Center for a renovation tour. He led us through water-intruded service tunnels and pointed out the issues that Trump had repeatedly invoked in the past year: crumbling concrete; corroding steel; deteriorating slabs of marble; outdated chillers, boilers, and other equipment.

The dark and damaged corridors certainly looked bad. But I left the tour scratching my head, wondering whether this was all normal wear and tear for a 55-year-old building, and whether its repair ought to really render the 1.5 million-square-foot complex uninhabitable for two years. (Arts leaders generally prefer phased renovations over complete closures to keep audiences in the habit of showing up.) Either way, the Kennedy Center got the result it wanted: Media reports published within hours of the tour presented Floca’s claims without rebuttal and prominently featured photos of rust and decay.

Step two was court.

The two lawsuits, from Beatty and the historic preservationists, had back-to-back hearings in U.S. District Court last month. When Floca took the stand in the preservation suit, he made a case loaded with technical detail and specifics about the renovation timeline—that is, he tried to describe a serious-sounding plan, not a Trump vanity project. From the moment he arrived at the Kennedy Center in 2024, Floca had said that he was “dumbfounded” to see its disrepair.

“Leadership, at that time, knew that we were not telling people the true needs of the campus,” Floca testified. Of phasing the upgrades, he said, “It is just impossible and irresponsible.” In a court declaration, he said that he had come up with the idea of a shutdown, pushing back against the perception that Trump made the call in order to cover for the center’s failings.

Floca sought to create distance from Trump’s implication that he would dramatically overhaul the structure, denying any plans to tear down or rebuild the center. He also characterized the center’s new moniker—the Donald J. Trump and the John F. Kennedy Center for the Performing Arts—as a “secondary” name for the Kennedy memorial.

On the tour, Floca addressed the criticisms that the closure is a smoke screen for the Kennedy Center’s dire finances. “Across the industry, it’s been said many times that sales are difficult for performing-arts centers and that this building, this organization, is no different,” he told reporters. “But the decision to close the center is completely founded in the maintenance needs of this building and not the mission, or not the programming, or not being able to achieve that mission.”

Floca is one of the few executive-level leaders who predate Trump’s takeover and has admirers among the rank-and-file staff.

But he’s not an arts administrator, and there is little sense of how the center might come back to life at this point. This spring, the center had been exploring how to continue some programming efforts in its Reach complex, such as orchestra rehearsals, educational programs, and artistic performances through the Millennium Stage.

Yet the calendars for the venues in its main structure—three massive performance halls as well as several small spaces—are about to be empty. It’s unclear how soon new seasons of programming could even be booked to fill them. Or if audiences will ever show up en masse again.

The Apple Car Is Finally Here

2026-05-29T15:16:49-04:00

Sign up for Ordinary Extraordinary, Ian Bogost’s guide to making everyday life vivid again. You’ll receive the first edition of the limited-run newsletter course in early July.

Transportation has never been a Ferrari’s real purpose. Sure, you can drive one—although not literally you, because you probably can’t afford one. For the few who can, it is an automobile to be seen idling at a stoplight before prancing away, or parked at a luxury-hotel valet stand, inspiring desire and jealousy. For normal people, a Ferrari is a symbol: of power, control, precision, and wealth—but also of the longing for those virtues, and of the idea that they are virtues in the first place. The Ferrari is the quintessential bedroom-poster car, captured in a glossy photo pinned on a wall in a teenage boy’s bedroom like a photo of a scantily clad woman: an unachievable object of desire.

If a Ferrari is an object of spectacle, an Apple device is an object of function. The Apple product, whether it’s a laptop, music player, smartphone, tablet, speaker, or watch, is designed to dissolve into its context and melt into ordinary life. Frictionless, intuitive, and transparent—in its ideal form, an Apple product ceases to feel like an object at all, and instead facilitates an activity. An iPhone or MacBook expresses style, but through minimalism, an aesthetic concerned with vanishing into the background and becoming obedient to intended purpose. This approach to design transformed the traditions of industrial modernism that it had inherited—from Dieter Rams, Ludwig Mies van der Rohe, Walter Gropius, and others—into an ethos that was demure instead of forward. The best technology would become softened, domesticated, and emotionally deodorized.

[Ian Bogost: Apple is boring now]

The old world of automotive desire and the new one of glass rectangles collided this week, when Ferrari unveiled the Luce, its first electric supercar. The vehicle looks like a Ferrari on the inside but an anonymous lozenge on the outside, a design that some Ferrari fans hate. Does it mean the end of the house of the prancing horse? No. Rather, Ferrari’s first EV is a delightful if wistful marriage that nobody could have predicted. Through this pairing, the Ferrari Luce signals the final victory of the smartphone over the automobile. Nothing aspirational remains that isn’t an expression of the Silicon Valley technology industry.

Although cars remain important in America, they have declined as an expression of identity, replaced partly by online life, where self-expression can go global. Young people don’t care about driving, in part because teenagers aren’t allowed to go anywhere, but also because smartphones made doing so less necessary. Silicon Valley had gone into the transportation business, first with ride-sharing and then with autonomous vehicles. It seemed reasonable that tech companies might play a large role in the future of transit.

From 2014 to 2024, Apple tried and failed to make a car. At first, it was meant to be an actual car, with wheels and everything. Details were scant, but Apple hoped the vehicle could do for the automobile what the iPhone had done for phones—reinvent the category, and with it, the way people lived. Apple hired people from traditional car makers, from Tesla, from battery companies, from autonomous-driving start-ups. Thousands of people worked on the project, code-named Titan, at a reported cost of a billion dollars a year or more.

Apple was in over its head. A car, it turned out, is not like a personal electronic device. Apple tried to pivot Titan to a platform for autonomous driving. But in the end, after a decade, the company gave up. It canceled Project Titan. An Apple automotive future would be left to CarPlay, the software platform that can make your iPhone operate your car stereo and, soon, your climate control and speedometer.

Jony Ive spent nearly three decades at Apple, where he served as chief design officer from 2015 to 2019. He had a hand in nearly every major Apple product from Steve Jobs’s return in the late 1990s through the 2010s—the iMac, iPod, iPhone, iPad, MacBook Air, Apple Watch, and even Apple Park, the company’s headquarters. Ive reportedly became bored at Apple, and he cut ties with the company in 2022. Now he runs LoveFrom, an industrial-design consultancy.

Ive connects Apple’s legacy with Ferrari’s future. The sports-car company hired LoveFrom to design the Luce, inside and out, giving Ive and Marc Newsom, his LoveFrom partner (and fellow Apple alumnus), freedom to design a wholly new automobile. Car and Driver reported that this newness extended to the car’s form factor, electric motors, batteries, steering wheel, physical controls, and digital displays. The car produces more than 1,000 horsepower and costs $640,000.

Due to its price, the Luce operates mostly as a symbol rather than an automobile, as did all Ferraris before it. Yet the car looks nothing like a Ferrari, or at least nothing like the received idea of a Ferrari. It is a four-door hatchback, a configuration that, though not new for the company, is highly unusual for an Italian supercar. It is also the first Ferrari that seats five, betraying the company’s apparent principle of inutility—a Ferrari is supposed to be excessive instead of useful.

But mostly, the Luce is smooth and rounded, resembling an aerodynamic suppository more than a big-haunched cavallino rampante, the rearing horse that serves as Ferrari’s logo. That design produces performance—a “drag coefficient lower than any prior roadgoing Ferrari,” according to Car and Driver—which helps the car accelerate from zero to 60 miles per hour in about two seconds. But lost in the process is the typical Ferrari style: low, taut, and animalistic, like a machine stretched over the musculature of a ferocious creature.

For this reason, the Luce has produced a backlash. Some “Ferraristi,” The New York Times reported, “are finding it difficult to embrace the Luce’s bubblelike exterior.” The former Ferrari chairman Luca di Montezemolo said, according to The Wall Street Journal, “At least, I hope they take the horse off that car.”

One mocking social-media post depicts the car on its back, with a charger inserted into its underside. The joke refers to the Apple Magic Mouse, whose now-infamous design requires plugging it in upside down while it charges, preventing it from being used. The message: A Jony Ive–designed Ferrari brings an unwelcome Apple-design sensibility to an incompatible product and brand. The Ferrari Luce looks like exactly the sort of car that Apple would have made. Now that the smartphone-car is actually here after more than a decade of anticipation, people aren’t sure they actually want it.

In part, that’s because the whole supercar market has been on the wane for at least a decade. In 2015, when Tesla began delivering the Model X, automobiles had already ceased to be an object of desire. A Tesla could keep pace with a Ferrari or Lamborghini back then, but it did so in a humdrum way, stripped of the carnal passion that had imbued its Italian precursors. No teenager would ever hang a picture of a Tesla on their bedroom wall. Nor, for that matter, the Ferrari Luce.

Some critics accuse dinosaur-burning supercar purists of “petro-masculinity,” a misplaced and retrograde attachment to gasoline combustion and climate-damaging excess. Lamborghini dropped plans for its all-electric supercar, the Lanzador, after concluding that demand for it was “close to zero.” Pagani scrapped an electric version of its multimillion-dollar Huayra on the grounds that EVs “lack the emotion” of internal-combustion cars. Gordon Murray Automotive, led by the designer of the McLaren F1, sold its EV division to focus on V12 gasoline automobiles. Aston Martin, Porsche, and Lotus have also scaled back their electric ambitions.

But as Tesla’s and Ferrari’s examples attest, not to mention the Formula E electric-racing circuit, EVs can be just as—or even more—powerful than gas-burning vehicles. The problem with EVs was never their performance on the road.

Ferrari appears to have realized that electric vehicles are the future, and that pursuing that future demands the reinvention of the supercar itself, as well as the supercar company that makes them. Actually taking that risk by designing the Luce as a production model that will be released rather than scrapped or relegated to concept-car purgatory is worthy of praise.

But that kind of risk taking has consequences—Ferrari’s stock was down as much as 8 percent after the Luce reveal. Even so, a Ferrari was always an out-of-reach toy for the ultrawealthy, and owning such a car let the driver forge new and hazardous paths, much like taking risks in business. Seen in this way, the Luce embraces the symbolic spirit of the supercar better than the V12 Pagani or the Gordon Murray T.50.

[Andrew Moseman: A new kind of hybrid car is about to hit America’s streets]

Ferrari may have realized that its old way of chasing wealth and symbolizing power has ended. Apple, Ive, and their kindred beat it years ago. Lamborghini and Aston Martin might see dying on their own terms as more noble than caving to incompatible values. But Ferrari has steered a more sensible course, which also makes its track appear unexciting and even unprincipled. The company has embraced an important virtue, which is that electric vehicles are the future, even for supercars, and embracing that aspiration at the top of the market will help adoption trickle down to the bottom.

Silicon Valley still sees risk in business as a virtue, but its successful industrialists seem to value utility, simplicity, and intelligence over ornament or conspicuous luxury. That ethos is consonant with the design sensibility that pervades the sector. The minimalist principles that Ive brought to Apple became doctrine in the tech industry. Technology was deemed good if it was smooth, quiet, seamless, and emotionally reassuring. Like the Bauhaus and International Style that influenced it, monochromatic, high-tech minimalism is anonymous and somewhat generic, and its capacity to operate anywhere contributes to its ability to scale globally. Ostentation and idiosyncrasy—of the kind that a traditional Ferrari represents—never had much place at Ive’s Apple. Instead, technology was meant to disappear, to conceal complexity, to deliver emotional calm, and above all to present itself as inevitable.

This most famous of Italian-sports-car makers may have realized a more practical truth as well: The tech sector’s ultrawealthy are one of the only markets left for a Ferrari anyway.

Condemning a Nazi Tattoo Shouldn’t Be This Hard

2026-05-29T15:00:00-04:00

For decades, Nazism and the anti-Semitism underlying it have marked zero on the Kelvin scale of villainy—the metric against which all other forms of evil are compared. This is so well understood that we now have cultural phenomena such as Godwin’s Law, the theory that online debates inevitably lead to Nazi comparisons, and the “everything I don’t like is Hitler” meme. But their existence proves the point: If one wishes to say that something is irredeemably bad, Nazis are the benchmark, the absolute.

Yet recently this understanding seems to have grown less universal. Nazi symbolism and more modern versions of the ancient conspiracy theories behind this intolerable ideology have found a degree of toleration within American political movements desperate for shortsighted victories. The underlying hatred that, among other things, motivated the killing of more than a third of all the Jews on the planet eight decades ago is viewed no longer as unacceptable, but rather somewhere on a scale of “problematic” issues that can be either explained away or ignored.

The most recent case is that of Graham Platner, the 41-year-old Democrat who is hoping to unseat Senator Susan Collins in Maine. Platner has a unique personal story, having reinvented himself from high-born prep-school student to blue-collar oyster farmer, and from willing Marine who talked about wanting to go to war to kill people (and who later worked for a military contractor) to a victim of Collins’s vote to authorize the Iraq War. Although Platner is by no means the first politician to reshape his personal narrative during a campaign, he is likely the first to attempt an innocent explanation for having had, for 18 years, a tattoo of a Totenkopf, the insignia of the Schutzstaffel, or SS—the most dedicated and fanatical component of the Third Reich, whose members were the architects and executioners of the Final Solution.

Platner has said that he got the tattoo while “carousing” with other young Marines in Croatia, that he thought of it as simply a skull and crossbones that “looked cool,” and that he was horrified when he learned of its significance. He got the image covered up once it became public. But the idea that he remained blissfully ignorant of the Totenkopf’s meaning strains credulity. CNN found evidence suggesting that he was aware of its significance for years and had spoken with an acquaintance about it. Platner’s former political director made comments to the same effect.

[Tyler Austin Harper: How ‘big tent’ are Democrats willing to go?]

What’s most incredible is not that Platner would attempt to spin something that is obviously disqualifying. It’s that the leaders of his party are accepting that spin. One by one, Democratic politicians have lined up to campaign with Platner or post messages of support. The Pod Save America hosts, former officials in the Obama administration, have gone on a campaign to deride anyone who expresses concern over the tattoo.

This is clear hypocrisy. Democrats once pointed to every example of the “okay” hand signal as ironclad evidence of white supremacism; they cannot suddenly be just fine with Nazi logos. I feel confident in stating that Elizabeth Warren, Seth Moulton, or the Pod Save America crew—or any other Democrat currently defending Platner—would be entirely unsympathetic to a Republican candidate with a Nazi tattoo. To these Democrats, apparently, nothing—not even condemning the greatest evil to have existed on Earth—is more important than a victory against the opposing party. This is what excessive partisanship—and callousness—does to the human brain.

Platner’s best explanations for the tattoo require us to believe that he’s too ignorant about the world around him, too incurious about what he had permanently inked on his body, too impulsive, or too dismissive of what the Nazis represent to know or care that he was wearing their symbol. Even if one accepts the explanation that a drunken 20-something made a bad decision, the greater issue is the discernment of the 40-something who kept the tattoo until it became a political liability. And if he is merely an impetuous, uninquiring dolt, those are not the traits anyone should want in a senator.

Only recently did Massachusetts Representative Jake Auchincloss do what many in his party lacked the wherewithal to do: speak out against Platner and his candidacy, arguing that the SS tattoo and Platner’s comments about it are “disqualifying.” Auchincloss can claim a unique position from which to criticize Platner. Much of Platner’s defense has been that the tattoo was just an example of typical Marine hijinks; he has also spoken more generally about the stress that war caused him. But Auchincloss himself is a former Marine and a veteran of the global War on Terror. He is also Jewish. Rather than follow Auchincloss’s example, however, some Democrats criticized him for stepping forward, labeling him a traitor or calling for him to face a primary challenger. In their eyes, the problem within the party is not the man who bore a Nazi logo; it’s the Jew who said that Nazi logos are bad.

The Democrats are not alone in this malignant thinking, of course. The problem of excusing abhorrent ideologies for the sake of electoral gains is growing more pervasive and universal. In the run-up to last week’s Kentucky primary elections, numerous Never Trumpers rallied behind Representative Thomas Massie against his Trump-aligned challenger, Ed Gallrein. The movement of Never Trump conservatives, in which I count myself, began out of an adherence to the intellectual roots of conservatism, rejecting the fluctuating definitions of it that came with Trump’s brand of populism. But for many, “Never Trump” has become not a statement of the ideas that they stand for, but rather an endorsement of anything, or anyone, that Trump is against.

Massie’s late-arriving Never Trump fans were so singularly focused on his willingness to be a burr in the president’s saddle that they overlooked his many disqualifying traits. In his recent campaign, Massie and a PAC supporting him targeted Jewish American donors who backed his opponent, invoking shadowy anti-Semitic conspiracies and accusing one such donor of being an agent of a foreign power. One of his supporters, William Paul, the son of Senator Rand Paul, publicly and drunkenly shouted at Representative Mike Lawler that if Massie lost, it would be the fault of the Jews. William Paul released a public statement apologizing for having a drinking problem but leaving unaddressed the underlying hatred that his drunken statements revealed; Massie didn’t comment on the report about the incident.

Massie used his concession speech as one last opportunity to stir up blame and animosity for the Jews. “I would have come out sooner,” he said, “but I had to call my opponent and concede. And it took a while to find Ed Gallrein in Tel Aviv.” Such comments echoed old tropes about Jews as “rootless cosmopolitans” or disloyal citizens.

Like Massie, former Representative Marjorie Taylor Greene has been met with waiting open arms from opponents of the president since she broke with him, and newfound condemnation from Trump and his supporters. This is an indictment of both camps. Trump never said a harsh word when Greene spread rumors about Jewish space lasers. Likewise, the MAGA influencers never had a problem with Massie until he challenged the president over the Jeffrey Epstein case—something Massie made less about justice for the victims than about claims of Mossad manipulation. Only criticism of Trump made Greene and Massie worthy of excommunication. Likewise, in the eyes of some Democrats and Never Trumpers, the act of opposing Trump was enough to wash away Greene’s and Massie’s contemptible statements toward Jews.

[Ali Breland: Who gets panzer tattooed on their arm?]

Even having cast out these specific conspiracy mongers, the Trump movement still has plenty of other examples of the same rot. Heritage Foundation President Kevin Roberts refused to sever the organization’s ties to Tucker Carlson after Carlson conducted a friendly interview with the anti-Semitic influencer Nick Fuentes. Since then, only more examples have emerged of Jew hatred among GOP staffers and party members. Nate Hochman, a speechwriter for Ron DeSantis’s presidential campaign, shared a video with a Nazi Sonnenrad and was fired, but then got hired by Senator Eric Schmitt. Paul Ingrassia, who called the October 7, 2023, Hamas attack on Israel a “psyop” and wrote in a private group chat that he had “a Nazi streak in me from time to time,” has been the subject of one controversial report after another but keeps getting moved into new roles in the administration. (A lawyer for Ingrassia refused to “concede the authenticity” of the group-chat messages.)

Rather than purging those who have used slurs, distancing party leadership from such behavior, or even making a clear statement of condemnation, Vice President Vance has suggested that no one on the right should be considered an enemy of MAGA. “When I say that I’m going to fight alongside of you, I mean all of you—each and every one,” Vance said at a Turning Point USA gathering last year, after the Heritage controversy. “President Trump did not build the greatest coalition in politics by running his supporters through endless, self-defeating purity tests.”

Right, left, and center, political parties and movements seem to have lost their ability to affirm a simple statement that shouldn’t be hard to say: Anti-Semitism is bad, and those who traffic in it have no place in our party and will not get our support. Instead, candidates or influencers who flirt with or fully embrace the well-worn tropes and imagery of Jew-hatred have their behavior excused, explained away, or ignored. Nazism, whether diluted in the form of imagery or conspiracies, or concentrated, cannot be tolerated. It’s shameful that it is left to Jewish figures such as Auchincloss to have to speak out against this by themselves, and that they are attacked and criticized when they do.

And it’s shameful that our leaders are willing to accept this over such small stakes. Graham Platner is not the only Democrat in Maine who’s at least 30 years old, nor will the Trump administration fall apart without Paul Ingrassia. These people do not represent the best possible candidates to fill the roles of senator or government staffer. Instead, they represent a human manifestation of their parties’ refusal to let the other side win. In a political environment in which everything is couched in absolute terms, no concession can be offered, no defeat can be accepted, and no standard can be enforced if it could offer the slightest benefit to your opponent.

But if one’s movement or political party is best represented by someone who traffics in hatred, then that party is not worthy of support. It should go without saying, but yes, the organization that manned the gas chambers is worse than Susan Collins, and stoking mistrust of American Jews is worse than Trump getting a reliable vote representing 1/435th of the power of the House of Representatives.

Some might think that I’m being hyperbolic or alarmist in my outrage, that a tattoo need not represent a person’s authentic ideology, or that red-pilled edgelords are just making harmless jokes, as the vice president has suggested. But as my father taught me when I was young, walking past a problem is an endorsement of it, and the new boundaries of what is tolerated by a movement become the new definitions of what is welcomed within it. As with so many things, the rise of extremism happens “gradually, then suddenly.” Like cracks in a dam, each individual flaw might not be large, but the tolerated imperfections grow, and the damage compounds, and without maintenance the entire structure gives way, releasing the destructive force that had once been held back.

Toleration of pure evil, as anti-Semitism is, in small quantities is toleration of the whole. Some things are more important than winning elections.

Why Everyone Hates AI Data Centers

2026-05-29T13:00:00-04:00

Subscribe here: Apple Podcasts | Spotify | YouTube

Data centers are quickly becoming the most polarizing buildings in America. On this episode of Galaxy Brain, Charlie Warzel speaks with the reporter Jael Holzman about the backlash to the buildings powering the AI boom. Why have data centers become controversial? What are the environmental, economic, and political impacts? How does the backlash track along left/right party lines? This episode demystifies the data-center fight.

Get more from your favorite Atlantic voices when you subscribe. You’ll enjoy unlimited access to Pulitzer-winning journalism, from clear-eyed analysis and insight on breaking news to fascinating explorations of our world. Atlantic subscribers also get access to exclusive subscriber audio in Apple Podcasts. Subscribe today at TheAtlantic.com/Listener.

The following is a transcript of the episode:

Jael Holzman: The conversation around data centers is easily a pain sponge. But you gotta wonder where that water is coming from and what that sponge really is made of, because what that sponge is made of is so many local conflicts stitched together.

[Music]

Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain, a show where today we’re going to talk about the most polarizing buildings in America: I’m talking about data centers.

Data centers have been around for decades; they’ve been powering much of what we do online. But the AI boom has created this ravenous need for more computing power, and, in the process, something extraordinary seems to be happening: People across the political spectrum are coming together in opposition to these data centers. The AI backlash has galvanized people like Bernie Sanders and AOC [Alexandria Ocasio-Cortez]—they’ve proposed a data-center moratorium in Congress. But also the populist right, where figures like Steve Bannon are arguing that tech elites who are investing billions in the AI-infrastructure build-out are “totally out of control.”

Most notable, though, is the reaction from regular citizens. A May Gallup poll found that 70 percent of Americans oppose the construction of an AI data center in their community. Across the country, in local town halls and community meetings, grassroots activists and concerned residents are coming together to protest these projects, and, in many cases, they’re winning. As the website Heatmap reported this month, at least 20 proposed data centers were canceled following local opposition in the first quarter of 2026.

The stakes of this fight on either side are reasonably clear. The Big Tech AI hyperscalers are investing historic sums in these buildings. According to my colleague Matteo Wong, “Microsoft, Amazon, Meta, and Google alone have already spent more on data centers since the launch of ChatGPT than the federal government spent to build the entire interstate-highway system.” Tech companies need to keep building these facilities. And they need to do it fast in order to keep up with demand—but also this expectation that these models get better and better.

But data centers are expensive. They take a lot of time to build. And data centers have become this potent symbol for those who are skeptical of AI. They represent a physical incursion of big tech into the communities. Many data centers are loud; some are powered by natural-gas turbines. There are local fears here—about energy use, water use, pollution—and there are national fears about data centers driving up energy prices.

Some of these issues are clear-cut: Data centers are huge, they’re loud, they’re industrial. But other issues, like water use—those are highly contested by people who say that the concern may be overblown.

Like so many political issues, the data-center fight seems to cleave people into two distinct camps. Those who see the buildings as wasteful, polluting, as the engine of a technology they’re anxious about. And those who see data centers as an engine of progress, part of an American infrastructure boom.

This conflict is still incredibly new, and there’s a lot of confusing or bad information out there about data centers. It seems clear that AI is about to collide with electoral politics, both in the midterms and in the 2028 race. What’s the real economic and environmental impact of these buildings? How do the politics of data centers track against left/right party lines? What do people stand to lose and gain when these buildings pop up in our towns?

Jael Holzman is a reporter for the climate website Heatmap and author of its newsletter, The Fight. She’s been covering environmental conflicts in politics like mining, renewable energy, and industrial decarbonization. And for the last few months she’s turned her focus full time to reporting on the data-center backlash and the policy fight therein. Her work has been instructive in demystifying what’s actually going on inside these buildings and across the country. She joins me now to talk about it all.

[Music]

Warzel: Jael, welcome to Galaxy Brain.

Holzman: Thanks for having me.

Warzel: So I want to start extremely basic here. What exactly is a data center? Like how do these buildings, very broadly, work? And why in really the last year, last six months, as a structure, have they become so controversial?

Holzman: Well, data centers aren’t a new thing. Their invention goes back to the early to mid-20th century, the rise of computing in general. Now we’re seeing a rise of data centers specifically because of artificial intelligence, and the sheer amount of compute—computers whizzing and buzzing, et cetera, et cetera—needed to do all of what Claude and ChatGPT enable in our modern society.

Why are people upset about data centers? I’ve been spending a lot of time as a journalist trying to understand that question as of late. I think people have a lot of reasons they say they don’t want these projects. We find that it’s an incredibly bipartisan concern right now. Recent data from Heatmap News’s pro platform has found local opposition exploded in the first quarter of this year to record highs. That’s just registering examples where local data-center fights were showing up in local media or in local-community meeting minutes, things like that.

This is happening not just because of the impacts that people claim, but it’s worth saying that just a broad social change like this is almost necessarily going to invite a huge amount of upset and angst. We’ve seen that with all other sorts of technological innovations. I see a lot of people who are saying they’re upset about data centers, who were upset about vaccines before that, who were upset about masking before that, as well as about renewable energy.

People who were worried about getting cancer from wind turbines, now worried about getting cancer from a data center. Picking apart fact versus fiction, understanding people’s motivations, has never been more important. And we’re honestly still in the beginning phases of getting any of this. Just as new as AI itself.

Warzel: Talking about that backlash, there’s this recent Gallup poll that’s been passed around. That seven in 10 Americans opposing constructing data centers for AI in their local area. Forty-eight percent coming in as strongly opposed. Barely a quarter in favor. What are those fights, since you’ve been covering them—what do they look like on the ground level? Is this response inside of communities’ really loud, boisterous city council meetings? Are we talking about on-the-street protests in these communities? What is the shape of the fight on the ground?

Holzman: So here’s how it looks in so many rivers and dells across the United States right now. A real-estate company or a shell company shows up and says that they would like to develop property either for a data center itself or for a tech campus. And it’s broad, but then it eventually becomes a data center.

Who the ultimate inheritor of that property is going to be—whether it be one of the several but not that many tech giants that are using these facilities or whether it be, you know, an individual magnate who’s going to benefit enormously from the transaction—a lot of that information is actually shielded from the public view. There’s not the kind of years-long disclosure process that you and I are used to seeing in things. Like, They’re going to build a giant wind farm. It’s going to go through a big permitting process, et cetera.

A lot of the time, because this stuff is showing up as real-estate projects first, the public is learning about these projects as they’re being approved, many of the time. And I’m sure companies would argue that they are trying to get out ahead, but I’ve yet to see a lot of instances where that’s really the case. Too often I find it’s a shell company, or it’s a “startup” that showed up and then is ultimately going to give its property to Amazon or to Google or somebody else.

If you’re living around that area, you almost naturally are going to be upset. You’re like: Wait, I didn’t sign up for this. I didn’t have a vote on this. And you’re also not really accustomed to your local leaders having such consequence in your life. We haven’t had a big industrial build-out like this in so long. And so you’re seeing people flood out to local meetings where, before this, people weren’t even paying that much attention to their city council or their county commission. Suddenly they are showing up; they’re learning who those people are, and this is how they get to know them.

You know, I was reading this morning about a someone running for municipal office who left the Republican Party, in a small county in the middle of Kentucky, because he was so upset, and how the political alignment just isn’t responding to the angst over data centers. And so you’re seeing this pushback—these forces directed and vented in new ways that we’re not used to seeing. Recent politics being nationalized, and these fights being nationalized. This is going to have impacts throughout our politics for years to come. And this is only the beginning.

Warzel: I think it’s really interesting to talk about the structure of some of those real-estate transactions. The nondisclosure of it all, or the shell-company part of it.

Holzman: Sure.

Warzel: Because even if it’s not necessarily structured to be nefarious in any way, it feels that way when it comes into your community. If you’re like, Whoa, whoa, what is this? Now you’re telling me this is Amazon? Whereas an Amazon facility, you know, a distribution facility with all the trucks, you kind of do know what it is. And there is that way in which the secrecy behind this is a real factor here that I wasn’t really thinking about.

Holzman: I think also people are just not used to, the ordinary Joe Shmoe is not used to dealing with the energy and tech-development space, right? So on the left, what this reminds me most of is actually the anti-mining movement in the U.S. and where it has sat for the last 10, 15 years. We’ve tried to have a bit of a mining boomlet, in part due to the demand for things like batteries for electric cars and cell phones, mining stuff like lithium, cobalt, graphite. And the progressive left in this country really couldn’t define the evil there. It was kind of running headlong into the climate movement and the push to develop more stuff to decarbonize.

At the same time, there was this push against renewable energy on the very far-right, conservative sect that started to gain steam over the past five years, in part driven by the push for this energy transition away from fossil fuels. You saw this push against solar farms and wind farms on farmland, these concerns about battery-storage technology and potential Chinese control there. Which are baseless, for the most part. Once again, we’re talking about movements. We’re not necessarily talking about facts, right?

Either way, both of these movements failed to really define a political mainstream. They could affect things on a local level. They could affect individual politicians’ conversations on these issues, but they didn’t really have a place for them all to congregate at one time. And when you boil these movements down, they really do represent a left-right horseshoe theory of politics that is far more complex than NIMBYism, but does ultimately wind up in a similar place. Where you’re kind of just arguing against development that is happening in other countries, that other countries are going to do to try to dominate. And it’s not like AI is going away.

The folks who push on this movement oftentimes would argue that the solution is to just not use artificial intelligence, but that rarely works in this country, let alone elsewhere. And so you wind up in this place where there’s ripe, fertile ground in pre-existing political movements on both sides of the political spectrum that do sort of operate in a similar space, which allows for this movement to be so powerful. The time has never been more apt for a very grassroots populist movement against very wealthy tech magnates, energy-industry magnates, with fuel prices going higher. The issue here is like—what is the end goal, right?

Warzel: Right.

Holzman: And this is what I’m trying to best understand. What is the end goal of the anti-data-center movement right now, aside from just banning it indefinitely? And on top of that, where is the movement to get the industry to actually be more socially responsible, to get more socially responsible projects? Less kind of Colossus xAI situations, where you hear about the NAACP suing for environmental racism.

Warzel: And that, by the way, just for listeners, the Colossus is one of the sort of best-known data centers. It’s in Memphis, right?

Holzman: There’s two facilities in the Memphis metro area, and there have been reported issues around them operating gas turbines without proper permits. Alleged in part because Elon Musk and xAI have just kind of publicly moved forward in this way. It’s not exactly like they’re hiding the situation. It’s more of a “move fast and break things” approach to land development and large industrial facilities that Silicon Valley is exporting into rural areas. And now we’re seeing the results.

Warzel: But, just for the record, Musk has also said that the turbines are mobile. He said they’re temporary, and thus they are exempt from the more stringent air permitting. One of the things that I’m seeing is this constant conversation about resources. Can you talk to me a little bit about, like—using water as an example—how this conversation is so polarized?

Holzman: What I find is that when I cover a community that’s upset about water, this is also a community that was already upset about how much water was being used. It’s additive, right? I think the headline becomes “People Hate Data Center Because of Water Use,” when in reality, people hate data center in that case because: Here’s another water user. Right? I don’t think people are just looking at it in a vacuum the way that some folks, the loudest folks on the internet, might seem.

That being said, water is far less of an issue to your day-to-day life in data-center development—as far as what my reporting shows—then the energy impacts, then the noise impacts. I think one of the things we don’t talk about that is one of the most profound impacts of data-center development in a particular community is the noise. And there can be many forms of noise pollution from a data center. Both heard as well as what some people claim is perceived from the vibration of things like large gas turbines or the whizzing of an air-cooling facility. You’re in a position where these projects really do impact communities.

And there are too many examples now like the xAI data center. I visited a project, a Vantage data center in Sterling, Virginia, the VA2 project, where the noise pollution is so profound that when you go there, you feel it in your body. You can’t even hide it. And you feel the vibration, and then you smell the air, and it doesn’t smell right. It smells tainted. There’s something to be said for how there aren’t enough good cases. And as a journalist, I’m looking for those, and I’m excited to tell those.

I hate to bring them up again, but it reminds me too much of the hard-rock mining industry. The hard-rock mining industry has modernized over the past few decades, and there are much cleaner mines than what people today think of when they think of “Guy in hard hat goes and creates open-pit gold mine.” Especially in technologies like lithium. The extraction processes, they do have negative impacts, but it’s not this sprawling, for lack of a better term, colossus of a site, right?

I feel like the data-center industry, and we’re starting to hear people even in VC world explain the data-center sector, the tech companies behind it. There is a need to get better at storytelling, to hire people in community relations who have a background in those communities. You know, the oil industry has said for a long time that they’re able to work so well in places like Texas because they get good land agents. You know, they’ve invested their time and livelihoods for a very long time into getting the land and using the resources.

People are so upset in part because the tech sector is treating data-center development like bug testing. You know, it’s like A-B testing. You’re trying out a community, and you’re gonna see if it works and if it doesn’t, then they go away. And philosophically, that makes sense if you’ve been doing it for a long time. But that’s not how large real-estate projects work. That’s not how large energy projects work in this country. Or at least it hasn’t been for a very long time.

Warzel: What is the flip side here? What is the upshot of a data center coming to your town? Economic benefits, tax breaks? What does a town stand to gain from letting a data center get built here?

Holzman: I think data centers offer a lot to a community in a moment with declining revenues from state and federal governments to fund crucial social services for you and I. There is an argument to be made that the sheer magnitude of tax revenue going into some of these communities could replace cuts to Medicaid, could replace cuts to Head Start. One can easily make an argument that this is where the money is coming from; you should really consider taking that money and rolling with it.

There is a credible argument there. Look at northern Virginia. So much tax revenue goes to northern Virginia now from being Data Center Alley, as some folks colloquially call it. The flip side can be found in Loudoun County [Virginia], or in a state like Alaska. I mean, the corollary here is that if you rely on oil and gas revenue for your state’s budget, eventually you will kind of be reliant on that industry to a point where you’re not able to successfully regulate it effectively to what the people want. And that’s what’s happening in Loudoun County, where within the past couple of weeks, staff for the county have started warning that their budget is on track to be 60 percent data-center revenues. And they’re actually very concerned about how much data-center money is providing a foundation for their living, and recommending against relying on so much data-center money.

But I would argue there is certainly benefit to a tax base in particular. Aside from that, these facilities don’t really do much. It’s not like they’re providing milk or power themselves. One argument that I have heard, though, from this guy Duncan Campbell from DER Taskforce—smart energy wonk that I follow on Twitter—this idea of if they’re going to build on-site power, you could connect them to the grid. And then when they are not fully operating, or if you force them to curtail their power for a bit, all of a sudden you’re building all this new electricity generation, riding on the coattails of the data-center industry. Stuff like that, which I think is pretty novel, could help to make this a win-win situation. But that’s going to require policy makers to do a lot more than they’re doing now.

Warzel: Well, let me ask about the electricity part of this. I feel like you see data centers connected to driving up these electricity costs. And I know it’s complicated. But how do you, if you’re at a bar with someone who’s concerned about this, like how do you calibrate that concern?

Holzman: I think because energy prices are up across the board, and because the economy currently for a lot of people feels like it’s not properly designed to benefit those in the lower and lower-middle class, I think you’re going to just incessantly find people upset about the new thing. Whether it’s a data center or Walmart.

If I’m talking to someone about data centers at a bar, the first thing that comes up is like Jeff Bezos and Bill Gates, not the energy bill. And the reason for that is because data centers make a really, really good symbol for populist politicians to rail against the, for lack of a better term, “Epstein class” or “Bill Gates class.” Or, you know, name your very wealthy controversial person.

Warzel: The oligarchs, yeah.

Holzman: Yeah, the oligarchs. It’s like, Here come these big tech guys—coming in and putting in a giant thing I didn’t want in my community. No one listened to me. I didn’t know about it until later. And there’s secrecy, too. I’m doing a lot of reporting now on the online right wing and how they’ve embraced the anti-data-center movement. Folks like Matt Walsh and Tucker Carlson have even picked up the mantle. And it’s hard not to look at this from a layperson’s perspective as—absent some intervention, absent some very smart political communication and PR from the tech industry, it’s going to be hard to back this “just elites coming in and trying to take over our community” argument for a lot of low-information, laypeople, Joe Schmoes in America.

Warzel: This brings me to how, and we’ve touched on this a little bit already, but how this fight breaks along—because it is sort of united on the left and the right, as we’ve said—but how it breaks along these partisan lines. Can you describe for me a little bit of where the left splinters off on this, and where the right does? You sort of alluded to it with the Matt Walshes and the Tucker Carlsons there for the right. But I’m curious to make that a little more legible for people.

Holzman: Yeah, sure. So what Heatmap News’s data shows is that one of the top predictors, if not the top predictor, of opposition to whether it be a renewable energy project or a data center project is if they voted for Barack Obama for president and then Donald Trump. It is, without a doubt, a fantastic nexus for the left-to-right horseshoe politics that looks a little bit like NIMBYism, is far more complex and intermingled in class-grievance politics, paranoia about surveillance, paranoia about elites.

Where the right and the left converge here is a political coalition that might prove quite powerful in the future. You have, you know, campaigns like what we’re seeing with Graham Platner for Senate in Maine. I recently interviewed the candidate, and he explained to me how terrified he is of AI and the idea that AI data centers are just going to be built out without any regulation, as he put it, in place at all. It’s the same thing that folks like Congresswoman Nancy Mace, who’s running for governor in South Carolina; she tweeted yesterday asking if she should ban data centers in South Carolina for at least a year. You’re seeing this animus picked up on both a far-left and a further-right convergence that, to me, where it leads is the bigger question.

I’m not entirely sure what drives this convergence, except a lot of commonalities that both parties had and may not have realized until right now. And the last thing I’ll say, on that note, is I don’t really know where the constituency is for the pro-data-center movement in the United States. I think that’s yet to be fully determined.

Warzel: You talked about your interview with Graham Platner, presumably the Democratic nominee in Maine. He talks about the AOC/Bernie data-center-moratorium idea. I’m curious about that. This idea of pausing, and where you see that. Because I don’t quite know whether the idea of the moratorium is more symbolic, or is it actually in your mind much more of a political opposition? Of “No, no, no, no, no; we do not want these being built”?

Holzman: It’s a “yes and.” There is a negotiating argument, as the proponents will say, and I’ve spoken to folks throughout the movement for a national moratoria. That it provides leverage in the push to further regulate the sector. That maybe by calling for a moratorium, some people in the middle will go, “Okay, well, we don’t want to ban this, but we should probably regulate it more.” But what’s interesting is that I’m not just seeing people calling to ban it from the left. You know, like I’m starting to see people call for banning it on the further right of the U.S. political spectrum, in the mainstream.

And that, I think, is rooted in people’s concerns about their utility bills. I think that is just, when you get down to it, enough people are afraid of this and want to push pause. Local governments have been enacting moratoria on developing certain kinds of things until they develop zoning ordinances as long as local government has existed. This isn’t exactly a newfangled thing on that scale. The idea of a national AI data-center moratorium sounds ridiculous if you think of the federal government as divorced from that.

But maybe AOC and Bernie Sanders want to turn the federal government into a zoning opportunity. I’m not entirely sure. I think we still need to learn more about where they want this to go. You know, even when you dig into the legislation, like the bill itself on an AI data-center moratorium, it’s not entirely clear on when it would end. You know, it references the need for more study. It references until such regulation is in place. But it’s not really clear-cut. And I think even folks like the Senate candidate, Graham Platner, in our interview, pointed that out to me, saying: “I don’t want a moratorium for the sake of just a moratorium. We need to regulate this industry. We need to make sure it’s actually responsible.”

And it wouldn’t surprise me if we see more voices on the left and the right calling for, you know, “We don’t want a ban just for a ban. We want to actually have smart policy here.” That is, we’ve yet to really get to that discussion, but that’s where I think it’s going to head.

Warzel: A climate of the data-center fight that is very fascinating to me is the idea that for the time being, these data centers, many of them are being powered by decidedly not green-energy means. It seems to me that, broadly speaking, a fear here would be that the proliferation of these data centers is essentially halting any possible clean-energy build-out, right? Wouldn’t that necessarily be something that someone on the right would want to latch on to? To say, “Hey, yes, let’s get more reliant on the natural gas, on this beautiful clean coal.” Where’s the tension there?

Holzman: Well, that’s what the administration is doing. That’s what the Trump administration is doing. That’s what I would call “conservative energy world” is sort of bandying around. At the same time, I talked to folks who are worried about the fact that a lot of the capital that was going to go toward renewable-energy build-out when the Inflation Reduction Act, the country’s first climate law, was passed under Joe Biden is now going toward building these large data centers. At least that’s what they say. The data there is actually a little more mixed. I mean, the data centers were already going to be built out. AI was already coming. So there’s a complicated argument there.

And now the debate within the environmentalist space, within the climate-advocacy space, which is not the same thing, is around: Do you ride this wave and try to make it as clean as possible in the hopes that it can be a gush of demand on the grid that then brings about the kind of rapid energy build-out that folks had long wanted? Or do you oppose this, because it’s not possible to make it cleaner? That there is no amount of regulation that’s really going to happen here that will suffice? And by slowing it down, you delay what could be dirtier in order to make it cleaner, right?

I think that there’s a growing number of people calling for moratoria until there’s a regulation in place that says, “Hey, all of the power needs to be solar, wind, nuclear.” The issue here is I don’t think that the climate movement calling for moratoria, for a ban on data centers, is in any way the same as the anti-renewable-energy people that are also calling for a ban on data centers.

And so my concern as a journalist is that I talk to too many people who are calling for a moratoria who don’t seem to understand what ramifications could come with them getting in bed with a movement that wants to ban solar energy and wind energy for conspiratorial means. I don’t see anyone really reckoning with that kind of game with fire.

Warzel: To complicate the data-center part, though, even more is you have someone like Sam Altman making arguments on podcasts that like, “Yes, unfortunately there’s, you know, gas and turbines powering these things. That’s not ideal. But if we can get enough compute to train the models to be good enough and to be powerful enough, perhaps they will help us, you know, with perfectly clean fusion technology.” Is anyone buying that argument?

Holzman: When I was talking to someone this morning for my newsletter, we were struck, both of us, in how we agree there’s not really anyone leading right now through this. Ordinarily, I think with an industry this insurgent, you would expect someone like the U.S. president to be kind of guiding the country through this. But it feels like the administration is so supportive of AI data centers. It’s a “let a thousand flowers bloom” situation where, yeah, there might be a couple fights. Also by approaching it in a deregulatory way, maybe you can just have such a booming economy that people aren’t really complaining as much about that. I don’t know how that’s going to work in an era with higher fuel prices, but we are where we are.

And so I think someone’s going to need to step in, either through a lower political level or through a race for the presidency or through industry vectors or maybe even new media to be kind of a leading voice on this issue. It feels like we don’t really have one. We have a chorus of pro-AI voices who say a lot of things that ordinary people, because there’s so much suspicion about technology, just don’t really take on face value. And then we have an environmental constituency, an activism constituency, that a lot of people also see as one-sided and motivated in part by specific aims. So I feel like we’re still in this open field with a lot at stake, but no one really driving the car through the field—if this metaphor is going to end with us in a ditch in a field.

Warzel: We’re somewhere in a field. That’s all we know.

Holzman: We’re somewhere in a ditch in a field.

Warzel: I’m curious how much of this you feel is totally genuine. Or if there’s an element of the tails wagging the dog here, right? Of media coverage about data centers then creates more media coverage, which leads to more ambient anxiety and anger, and this politics and this movement. And I’m curious, when you see it—going into these communities and reporting—how much of this moment right now feels entirely grassroots, versus a product of the cycle that I’m describing?

Holzman: I do not in any way wish to diminish the many people who just learned about this large thing coming to their community, and then told everyone in their community on Facebook. That does seem to be quite often the case. What I do think is happening now is that there are digital media outlets, organizations by the names of More Perfect Union, for example, that do get a lot of internet engagement off of telling people stories of conflicts around artificial-intelligence data centers.

What I also know is that when people talk about this stuff on social media, it gets a lot of traction. And so you have these authentic moments of virality, and then other figures do come in and then see that and decide to start talking about that too. The thing is, your average person wasn’t paying attention to a local government meeting before that data-center conversation came in. And I still don’t know how an online din is leading to people showing up in their city-council hearing.

I’ve spoken to folks in places like rural Pennsylvania who worked in health care and then suddenly found themselves making videos where they exposed the inner conversations between data-center developers and the state government. These people do exist, and they’re not being bankrolled. I think it’s just the internet conversation is forming its own kind of Ouroboros. Like I think Twitter is eating its own tail on data centers. And then in the meantime, with that din going on, there are just real fights happening in so many communities around this issue, that does bind so many motivations on not just the left but also on the right.

Warzel: I find myself wondering whether the anger and the backlash at all of this, and all of that, what we’re talking about, the algorithmic salience of all of this stuff. I wonder if it’s correctly calibrated, or whether people are mad at the wrong thing. Or whether any of that really matters, because, you know, there’s this low-grade concern about AI that people have everywhere, right? My kids are using it in school. Or whatever. My boss is making me use it. I love it. I hate it.

Holzman: Yeah. And that’s driving some of this too, right? I mean, some of this is just people upset at data centers because they know the more data centers, the more AI compute.

Warzel: Right. So you talked about them as this symbol. I have been thinking of them as a national pain sponge, right? They just represent whatever fears or concerns you have, and you can direct it at the project. But part of me wonders if there isn’t just the bigger, like the omni-fear—which is just the idea that there are going to be winners and losers in the AI boom. And there has been very little messaging and very little indication from these big AI firms and the culture that regular, average Joe Schmo is going to be the winner.

Holzman: Yeah.

Warzel: And so what all of this really is, is just this reaction to who stands to profit, right? Who stands to consolidate power.

Holzman: The conversation around data centers is easily a pain sponge. But you gotta wonder where that water is coming from and what that sponge really is made of, because what that sponge is made of is so many local conflicts stitched together. And it feels wrong to lose sight of the real concern, the very real fear, that your average farmer in rural Pennsylvania and rural Ohio is having. And I think dealing with that is something that’s generational. Like, this is going to be a generational fight over how we even reckon with the politics around data centers.

Warzel: Do you feel that the coming national-politics part of this is going to distort what this whole fight is about? And muddy the waters for the generational fight that is to come?

Holzman: Well, before we talk about the generational fight—and the politics that are to come in the midterms and in the 2028 presidential race around AI data centers—it’s also worth noting how much money the AI sector is spending on elections right now. And how much of an impact they’re trying to have on not just federal races, but state races, even local races. I think you’re going to see an even greater role played in 2028 than in this election cycle.

And I also think that you won’t always find the critics of the AI data-center sector winning out, because I don’t necessarily see the far right having a different view than the far left on AI data centers right now. I think both poles really want this to stop. It’s just for entirely different reasons, which I think those sides are going to need to sort out and figure out how they feel about that.

What’s going to happen, I think, is you’re going to have this very techno-optimist middle rise up. And it may be such a large tent that it brings in enough voters to be a durable political coalition. And that techno-optimist future goes: “No, let’s not ban data centers. Let’s figure out a way to build a better society with it. Let’s figure out how to take that tax revenue, put it toward transmission-infrastructure upgrades, wires. And then bring the data-center power back onto the grid. And then all of sudden, look, we have cheaper power—and I can cut your taxes now, because the data center helped with that.” I think you’re going to have that push.

And the question is whether or not, in this very populist political environment, you see this middle win out. Or if those two diametric poles that are very, very angry at this elite class and a infrastructure build-out that they don’t really think they’re going to benefit from at all, whether or not those poles are actually the more powerful ones. I think that because the Obama-Trump coalition is such a good predictor and touchstone for the anti-AI-data-center angst, it might be that the same forces that elected Donald Trump are the forces behind this backlash.

And it’s yet to be determined, but I feel like that’s honestly more likely than the left being the anti-data-center camp and the right being the pro-data-center camp. So, I mean, I think in the future, it may be that there’s actually a race for president in 2028 where both candidates of both parties are criticizing data centers, and trying to figure out who is the most anti-data-center candidate out there. That feels like more likely than, you know, a polarization over whether or not to develop them.

Warzel: Hearing you say that, I think it’s a very safe bet to assume that the politics are going to be fractured and weird and potentially incoherent in ways, and very coherent in other ways. I think that’s like a very good bet. It’s going to be really fascinating to watch this play out. Hopefully we can have you on again to talk about it as it goes. But Jael, thank you so much for coming on and demystifying the next great local/national fight.

Holzman: Thanks for having me on, Charlie.

[Music]

Warzel: That’s it for us here. Thank you again to my guest, Jael Holzman. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe on The Atlantic’s YouTube channel, or on Apple or Spotify or wherever it is that you get your podcasts. And if you want to support this work and the work of my fellow colleagues, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

This episode of Galaxy Brain was produced by Renee Klahr and engineered by Miguel Carrascal. Our theme is by Rob Smierciak. Claudine Ebeid is the executive producer of Atlantic audio, and Andrea Valdez is our managing editor.

AI Slop Is Coming for Your Playlists

2026-05-29T12:46:00-04:00

This is an edition of The Atlantic Daily, a newsletter that guides you through the biggest stories of the day, helps you discover new ideas, and recommends the best in culture. Sign up for it here.

Late last month, a swarm of songs with near-identical names, lyrics, and melodies started to go viral on streaming platforms across the world. These tracks were not exactly the same—some have a little more guitar than others, some are more dance-oriented—but they’re all named something close to “Angel Above Me” or “Run Run River,” after the song’s first line. They’ve accrued millions of streams on Spotify and TikTok, and versions have hit No. 1 on iTunes in Germany and Austria.

Many of them appear to have been generated by AI. It turns out that they’re based on a human-made song, “Angels Above Me,” which was released in 2019 by the reggae band Stick Figure. That track has enjoyed a streaming bump in recent weeks too—but many of the people listening to the new remixes may not even know about the original, because the song’s actual co-writers aren’t always credited. AI music has gone viral before and charted before, but song generators are now good and fast enough that they can flood the zone, creating tracks that slip past the safeguards of major streaming platforms and distributors. Spam-filtering systems can do only so much to stem the flow; according to data from the analytics firm Luminate, 106,000 songs (both AI-generated and not) were uploaded to streamers and other platforms every day in 2025.

Musicians copy one another for legitimate reasons all the time. Parody law, as the comedy albums of Weird Al Yankovic remind us, is expansive; you can often get away with ripping off melodies and lyrics as long as it’s clear that you’re mocking them. Covering another artist’s song is also legal, as long as you get the right license before you do so. The same goes for sampling—knitting together different musical clips to create something new—and interpolating melodies. When Ariana Grande repurposed a tune from The Sound of Music on her song “7 Rings,” she ceded 90 percent of songwriting royalties to Rodgers and Hammerstein. The process of vetting your song in some way isn’t always simple or cheap, but it ensures that the original artists are getting paid as appropriate.

AI remixes exist in a legal gray area. It took me all of 30 seconds to generate an AI clone of Kendrick Lamar’s voice for a “Not Like Us”–style diss track against the color blue. And it was just as easy to spin up an AI version of an Elliott Smith song featuring the exact lyrics of the original. Although I chose not to upload these to any streaming platforms for obvious legal and ethical reasons, I almost certainly could have: DIY distribution programs will push any song to Spotify, YouTube, Apple Music, or others for a small fee, and not all of them are equipped to vet each upload for potential copyright violations. That makes it all the more important for streamers to ensure that AI-generated music is both legal and properly labeled, and that a portion of the money flows to the original creators.

The problem isn’t AI music itself. Many of the songs that AI generates could be legal, if they are distinct enough from human musicians’ published work to avoid copyright trip wires. And some artists, such as Timbaland, Kanye West, and Diplo, are now openly using it as part of their own creative processes. Few data are available about the total number of purely AI-generated songs on streaming platforms, but it’s clearly enough to spook creators. Takedown requests through the Digital Millennium Copyright Act—one of the main tools for fighting intellectual-property infringement—offer only a piecemeal remedy, striking songs one by one rather than banning unauthorized material wholesale.

A representative for Suno, a leading AI-music-generation platform, told me that the company uses filtering technology to try to prevent unauthorized uses of artists’ preexisting songs, but she also said that the company didn’t know whether the unauthorized Stick Figure remixes had been created with their tools. A spokesperson for Spotify told me that “in the past year alone,” the company has “removed over 75 million spammy tracks from the platform” and “introduced a suite of new policies” regarding AI. She also wrote that “for any manipulated streams on Spotify, we remove those streams from play counts and withhold royalties.”

Companies take different approaches to the problem of labeling AI-generated content, but in recent months, a consensus has emerged that it might just be easier to verify human-generated content instead. Adam Mosseri, the head of Instagram (another platform that’s rife with AI-generated material), wrote in December that as AI gets better at imitating reality, “it will be more practical to fingerprint real media than fake media.” OpenAI CEO Sam Altman is backing a start-up that scans people’s eyeballs in order to provide “proof” of humanness. And a few weeks ago, Spotify began rolling out a verification-badge system for artists who meet their “authenticity” criteria. For now, the system largely excludes purely-AI artists—although “artists who use AI tools responsibly” are eligible to be verified.

These companies will need to fight the incoming wave of AI slop even as they lean into AI as a product. Last Friday, Universal Music Group announced a licensing deal with TikTok that promises to expand protections against AI music. But a day before, UMG had announced a partnership with Spotify that allows users to create AI-assisted remixes of certain songs. Spotify’s co-CEO Alex Noström described it as “grounded in consent, credit, and compensation for the artists and songwriters that take part.”

Stick Figure’s Spotify page is now verified, but whether this will actually redirect listeners toward its music is unclear. The fuzzy line between human- and AI-generated material poses a particular problem in the streaming industry, which has long encouraged passive listening habits. As the writer Liz Pelly lays out in her book, Mood Machine, one of the ways Spotify makes money is by encouraging users to listen to music at all hours of the day; hence the hundreds of “chill” playlists designed to soundtrack our lives without intruding into the foreground.

By framing music as sonic wallpaper, Spotify and other streamers have effectively set the stage for today’s confusion. We’ve been trained not to think all that hard about what we’re listening to. A recent study from Luminate found that people are growing less interested in AI-generated music—but what if they can’t tell they’re listening to AI-generated music in the first place? Plus, when people hear songs on streaming platforms or in social-media videos, they’re not typically thinking about the nuances of copyright law and royalty payments (Swifties are the possible exception to this rule). When AI enables unauthorized remixes that streamers don’t catch, human musicians and writers lose out. It’s not some coming danger; it’s already happening. And listeners may not even notice.

Related:

A Book I Wish I’d Read at 22

2026-05-29T11:30:00-04:00

This is an edition of the Books Briefing, our editors’ weekly guide to the best in books. Sign up for it here.

The earliest years of adulthood—ages 18 to 22, give or take—are deeply formative. Some people spend that period in higher education, where they’re encouraged to read broadly and think deeply about their path forward. But I think anyone in that phase of life can take inspiration from the list of books Anna Holmes named this week as great for graduates. “‘Figuring things out’ is a lifelong endeavor with no guarantee of success,” she writes. But “the best way to locate inspiration is by looking to writers who illustrate what you might want to emulate, rather than those who lead by edict or exhortation.” Her choices are surprising and wise, full of picks I’d pass along to many of the younger people in my life. Around this time last year, I wrote about a book that was important to me in my transition from undergraduate to grown-up. But this year, Holmes’s suggestions brought to mind a different Atlantic list—the books we found too late.

First, here are four new stories from The Atlantic’s Books section:

For that article, I recommended Vigdis Hjorth’s A House in Norway, which I had been assigned in a Norwegian-literature class I was taking in Oslo. That book—and that course—made me feel that I’d lost precious time I should have spent learning Norwegian, my mother’s native language, and getting to know the country she’s from. (My anxiety about wasted years is hilarious now, considering I was only 20 when I read it.) Judging by my colleagues’ recommendations on that same list, I’m not alone in feeling regret over not having read the right book at just the right time. And as I’ve gotten older, I’ve experienced these kinds of missed connections again and again.

About a month ago, for example, I was standing with a friend in the packed aisles of the Strand, New York’s famous, cavernous independent bookstore. We’d examined some rare, leather-bound Anthony Trollope novels; I’d tried searching for Garry Wills, before the crowd jostled me away from the biography section. After regrouping in General Fiction, we decided to head for the exit and on to dinner. Then I saw Sarah Waters’s Tipping the Velvet on a table. The American cover of the British novel is more titillating than it needs to be—two women wear nothing but stockings, sharing some kind of swing or trapeze—but it’s a striking image, and I half-remembered the book from years of word-of-mouth recommendations and Wikipedia sessions. On the strength of that vague recollection, I bought it.

If only I had sought it out sooner! I tore through the novel on the subway, then on the Amtrak back to Washington, D.C., and I even stealthily turned pages during conversations. Set in the 1890s, it takes the form of a Victorian picaresque. The book’s hero is a young girl, Nan, who falls in love with a male impersonator named Kitty; their relationship (and messy breakup) propels Nan into London’s queer underworld. She tries on a set of different identities—stage performer, sex worker, kept woman, socialist agitator—but all the while, she’s searching for a place, and for people, who feel like home. I have not had more fun reading in ages. Yet as I finished it, I felt a moment of sadness—in part because I was no longer Nan’s age. I’ve already done much of the self-discovery and self-definition she’s undertaking in the story. But if I’d read it in my early 20s, when I was also figuring out whom I loved, how I wanted to look, and who I wanted to be, I might have treasured it even more.

Illustration by Alisa Gao / The Atlantic

Read These Books by the Time You Graduate

By Anna Holmes

“Figuring things out” is a lifelong endeavor, but these titles offer inspiration for young adults finding their way. Read the full article.

What to Read

A Visit From the Goon Squad, by Jennifer Egan

Egan’s rightly lauded collection of linked stories found its way into my hands just as I was crawling out of a midlife mess in which I was making a lot of questionable choices. The book drops in on a highly populated world revolving around the music business, and for obvious reasons, I found myself drawn to the endearingly disastrous producer’s assistant Sasha. Paradoxically, her story gave me a tremendous sense of hope that, regardless of my mistakes in the moment, everything would be okay in the end. We first meet her as a 20-something living in New York who steals a wallet while on a date. We see her teenage years as a runaway sex worker in Europe, watch her as a misanthropic college student, and ultimately glimpse her as a content and loving mother, living in California and channeling her love of music and curiosity into her children as well as artwork of her own. Sasha’s life, like mine—and like all of ours—is full of low moments, but while those times shape us, they don’t need to define us. — Xochitl Gonzalez

From our list: Seven books for people figuring out their next move

Out Next Week

📚 Whistler, by Ann Patchett

📚 The Typing Lady, by Ruth Ozeki

📚 Something We Said: Richard Pryor, a Notorious Word, and Me, by Elizabeth Stordeur Pryor

Your Weekend Read

Illustration by James Lee Chiahan

The Night My Marriage Fell Apart

By Chris Jones

I snapped like I never had before, swinging my rake as hard as I could against our fence, breaking both, everything in splinters. I stood in my yard, still surrounded by leaves, and now with a fence to repair and half a rake in my fist. I flung it away and got into my battered little pickup to drive to the hardware store. Two blocks from home, I made up my mind that I didn’t like Amy very much anymore. Another couple of blocks, and I realized that she must have come to the same conclusion about me, a little sooner than I’d arrived at mine.

Read the full article.

When you buy a book using a link in this newsletter, we receive a commission. Thank you for supporting The Atlantic.

Sign up for The Wonder Reader, a Saturday newsletter in which our editors recommend stories to spark your curiosity and fill you with delight.

Explore all of our newsletters.

Photos of the Week: Memorial Day, Spelling Bee, Cheese Rolling

2026-05-29T09:00:00-04:00

Jose Perez / Bauer-Griffin / Getty

Olivia Rodrigo is seen filming a music video in Manhattan, in New York City, on May 26, 2026.

Christopher Furlong / Getty

Competitors take to the floor during the Blackpool Dance Festival at the Empress Ballroom in Blackpool Winter Gardens on May 26, 2026, in Blackpool, England. The festival is celebrating 100 years of competition in what organizers say is the world’s largest and longest-running dance event.

Jose Luis Magana / AP

Louis Avetis of Orlando, Florida, jumps after spelling his word correctly during the quarterfinals of the Scripps National Spelling Bee at DAR Constitution Hall, in Washington, D.C., on May 27, 2026.

Muhammad Reza / Anadolu / Getty

Muslims gather to perform Eid al-Adha prayers in the Eidgah Sharif neighborhood in Rawalpindi, Pakistan, on May 27, 2026.

Zain Jaafer / AFP / Getty

Muslims perform the evening prayer around the Kaaba, Islam’s holiest shrine, at Grand Mosque complex in Saudi Arabia, on May 24, 2026.

Dan Kitwood / Getty

Reflections of sunlight are seen as a northern gannet flies over the sea at the Royal Society for the Protection of Birds’ Bempton Cliffs reserve on the East Yorkshire coast, on May 26, 2026, in Bridlington, England.

Adam Gray / Getty

An ICE agent sprays chemical irritants at protesters and media outside the federal immigration center at Delaney Hall, where ICE is housing detained immigrants on May 27, 2026, in Newark, New Jersey. Ongoing protests, which became tense over the holiday weekend, come amid reports of a hunger strike by detainees.

Fayaz Aziz / Reuters

A blacksmith sharpens a sacrificial cleaver ahead of Eid al-Adha celebrations in Peshawar, Pakistan, on May 26,2026.

Aaron Favila / AP

Residents use buckets of water as they try to control a fire that hit a poor community in Manila, Philippines, on May 23, 2026.

Ahsan Mohammed Ahmed Ahmed / Anadolu / Getty

A view of Bekhal Waterfalls in Rawanduz Canyon, situated in Rawanduz district, Erbil Governorate, in Iraq’s Kurdistan Region, seen on May 21, 2026

Remko De Waal / ANP / AFP / Getty

This aerial photograph shows people being pulled on an inflatable raft by a motorboat during a heat wave, on Gooimeer, a lake near Muiden, Netherlands, on May 25, 2026. Temperatures hit record highs for May as forecasters warned of a prolonged period of extreme heat across Europe throughout the week.

Brook Mitchell / AFP / Getty

Moon, a two-year-old Bengal rescue cat, poses for a picture while being taken out for exercise by his owner, at sunrise at Pen Ponds, in London’s Richmond Park, on May 24, 2026.

Sebnem Coskun / Anadolu / Getty

A view of sunset illuminating the Pyramid of Khafre and the Great Sphinx of Giza, on the outskirts of Cairo, Egypt, on May 25, 2026

Ulet Ifansasti / Getty

Indonesian Muslims perform Eid al-Adha prayers at Butuh village in Wonosobo, Central Java, Indonesia, on May 27, 2026.

Eric Lalmand / Belga / AFP / Getty

A man dressed as a member of the French Imperial Guard stands guard at the foot of the Lion’s Mound monument, in Waterloo, after its renovation, in Braine-l’Alleud, Belgium, on May 22, 2026.

Anthony Upton / AP

Participants compete in the men’s-downhill-race category of the annual Cheese Rolling contest at Cooper’s Hill, in Brockworth, Gloucestershire, England, on May 25, 2026.

Fran Santiago / Getty

The Virgin of El Rocío is carried by devotees during a procession on May 25, 2026, in El Rocío, Spain.

Tim Phillis / AP

Members of the New York Knicks hold the Eastern Conference Championship trophy after Game 4 in the Eastern Conference–finals NBA-basketball-playoffs series against the Cleveland Cavaliers, in Cleveland, Ohio, on May 25, 2026.

Heather Diehl / Getty

U.S. Naval Academy graduates celebrate with a cap toss during the 2026 Naval Academy Graduation and Commissioning Ceremony, on May 22, 2026, in Annapolis, Maryland.

Samuel Corum / Getty

A mourner kneels at the headstone of a fallen loved one in Section 60 at Arlington National Cemetery on Memorial Day, on May 25, 2026, in Arlington, Virginia.

Jorge Guerrero / AFP / Getty

A pilgrim lights candles in the village of El Rocío, Spain, on May 25, 2026.

Axel Schmidt / Reuters

A participant of the Wave Gotik Treffen gothic festival poses in a park in Leipzig, Germany, on May 22, 2026.

Andreea Alexandru / AP

Rakele Menjivar poses for photographers upon arrival at the premiere of the film Bitter Christmas at the 79th international Cannes Film Festival, in Southern France, on May 19, 2026.

Adriano Machado / Reuters

The Indigenous leader Megaron Txucarramãe poses for a picture during an interview with Reuters in the village of Pykany, in the Menkragnoti Indigenous Territory of the Kayapo people, Para state, Brazil, on May 20, 2026. Megaron is Chief Raoni’s successor and is preparing to continue his uncle’s legacy, in addition to fighting to protect Indigenous people’s rights and ensure the demarcation of his people’s land.

Kim Hong-Ji / Reuters

A humanoid robot and a model present creations during Galaxy Corporation’s Mach33: Physical AI Fashion Show, in Seoul, South Korea, on May 28, 2026.

Sean Rayford / Getty

A member of the Them Cowgirls drill team rides in the arena during the 29th annual Black Cowboy Festival at Greenfield Farm, in Rembert, South Carolina, on May 23, 2026. The weekend event included agriculture and heritage exhibits, trail rides, dance classes, rodeo events, and live music.

Finnbarr Webster / Getty

Sheep are judged at the Royal Bath & West Show, in Shepton Mallet, England, on May 28, 2026. The historic show is one of the oldest surviving agricultural shows in England—the first show took place in Taunton in 1852 and then toured the country for more than 100 years, before a permanent home was found at Shepton Mallet in 1965.

Ian Forsyth / Getty

Oliver Alderson from Harrogate sits at the wheel of a vintage tractor during the Pickering Traction Engine Rally, in Pickering, England, on May 24, 2026.

Becoming a Perfect Wife, by Any Means Necessary

2026-05-29T07:30:00-04:00

Over the course of six seasons, the Hulu series The Handmaid’s Tale became known for its brutality. Women who revealed any hint of rebellion against their oppressors, including the government officials to whom some were forcibly betrothed, lost their eyes, their tongue, and sometimes their life. The image of red-cloaked women bowing their head to the ground is used as a blunt visual shorthand for female oppression in what the series depicts as a dystopian, totalitarian America.

The Handmaid’s Tale focuses on vicious persecution. Hulu’s sequel, The Testaments—which, like its predecessor, is based on a Margaret Atwood novel—examines a subtler tool for discipline: aspiration. The series, set four years after the events in The Handmaid’s Tale, unfolds in the authoritarian state that replaced the United States, which has been renamed Gilead. It follows a young, impressionable group of characters: the daughters of the ruling aristocracy. This group of teen girls is referred to as “Plums”; they are nubile and always immaculately dressed, and hope to become perfect wives to the nation’s most powerful men. To that end, they undergo strict training in how to be prim, proper, and hyperfeminine.

Gilead convinces its youth that this system is not just necessary but ideal. Here, The Testaments offers a shrewd observation: It’s easier to control people to whom subjugation seems desirable. The illusion of desirability does not hold for long, however. The more viewers see of the girls’ training—which toggles between simple and coercive, even violent—the clearer it becomes that Gilead’s messaging obscures a sadistic reality.

[Read: Slouching toward Gilead]

The Testaments unfolds amid a worldwide fertility crisis, which Gilead’s leaders have used as justification to establish a new, ultraconservative government order that strips women of their agency. Reproduction is a core value, and Gilead’s enforcers believe that their country’s survival depends on deploying a kind of soft power to influence its teen girls and convince them that homemaking is the only path to a life of godly bliss. The regime also resorts to more savage tactics: Fertile women must act as surrogates for wealthy couples struggling to get pregnant, and endure ritualized sexual assaults each month until they conceive.

The nation’s patriarchal values are baked into its school curriculum. Instead of English, science, and math, the Plums learn subjects such as embroidery, culinary arts, and scripture. Each girl seems to share the same desire: to marry a man who can provide stability and security, so that she worries only about running the home.

As part of their education, the Plums host elaborate functions for their peers and superiors—events that are, in fact, high-stakes sessions meant to rehearse the girls’ married lives, when they’ll be expected to throw party after party to proselytize about the joys of domesticity. During one crucial mixer, the girls offer carefully brewed beverages and fancy sweets to a roomful of matriarchs and instructors, all of whom have gathered to judge the Plums’ taste and poise. The adult women will decide which of the students has proved herself worthy of a high-status husband; thus the shortbread and tea cakes are conspicuously fussy, and reflect the obsessiveness that Gilead’s elite demands of its housewives. When one of the Plums trips on a rug while serving tea, she breaks into tears, convinced that her future could be ruined. Moments like these remind viewers that for all of their competitiveness and one-upmanship, the Plums are really just scared little girls.

The parties also test the Plums’ management skills. Although the spotlight is on the aspiring housewives, the show makes clear that each one has a household staff quietly handling the grunt work. Despite their anxiety over pastries, the girls don’t even bake the sweets—their servants do. And the stress takes a toll not only on the teens but also on their hired help—as when one student berates her servant for improperly preparing the tea.

It’s difficult to watch The Testaments without thinking about another group of women known for their performative hyper-domesticity: “tradwife” influencers, known for cultivating an internet persona based on their supposedly covetable, quaint home life. The show’s focus on the illusion of authenticity may recall, for some viewers, the real content creators who have been accused of disingenuousness about their lifestyle—by leaning on a robust staff, obscuring the use of prepackaged ingredients even as they claim to painstakingly make everything from scratch, or pretending to be someone they’re not. (Tradwives do differ from the Plums in one respect: Many rake in astronomical amounts of money through brand deals.)

Although the Plums enjoy a level of privilege, they clearly struggle to adhere to the standards that Gilead has set for them. And when they stumble, the regime’s harshest tactics bubble to the surface. If a wife-in-training breaks a rule (by, say, cursing at or hitting another girl), the administrators demand that her classmates participate in her shaming and corporal punishment. The message is clear: Anything less than perfection results in more pain. Still, fear is only one-half of the formula. Gilead needs its high-status women to want to live this way. When the wealthiest young girls embrace the regime’s customs, they create an aspirational reflection of its principles—inspiring others to want the same.

[Read: What happens when the tradwife dream goes wrong?]

This is why the Plums’ day-to-day lives are full of social gatherings, designed to illustrate the emotional and—more important—material perks of marriage. At one point, a married former classmate invites several girls over for a home tour that may seem familiar to any viewer who has watched an influencer show off her airy, well-appointed house. Although the Plums fantasize about finding husbands, they spend just as much time imagining all of the luxurious accessories that will soon follow: The bluntest girl among them says that she wants triplets because when one of their peers had twins, “her husband bought her a Mercedes and diamond earrings.”

Relying on marriage for security leaves these women’s fates entirely in their husband’s hands—and like The Handmaid’s Tale before it, The Testaments is not optimistic about how the men will use that power. In the series, a dentist abuses his young patients; fathers barely know their daughters; old men make suggestive jokes about the teenagers they’re trying to marry. By the end, one of the Plums has committed murder after having a world-shattering realization.

A lesser show might have taken a sneering approach, casting judgment on Gilead’s aristocratic wives for the Faustian bargain they’ve made. Instead, it evokes empathy, and in doing so lands a more nuanced point. Whenever someone’s life begins to look like an advertisement, it’s worth asking what, precisely, the product is—and if it will really benefit the buyer.

Trump Takes His Cut

2026-05-29T07:00:00-04:00

When it comes to the government’s relationship with business, Donald Trump is the most activist president since Franklin D. Roosevelt. He has wielded tariffs, the government’s purchasing power, and the threat of regulatory action to bend companies to his will. Over the past year, the president has even made the federal government a corporate shareholder across a range of industries. Just last week, the Commerce Department announced that it was taking stakes in a portfolio of quantum-computing companies in exchange for $2 billion in investment.

The federal government’s bold entrance into the world of corporate investing began last June, when Trump agreed to allow Japan’s Nippon Steel to buy U.S. Steel as long as the combined company granted the United States veto power over certain management decisions and ensured that key leadership roles went to American citizens. In the months since, the administration has taken sizable stakes in rare-earth firms and a new rocket company. Most dramatically, the administration insisted last year that the semiconductor giant Intel hand over nearly 10 percent of the company in exchange for $8.9 billion in grants that had already been earmarked in the CHIPS Act and other government awards but not yet paid. Trump now likes to brag that Intel’s stock has since surged by 300 percent.

This isn’t the first time that the federal government has taken equity in public companies. But in the past, it happened almost exclusively in times of crisis, as in the bailout of U.S. automakers and Wall Street banks in 2008, and the government promptly divested itself of the stock once the turmoil had passed. Trump, by contrast, is making government ownership commonplace, which means any company that accepts government funding now has to wonder if it’ll have to give up shares as a result. That kind of government investment in the fate of individual companies opens the door to a type of crony capitalism the United States has historically done a good job of avoiding.

[David A. Graham: Trump’s right-wing socialism]

The argument for Trump’s moves is easy to understand. If the federal government is going to subsidize the research and development of various companies, as it already does via loans and direct grants, then surely it should reap some of the benefits too—or “enhance the return for the U.S. taxpayer,” as the Commerce Department put it when announcing its investments in those quantum-computing companies. Trump isn’t alone in feeling this way. Some progressive Democrats, including Senator Elizabeth Warren, had already suggested that the government take a stake in companies awarded CHIPS Act grants. Senator Bernie Sanders, a democratic socialist, backed the administration’s plans to buy a stake in Intel, explaining, “If microchip companies make a profit from the generous grants they receive from the federal government, the taxpayers of America have a right to a reasonable return on that investment.”

But bipartisan support for these moves doesn’t make them right. Part of the problem is that the administration has not made it clear how these investments work, or even how they are being made. No law authorizes the Commerce Department or any executive-branch agency to take equity stakes in companies, which means there’s no legal protocol for how or when the government might sell those stakes, or where the proceeds from such a sale would end up. Nor has there been any transparency into how the government is picking these companies in the first place.

The administration can argue that it has targeted sectors connected to national security: steel, rare earth metals, defense contracting, semiconductors. But when Spirit Airlines recently tried to avoid liquidation, Trump talked openly about buying it, even though the airline’s fate is hardly a matter of national security. In practice, then, these decisions may be subject to Trump’s whims. Although all government spending is meant to be allocated by Congress, the president has made it plain that he believes the power of the purse is his.

But even a more rigorous and transparent process would not alleviate the fundamental problems of government ownership. A big federal stake in a company will invariably tilt the marketplace, which creates incentives for the state to tilt things further, perhaps via regulatory relief, exemptions from tariffs, or just plain presidential meddling. Apple, for instance, recently resumed using Intel chips in some of its products after Trump lobbied it to do so. Apple certainly could have done this regardless, as part of the company’s move to diversify its chip supply. But Trump’s willingness to use government power to help companies that support him and hurt those that oppose him means that for many companies it’s economically rational to simply accede to his wishes—even if the move would not be economically optimal in an actual free market.

[Read: What happens if Trump seizes AI companies]

Companies are coming to see these transactions as a cost of doing business. Trading away shares can be a way to gain favorable treatment, or at least avoid getting punished. Take the bidding war earlier this year between Netflix and Paramount for Warner Bros. Discovery, which Paramount won, contingent on regulatory approval. One can easily imagine Trump demanding a stake in the newly merged company in exchange for a blessing from the Federal Communications Commission. Paramount would be hard-pressed to say no.

When the government supports American companies with loans and grants, these companies can keep the administration at arm’s length. When the government takes a stake, the partnership becomes intimate and long-term. The upsides of this arrangement are greater, but so too are the costs. We want regulatory decisions and government policy to be shaped by what’s best for the economy, not by what’s best for the companies that the government happens to own. Federal stakes in public companies may be a good deal for the government, but they’re still bad for America.

The Biggest Tell That Something Was Written by AI

2026-05-29T07:00:00-04:00

A few weeks ago, where I live in Johannesburg, a man ran a stop sign and crashed into my Subaru. At the scene he was frantic, unable to gather his thoughts. Half an hour later, I received a lengthy, perfectly grammatical text from him elegantly explaining how he perceived the crash had happened. For a repair quote, I wrote to a mechanic I know, a man who used to text me in curt phrases riddled with shorthand. I got a response using just the same voice as the man who’d crashed into me—the distinctive voice of AI.

In surveys, people consistently say they distrust AI-generated writing. But that hasn’t stopped more and more of us from using it in everyday life—to compose work emails and personal texts, to make shopping lists, even to write scripts for arguments with our spouses. “I feel like I’m going nuts,” the writer Jason Koebler complained in the tech outlet 404 Media, under “the cognitive load” of trying to discern whether every piece of text he reads is real or fake.

AI writing is also creeping into our most elite literary spaces—newspapers’ opinion sections, books, literary magazines. I edit professionally, often working with authors renowned for their prose. Maybe two months ago, I began receiving a kind of submission I’d never gotten before: perfectly clean, without a stray comma; uniform in length, with evenly paced paragraphs and a distinctive tone that was simultaneously breezy and grandiose. At first I was surprised that people who prided themselves on their writing would turn to AI to write for them. Even six months ago, when I occasionally identified a paragraph in a writer’s work that seemed AI-generated, they would apologize.

Now some authors tell me they’ve embraced AI as a “writing tool,” no different from spell-check or a laptop. The phrase is protean and euphemistic, covering everything from using ChatGPT to find a quote to having it compose a long essay based on a two-sentence prompt. The reason for the change is simple: Competition in journalism and academia and grant writing and even YouTube influencing is insanely fierce. The edge goes to those who can stand out in a deluge of content, which is achieved through cleanly packaged messaging and sheer volume. Even professional communicators who are confident in their writing and unsure that AI is a perfect replacement are under increasing pressure to use it, so long as they feel they’re doing so within their profession’s boundaries. (The Atlantic, for the record, prohibits writers from using AI-generated text unless it’s explicitly identified as such.)

People who aren’t professional writers are making a similar calculation. AI programs’ efficiency in generating smooth, grammatical text is irresistible, whether you need a savvy sentence in a job application or a line of banter on a dating app. AI-generated writing can easily trick readers, especially if they’re only skimming. Tutorials exist for how to strip the telltale signs of AI use from your writing: Get rid of em dashes, colons, and of course the now-icky “It’s not X; it’s Y” formulations.

[Read: The typo vibe shift]

The problem is that the efficiency and frictionlessness that make AI appealing to writers are the same qualities that make it feel untrustworthy to readers. And readers are right not to trust it. No matter how much we may tell ourselves that AI is just a tool like spell-check, it isn’t. When we use AI to flesh out ideas, we lose the most important part of the writing process: thinking.

We tend to believe that efficiency is the highest virtue, the four-hour workweek the ultimate goal. Why sweat over the introductory paragraph of an essay if an AI program can sail over whatever argumentative obstacle you have in the space of 15 seconds? But the effort and the hang-ups are, as they say, a feature of the human thought process, not a bug. When human beings write, we judge ourselves; we stop; we backtrack. In published writing, the traces of this process are erased. But it is the process that makes human writing sensible and meaningful. Many authors describe how, when they’ve finally hit on the right idea, writing feels like going down a water slide; putting one sentence after another becomes easy.

When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.

AI can’t make that kind of judgment. Even if the companies that design AI programs could make them reason like a human being—a project whose hubris is underrated, given that we don’t fully understand the mechanisms behind our own thought processes—they won’t. After all, users consistently say they want AI models to be agreeable, compliant. That means AIs are reluctant to do what the voices in our head do all the time: interrogate the validity of our premise; misunderstand us, forcing us to explain better; insist that a query is stupid; refuse to answer a question at all. In a study published in March, a group of Stanford and Carnegie Mellon University scientists found that top AI models affirm their users’ ideas 49 percent more than humans do in conversation. They also discovered that participants rated more sycophantic answers as “higher quality” and said a sycophantic attitude made them more likely to use AI again.

So we end up with canned perfection—writing that can’t really be argued with, because it has no underlying deliberative reasoning process, no train of thought. As I wrote on X recently, AI writing is almost impossible to edit, because even when it sounds plausible, a closer look will show that every element is equally off: The tone is bland; individual word choices are baffling; the structure lacks sense; key pieces of the argument are missing; facts are false. Working on AI text, as an editor, is like trying to operate on a body whose skin, muscles, veins, bones, and organs are all compromised. There’s nothing to leave intact, nowhere to begin.

[Read: AI-writing scandals are getting very confusing]

Recently I asked ChatGPT Pro—which I paid for to summarize financial spreadsheets; I’m not a hater—whether it was capable of viciously dismantling an argument I’d read about the stock market. “I’m tuned to avoid being casually cruel, contemptuous, or needlessly snide,” it explained. “That doesn’t mean I can’t be sharp, skeptical, funny, or sarcastic … I can absolutely be dismissive of bad arguments, flabby prose, obvious spin, bureaucratic nonsense, or market delusions wearing a little tech-futurist hat.”

I wanted to look under the hood a little, to understand the reasoning behind how the program chose its phrases. “Why would market delusions be wearing a ‘little’ tech-futurist hat, rather than a big one?” I queried.

The program was incapable of pausing to wonder whether it had chosen the wrong word. Instead, it sought to justify itself in increasingly bewildering ways. “Because ‘little’ makes it funnier,” ChatGPT Pro answered. “The market delusion wants to stride into the room as Prometheus; ‘little hat’ makes it look like a raccoon that found a conference badge.”

Why a raccoon with a conference badge? I pressed. Why not a dog, or a frog?

“Raccoons live in alleys, storm drains, garbage nights, and morally humid environments,” it wrote. “Frogs have existential agency, while raccoons have logistical agency. A frog at a VC afterparty seems like an omen. A raccoon at a VC afterparty seems like it got on the guest list … Think: alley behind a conference hotel, the loading dock of a casino, the VC afterparty where everyone is saying ‘alignment’ but meaning ‘exit liquidity’ … Alignment sounds like ethics, safety, shared goals, humanity, the future. It has dry-cleaned blazer energy … Exit liquidity is socially clammy.”

All of these sentences are grammatically perfect. They also make no sense. And all substantially AI-generated writing is like this, under the hood. ChatGPT Pro agrees, though it has no idea why. “What if I said the whole raccoon thing made no sense as a metaphor?” I asked the program.

“I would agree,” it replied, ever servile. “At first, the raccoon worked as a throwaway image … A metaphor can survive one or two explanatory layers. After that, it starts looking like a raccoon in a trench coat.”

This is the kind of communication we’re becoming surrounded with. Its infiltration into every domain of our lives can’t be stopped. Even people who don’t use AI will begin sounding more like it. (A preprint by the Max Planck Institute for Human Development found that in off-the-cuff verbal conversations, such as podcast discussions, people are already exhibiting “a measurable and abrupt increase in the use of words preferentially generated by ChatGPT—such as delve, comprehend, boast, swift, and meticulous.”) After all, we remain so much smarter than machines, so much subtler, and thus so much quicker to learn and pick up cultural cues. The difference in how we operate will be extraordinary, and not at all hypothetical. Ten years ago I composed a reconciliatory email to a boyfriend but never sent it, because I couldn’t get the phrasing right. Only much later did I realize I simply didn’t mean what I’d been trying to write. If I’d had an AI program to help me get over the hump, I’d be married to a different person. A much less suitable one.

Maybe human writing will become like cloth-aged cheese or handloom rugs, an artisanal product created effortfully. Maybe we will come to treasure older writing. Herman Melville, George Orwell, Toni Morrison—all authenticated. Writing like this will be a fossil record for a kind of thought process we buried without realizing it. The other night, as I was drifting off to sleep, a 19th-century poem popped into my head:

Wynken, Blynken, and Nod one night

Sailed off in a wooden shoe,—

Sailed on a river of crystal light

Into a sea of dew.

A children’s rhyme, but it had a new beauty.

Or maybe smooth communiqués that arrive on time and betray no confusion, doubt, or internal struggle—that polish up our images as affable, efficient, and universally, if superficially, wise—is what we want. But at least we should know what we’re sacrificing.

Trump Might Already Be a Lame Duck

2026-05-29T05:00:00-04:00

President Trump has never really cared about the Republican Party per se. He basks in its adulation, and it’s beneficial to him when the GOP controls Congress. But he’s never adhered to its orthodoxies or honored its heroes. Neither has he been willing to brook internal dissent in the name of the party’s big tent. He demands absolute fealty but displays little loyalty. He can’t help obsessing over his personal priorities—such as his proposed ballroom or his retribution campaign against perceived tormentors—to the detriment of his party’s political interests. On ballots, Donald Trump (MAGA) would be more accurate than Donald Trump (R).

With little more than five months until the midterms, that divergence between what Trump thinks is good for Trump and what is good for the Republican Party has never been wider. Trump’s priorities appear in many ways to be hurting the GOP’s chances in November, when it already faces stiff odds of keeping control of Congress. The war he started with Iran put Americans’ economic struggles front and center when the price of gasoline jumped. Any semblance of a national legislative agenda has evaporated as he pushes long-shot bills that his own party declines to take forward. And his obsession with construction in and around Washington, D.C., it is safe to say, doesn’t suggest a chief executive focused on the problems of everyday citizens.

Meanwhile, Trump has wielded his clout inside the party like a broadsword, endorsing primary opponents in races against incumbents who defy him. Trump has a perfect endorsement record this year: All 118 candidates—for House, Senate, and governors’ races—he has backed in primaries have won, according to a Fox News count (though many of these races were not really contested). Even though Trump’s power over his party appears at its pinnacle, many Republicans believe that the president has actually accelerated his own political decline. Many of those primary winners may struggle in November, darkening the GOP’s prospects for keeping control of Congress. And at least some of the defeated incumbents, who will serve on Capitol Hill until next January, now feel liberated to push back on what they dislike in Trump’s agenda. Others in the Senate who are not up for reelection are bitter about the president’s role in their colleagues’ defeat and have shown little interest in helping him pursue his personal-grievance campaign.

“The problem is he has nobody around him who is willing to tell him, ‘Sir, the stuff you are talking about is not possible, and you are shooting yourself in the foot every time,’” one Republican Senate adviser told us. “He essentially has lame-ducked himself in pursuit of retribution, and either the staff has failed to make a reasonable argument against these actions, or they have told him this and he is no longer listening.” Either way, the party loses.

Since Trump returned to the White House, very few Republicans have dared defy him. Most have set aside private reservations to embrace his push on tariffs and mass deportations while professing ignorance about Trump’s efforts to enrich himself and his family. (“I haven’t seen the story,” is a common refrain.) On those rare occasions when a lawmaker has resisted his will, Trump has paid attention and waited for his revenge. Representative Thomas Massie of Kentucky, as measured by his voting record, was a reliable conservative. But he also was a prominent Republican voice calling for the release of the Epstein files. Trump opposed the release (unusually, he didn’t get his way), slammed the Kentucky congressman, and supported his primary opponent. Massie lost. Seven Indiana state lawmakers broke with Trump’s effort to redistrict their state in favor of the GOP. Trump backed their primary challengers. Five of the incumbents lost; one other faces a recount.

[Read: The ‘crazy’ plot to release the Epstein files]

The biggest ructions have been in the Senate. No modern president has endorsed challengers to two sitting senators from his own party. But Trump successfully backed Texas Attorney General Ken Paxton against the four-term incumbent John Cornyn in Tuesday’s primary runoff, and also helped oust Senator Bill Cassidy of Louisiana. Senator Thom Tillis of North Carolina, meanwhile, got so fed up with Trump that he decided to retire. But scorned senators can be furious foes. The Republican majority of 53 already was a tad precarious because of occasional defections from two relative moderates, Susan Collins of Maine and Lisa Murkowski of Alaska. Add in Tillis, Cornyn, Cassidy, and retiring former Majority Leader Mitch McConnell, and Trump’s grip on the chamber starts to look shaky.

The president’s purge of candidates who have shown they can win a general election in favor of newcomers who are focused on pleasing him means the party will now have to do more to have a shot at victory in November. Some strategists think that the Texas race for U.S. Senate with Paxton on the ticket could require as much as $100 million in additional Republican funding from out of state, both because Paxton is a less effective fundraiser than Cornyn and because his turbulent history leaves him more vulnerable to Democratic attacks. Although Democrats have often hyped but seldom delivered in the Lone Star State, they see Paxton as the weaker opponent for state lawmaker James Talarico. A Talarico win in Texas could hand the Senate to the Democrats; even if he loses, the diversion of GOP resources to Paxton could put other states in play.

Senate Majority Leader John Thune made no secret of his support for Cornyn. When Trump and Thune spoke on May 18, the call was so tense that Thune told his advisers afterward that he thought Trump would back Paxton. Senator Tim Scott of South Carolina, who runs the National Republican Senatorial Committee, followed up with his own pitch for Cornyn in a less contentious conversation with the president, people briefed on the exchanges told us. Trump endorsed Paxton the next day. (Internal polls were already showing Paxton ahead, but the president’s endorsement turned the contest into a rout.)

Thune in particular has at times thrown up his hands in the face of the president’s obduracy. Trump, in turn, has been venting to other GOP senators that the upper chamber is ineffective and insufficiently loyal. “There are definitely frustrations there that are not going away,” one person familiar with the exchanges told us, “and there is no appetite from Thune to resolve it.”

Trump grew so frustrated over the Senate’s inability to pass the SAVE America Act—legislation designed to crack down on issues as disparate as immigrant voting rights and transgender surgeries—that he embraced the idea of a “talking filibuster,” based on the recommendation of Utah’s Mike Lee. (The talking filibuster is a rarely used tactic during which senators delay voting on a bill by refusing to yield the floor, thereby forcing very lengthy debate.)

Thune had the thankless task of explaining to Trump that such an approach would also empower Democrats to offer their own amendments. That could have forced floor votes on issues such as tariffs, the Iran war, and abortion rights, where Republicans would have to choose between defying the president and giving Democrats ammunition for the fall. The legislation remains stalled.

[Read: A serious Senate debate about an unserious bill]

Senate Republicans have shown other flashes of independence. Tillis held up Trump’s nominee for chair of the Federal Reserve until the Justice Department stopped pursuing Jerome Powell. Rand Paul, Murkowski, Collins, and Cassidy voted to advance a resolution that would require Trump to get congressional authorization to continue the war in Iran. (House Republicans canceled a vote on the measure out of fear that it might pass.) And early hopes that Congress might authorize $1 billion in security funding for the White House ballroom were dashed after pushback from some GOP lawmakers and a ruling from the Senate parliamentarian. Trump ordered Thune to fire the parliamentarian; the majority leader refused.

Last week, still smarting from the Paxton endorsement, the Senate went into recess rather than consider Trump’s plan to create a nearly $1.8 billion fund for alleged victims of government “weaponization.” The plan was widely and immediately panned on two grounds: the prospect of recompense for the rioters who attacked Congress on January 6, 2021, and protections that would forever shield Trump, his family, and businesses from IRS scrutiny. “So the nation’s top law enforcement official is asking for a slush fund to pay people who assault cops? Utterly stupid, morally wrong — Take your pick,” McConnell said in a statement. The fund’s fate is now unclear.

Over the weekend, Iran hawks in the Senate who are usually joined at the hip to Trump—Lindsey Graham, Ted Cruz, Roger Wicker—fumed at the reported terms of an Iran peace deal that the president was touting as imminent. The White House tried to silence the objections, but the timetable for a deal notably decelerated, and nothing has been signed.

In the House, where the Republican majority is even more tenuous than in the Senate, Trump has also faced defiance. Massie, Kevin Kiley of California, and Don Bacon of Nebraska broke ranks to give Democrats a chance to oppose Trump’s tariffs on Canada. Bacon is retiring after criticizing Trump’s foreign policy. Kiley, meanwhile, found his congressional district eliminated as California leaders retaliated for GOP redistricting in Texas. Kiley declared that he would run for reelection in a new district—as an independent.

Steven Cheung, the White House communications director, downplayed the internal GOP strife. “President Trump is the unquestioned leader of the Republican Party,” he told us in a statement. “Look no further than his perfect and sterling record in the past year—a 100% success rate for his preferred candidates, proving his endorsement is the most powerful endorsement in history.”

For months now, Republicans have fervently hoped that Trump’s focus would shift to issues that could help the party in November. Instead, he has been consumed with an Iran peace agreement and with his projects: new paint for the Reflecting Pool, a triumphal arch near Arlington National Cemetery, the conversion of a Washington, D.C., public golf course into championship links, and, of course, the ballroom. The economy? Not so much.

A recent Quinnipiac poll found that 45 percent of voters said that affording gas is now somewhat or very difficult, up from 29 percent in December. The same poll found that 55 percent of voters, including 16 percent of Republicans, blamed Trump “a lot” for the rise in costs, and 56 percent of voters opposed the U.S. military action against Iran.

[Read: Trump voters are over it ]

Despite White House promises that the president would hold events across the country to promote economic fixes, Trump has largely stayed in Washington (or at Mar-a-Lago) and declared that the affordability crisis is a Democratic “hoax.” He seems uninterested in fulfilling his campaign promises to get prices down. Earlier this month, the president effectively gifted the Democrats a campaign ad by saying, “I don’t think about Americans’ financial situation,” when he was asked about the impact of the Iran war. On Wednesday, while insisting that domestic political considerations would not factor into his negotiations with Tehran, Trump declared, “I don’t care about the midterms.” Many Republicans likely nodded in resigned agreement.

The Brazenness of DOJ’s Reported Investigation of E. Jean Carroll

2026-05-28T18:45:00-04:00

This is an edition of The Atlantic Daily, a newsletter that guides you through the biggest stories of the day, helps you discover new ideas, and recommends the best in culture. Sign up for it here.

Updated at 10:51 a.m. ET on May 29, 2026

No White House is immune to hypocrisy. What makes the Trump administration’s approach to justice so astonishing is not just the depth of the hypocrisy but its brazenness.

Last night, CNN reported that the Department of Justice is pursuing a criminal investigation against E. Jean Carroll, the writer who has accused Donald Trump of raping her in the 1990s, and won nearly $90 million in civil judgments against him. The probe reportedly focuses on whether Carroll committed perjury during her testimony related to two civil lawsuits against him, both of which she won. (In a statement Thursday night, Andrew S. Boutros, the U.S. attorney for the Northern District of Illinois, denied that his office had opened a criminal investigation into Carroll; CNN reported that sources reaffirmed the existence of the investigation to them after his statement was released.)

The news comes less than 10 days after Trump—putatively acting as a private citizen—announced an agreement with that same Justice Department to create a $1.8 billion slush fund to reward his political allies, potentially including those who sacked the Capitol on January 6, 2021.

“The use of government power to target individuals or entities for improper and unlawful political, personal, or ideological reasons should not be tolerated by any Administration,” the DOJ official Trent McCotter said while announcing the settlement. That quote came in a written statement, which mercifully freed him from having to keep a straight face while saying it. In truth, using government power to target individuals for political and personal reasons seems like an apt description of the probe into Carroll as well as many of the Justice Department’s steps in recent months.

Carroll’s lawsuits infuriated Trump. The president has been accused of sexual assault by many women; he has denied all accusations, although he also boasts about nonconsensual groping in the infamous Access Hollywood tape. Carroll, however, brought a case where a court actually found him liable for sexual abuse. Judge Lewis Kaplan wrote that although “Carroll failed to prove that she was ‘raped’ within the meaning of the New York Penal Law,” the jury found that Trump “did exactly that” in the common understanding of “rape.” Trump insulted Carroll repeatedly on Truth Social as well as on the stand during one of the trials. He insisted before the trial that he did not know her, despite a picture showing them together, and said she was not his “type,” but when shown the photograph in a deposition, he mistook her for his ex-wife Marla Maples.

Trump couldn’t beat Carroll in court (though ongoing appeals efforts mean he has not yet had to pay up), but he does have a Justice Department that has shown a willingness to bring preposterous cases against his political enemies, and an acting attorney general who appears determined to prove he can succeed at political retribution where his fired predecessor did not. (CNN reported that Todd Blanche, who holds that title, was recused from this case because of his previous work as Trump’s personal lawyer in Carroll litigation.)

The accusation of perjury centers on financial support for Carroll’s legal efforts from Reid Hoffman, a LinkedIn co-founder and major donor to liberal causes. In a 2022 deposition, Carroll said she did not have any outside support for her litigation, but two weeks later, her lawyers told a judge and Trump’s attorneys that they had secured funding from a nonprofit Hoffman leads.

There are a couple of reasons to be skeptical of claims of perjury here. First, Kaplan, the judge overseeing the civil cases, already considered and dismissed concerns over the testimony. Carroll’s lawyers said she had no communication with Hoffman or the group, but Kaplan allowed Trump’s team to question her once more. The judge then concluded that Carroll’s credibility was not in doubt and barred Trump’s attorneys from questioning her about the funding during the subsequent trial.

Second, the DOJ investigation is reportedly being overseen not by a U.S. attorney in New York, where the trial occurred, but by Boutros, the U.S. attorney in Chicago. Although this is legal, it is unusual (or it was until this Trump administration, during which DOJ has repeatedly assigned faraway offices to handle political cases). The track record of the Chicago U.S. Attorney’s Office is a red flag of its own: The office was recently in the news when prosecutors dismissed the only remaining misdemeanor charges against members of the “Broadview Six,” a group of people arrested at a protest at an ICE facility last fall. They had already moved to dismiss felony charges, which turned out to be a result of misconduct by prosecutors.

April Perry, the federal judge presiding over the case, said she was “incredibly shocked” by prosecutors’ conduct during grand-jury proceedings. “I have never seen the types of prosecutorial behavior before a grand jury that I saw in those transcripts,” she said. Prosecutors personally vouched for the credibility of evidence before a grand jury, which is impermissible. When they failed to get an indictment, they excused grand jurors who voted against charges and tried again. They also spoke with grand jurors outside of a courtroom. Later, they redacted transcripts to hide it all from Perry.

Perry summoned Boutros to her courtroom, where she scolded him. “Your sole goal is to do justice. Your client is justice itself,” she told him. “I do believe deeply in the presumption of regularity and that most government attorneys are doing the best they can to do the right thing. That trust has been broken.”

Any investigation into Carroll faces the same problem: Boutros and the Trump Justice Department as a whole no longer have the benefit of the doubt that their actions are fair and impartial, and that they aren’t just attacking Trump’s enemies, real or perceived. Even if the probe sputters, a spurious criminal investigation is a form of extrajudicial punishment. Defendants must spend time and money on attorneys; the Southern Poverty Law Center also recently found itself cut off from financial channels because it is facing a dubious indictment.

Trump ran for office decrying what he alleged was the “weaponization” of the Justice Department, and he promised to reverse it. But what was apparent then and is beyond dispute now is that Trump had no problem with politicized justice—he just wanted it on his side. The Broadview and Carroll cases show just how effectively he has achieved that.

Related:

Here are three new stories from The Atlantic:

Today’s News

U.S. and Iranian officials reached a tentative agreement to ease tensions in the Strait of Hormuz and begin 60 days of talks over Iran’s nuclear program, though the deal still requires approval from President Trump and Iran’s supreme leader, U.S. officials said. The development came after the United States and Iran exchanged strikes overnight.
Yesterday, Trump refiled a $10 billion defamation lawsuit against The Wall Street Journal over a report alleging that he sent a sexually suggestive birthday message to the convicted sex offender Jeffrey Epstein. The new complaint comes after a federal judge dismissed Trump’s earlier case, ruling that he had failed to show the newspaper acted with the “actual malice” required to win a defamation claim.
In her forthcoming memoir, Jill Biden said she feared that her husband, former President Biden, was having a stroke during his widely criticized 2024 debate performance.

Dispatches

Time-Travel Thursdays: The Atlantic once warned that political violence was straining Colombia’s democracy. Ahead of another consequential election, some of those tensions feel familiar again, Rafaela Jinich writes.

Explore all of our newsletters here.

Evening Read

Martyn Goodacre / Getty

I Am Begging You to Read Terry Pratchett

By Helen Lewis

Will we ever live to see a successful screen adaptation of a Terry Pratchett novel? The Amazon television series Good Omens, which ended this month, came closest—but that book, a comedy about an angel and a devil teaming up to avert Armageddon, was co-written with Neil Gaiman, and the source material ran out after the first season in any case.

Pratchett is the funniest English writer since P. G. Wodehouse, with a sharp, satirical edge disguised by the trappings of the fantasy genre—vampires, dwarfs, witches, and wizards. Many fans thought the original covers of Pratchett’s novels went too heavy on busty maidens and strapping men with big swords, undermining their literary merit, and a similar problem has beset the various screen adaptations from Sky and the BBC. I suspect that casual viewers can’t compute the idea of watching something with the comic tone of a Charles Dickens or Tobias Smollett novel while being distracted by CGI trolls.

Read the full article.

A ‘Promising Democracy’ That Can’t Stop Fighting Itself

2026-05-28T15:37:00-04:00

This is an edition of Time-Travel Thursdays, a journey through The Atlantic’s archives to contextualize the present. Sign up here.

In April 1948, after the assassination of the populist leader Jorge Eliécer Gaitán, crowds poured into the streets of Bogotá. Buildings burned. Churches were looted. Armed mobs seized parts of the capital. Gaitán—a labor lawyer turned political phenomenon who seemed poised to become Colombia’s next Liberal president—had built a mass following among working-class Colombians frustrated by inequality and elite rule. An enraged crowd beat the alleged gunman to death before his motives could be revealed. Gaitán’s killing triggered El Bogotazo, the explosion of unrest that marked the beginning of La Violencia, the brutal conflict between Liberals and Conservatives that would kill more than 200,000 Colombians over the following decade.

In 1950, an article in The Atlantic warned that Colombia’s “promising democracy” was beginning to come apart. An unnamed writer noted that the country had functioned “more consistently and over a longer period than any other Latin American republic,” but that its government was faltering. Across rural Colombia, Liberal and Conservative elites backed armed supporters who fought to defend each party’s political power and economic interests. The country’s leaders seemed to govern by intimidation: opposition meetings broken up in small towns, armed groups terrorizing voters, emergency decrees restricting democratic life.

More than 70 years later, familiar patterns are emerging as Colombia heads into one of its most consequential elections in years. On Sunday, Colombians will vote for a successor to President Gustavo Petro, the country’s first leftist president and a former member of the Marxist M-19 guerrilla movement. Petro came into office promising to negotiate cease-fires with every major armed group still operating in Colombia, but many of these talks eventually stalled or collapsed. He suspended negotiations last year with the National Liberation Army, or ELN—now Colombia’s largest active guerrilla group—after it launched an offensive in northeastern Colombia that killed more than 30 people. Still, even as Petro’s peace agenda has faltered, several armed groups, including the ELN, have signaled that they may be open to restarting negotiations with the next government.

The election has become a referendum on Petro’s “total peace” strategy. His supporters say that Colombia cannot end decades of conflict through military force alone; the Petro ally and presidential candidate Iván Cepeda has promised to continue the negotiations. His Conservative rival Paloma Valencia and the right-wing outsider Abelardo de la Espriella are each campaigning on restoring security through a strong military response, arguing that Petro’s approach allowed armed factions to regroup and expand their territorial control, particularly in rural and border regions.

The debate has become inextricable from the country’s deteriorating security situation. Although the cities, where much of the country’s wealth and political power are concentrated, have become safer and stabler over the past decades, armed groups have staged dozens of bombings and drone strikes across Colombia in recent months. Dissident factions of the Revolutionary Armed Forces of Colombia, or FARC—the far-left guerrilla group that fought the Colombian state for more than 50 years before signing a landmark peace agreement in 2016—were behind several of the attacks, targeting civilians and military bases just weeks before the election. Some factions never fully demobilized after the accord, while others later splintered from the peace process entirely. Especially during election cycles, these insurgents use violence to protect their illegal economies and to demonstrate their continued power in regions where they often have a stronger presence than the state itself.

The race has felt, at times, like a dispatch from 1948. Last summer, a Colombian senator and presidential hopeful, Miguel Uribe Turbay, was shot during a campaign rally in Bogotá and died two months later. A 15-year-old hit man and eight others were arrested for the shooting, and the country’s attorney general has issued warrants against leaders of the Segunda Marquetalia, an offshoot of FARC, in relation to the assasination. Earlier this month, a former mayor and a staffer allied with the presidential candidate de la Espriella were shot dead (the shooters have not been apprehended). Colombia’s public-defender’s office warned that the killings could threaten “democratic participation” ahead of the vote.

When an unbylined Atlantic writer covered Colombia in 1963, the country looked very different. Exhausted by years of bloodshed, the Liberal and Conservative parties had agreed to share power through a coalition known as the National Front, alternating the presidency and dividing government positions between them. The article describes a country trying to steady itself after chaos, building roads and housing projects, attracting foreign investment, and projecting an air of stability after years of partisan violence.

Still, beneath that stability, the writer detected problems, such as economic woes and oligarchic tendencies, that had not so much disappeared as hardened. The coalition government’s success, the writer observed, would ultimately depend on whether it could address “urban squalor and rural poverty, whose victims are being aroused to a sense of their own strength.” Those lingering tensions would soon reshape Colombia again. A year after the article was published, the FARC and the ELN emerged as separate militias. A few decades later, presidential candidates, journalists, and judges were routinely assassinated by cartels, guerillas, and paramilitary groups warring with one another and with the state.

The 1950 article in The Atlantic ends without resolution: “What will remain of Colombia’s promising democracy after so long a period of restraint and turbulence remains to be seen.” Reading it now, amid another tense moment in Colombian politics shadowed by assassinations, bombings, and fear, that line feels like a question that Colombia has spent generations trying to answer.

More From the Archives

The Slow Food movement was born in 1986 when Carlo Petrini, an Italian environmental activist and former radio journalist, rallied a group of friends to protest the replacement of a beloved coffee shop in Rome with a McDonald’s. When a bystander asked what, if not fast food, he was in favor of, he said: “Slow food.” What did “slow food” actually mean? That was something Petrini, who died last week at the age of 76, would spend the next few years figuring out, eventually hatching an international movement that combined an embrace of sustainable farming and traditional cooking with an epicurean’s appreciation of good food. (Petrini would also found the University of Gastronomic Sciences, in Pollenzo, Italy, the first such institution in the world.)

In 1999, Corby Kummer, an Atlantic senior editor and a longtime food writer, helped introduce the Slow Food movement to America with his article “Doing Well by Eating Well.” “Appetite can join forces with radicalism,” he wrote, “and both sides can be the stronger” for it. That article would soon grow into the book The Pleasures of Slow Food: Celebrating Authentic Traditions, Flavors and Recipes.

As Kummer noted last week, Petrini showed people that they couldn’t enjoy a region’s best produce and cuisine “without recognizing the dignity and well-being of the people who make food, the importance of tradition and human contact, and social and environmental justice.”

— Scott Stossel, national editor

The U.S. Is Winging This Ebola Outbreak

2026-05-28T15:04:46-04:00

By the time African health officials confirmed the world’s latest Ebola outbreak, the epidemic had already spilled from the Democratic Republic of the Congo into neighboring Uganda. Within two days, the World Health Organization declared the outbreak a public-health emergency of international concern. Less than two weeks later, the potential case count has risen past 1,000, including more than 230 deaths, and 10 other African countries have been designated at risk of being swept into the crisis.

Countries and health coalitions from around the globe have quickly mobilized funds, medical resources, and personnel to the region. But one nation has been conspicuously absent from the core of the international response. Prior to January, when the United States officially withdrew from the World Health Organization, it was one of the coalition’s largest, richest, and most prominent partners, and its biggest funder. Now it has sidelined itself—limiting the potential effect any of its actions will have.

Were the U.S. still a member of the WHO, its federal health officials likely would have been able to start responding to the crisis sooner and better positioned to direct resources where they were most needed; were USAID still intact, its officials would have been in Congo, managing the outbreak before it ballooned. As things stand, American health officials did not learn of the epidemic until nine days after the WHO did. (When reached for comment, a State Department spokesperson wrote over email that the U.S. began its response within 24 hours of hearing about the outbreak and argued that “the WHO’s delay in informing the world of concerns until May 15 had a grave impact.”) Even as the U.S. has leaped into action, it has remained on the outskirts of the primary effort to control this outbreak.

American leaders “are not doing nothing,” Lawrence Gostin, a global-health-law expert at Georgetown University, told me. The U.S. government has announced that it is dispatching more than $160 million in emergency and humanitarian funds to contend with Ebola on the ground, deploying CDC personnel and a disaster-assistance-response team to the region, and bankrolling “up to 50” Ebola-treatment units in affected areas. Yet to public-health experts around the world, the U.S. response looks siloed, uncoordinated, and ultimately less effective than it would otherwise be. When one country holds itself at arm’s length from other global-health actors during an international crisis, “at best it wastes resources,” Jennifer Nuzzo, the director of the Pandemic Center at Brown University School of Public Health, told me. “At worst it winds up conflicting with or impeding the work of others.”

For this particular Ebola outbreak, the margin for error is even slimmer than usual. The viral strain causing the epidemic, Bundibugyo, is frequently missed by standard field tests and lacks both treatments and licensed vaccines. Local and international health officials were weeks late responding to it, which allowed the virus to spread more widely. Many regions of Congo, including ones at the center of the outbreak, have been fragmented by intensifying armed conflict, which has weakened health infrastructure. And the Trump administration’s gutting of domestic and international public-health infrastructure increased the region’s fragility, cut down on available health personnel, and likely delayed the initial detection of Bundibugyo, researchers told me.

Situations this dire, Nuzzo said, call for an incident-command system, in which roles are carefully delegated “so we aren’t showing up and stepping on the toes of others who are already in the area.” American leaders are still communicating with relevant countries to some degree—setting up bilateral financial agreements, for instance, with Congo and Uganda. The government also has contributed to an emergency-response fund through the United Nations Office for the Coordination of Humanitarian Affairs and is working with “implementing partners, Africa CDC, and technical channels on the ground,” according to the State Department spokesperson; Andrew G. Nixon, the deputy assistant secretary for media relations at the Department of Health and Human Services, wrote in an email that the U.S. has “activated an aggressive, coordinated response.”

Meanwhile, the WHO, a UN agency that marshals responses across its nearly 200 member nations, is spearheading collaborative efforts on a much larger scale and leveraging its own technical expertise—capabilities that the U.S. does not have on its own. The Trump administration has also reportedly placed restrictions on the number of federal health officials who can attend virtual WHO meetings.

As a result, the U.S. is now a side channel to the main event, risking redundancies through its bespoke response. “You’re going to get massive confusion and duplication,” Salim Abdool Karim, who chairs Africa CDC’s Emergency Consultative Group, told me. (The WHO did not respond to requests for comment.)

The WHO has never been the sole or perfect arbiter of public-health response. In recent years, experts have criticized aspects of the WHO’s delayed responses both to the Ebola outbreak that began in 2014 and to COVID-19. (When justifying the U.S.’s withdrawal from the WHO, the White House specifically cited the organization’s “mishandling of the COVID-19 pandemic.” President Trump has also said that the WHO asked the U.S. to contribute too much money, insisting that “World Health ripped us off.”) But few agree with U.S. leaders that improving global health involves withdrawing from the organization. As Ebola rips through Congo and threatens to overflow into neighboring regions, coordination is the only viable path—and the WHO is the main channel through which coordination occurs. “Trying to imagine how you would do this response without WHO? It boggles my mind,” Abdool Karim said. Yet that’s exactly what the U.S. is now attempting to do.

As things stand, Gostin, who has been in constant contact with colleagues in Congo and at the WHO, said that he and many of his fellow public-health experts have little knowledge of what actions that U.S. officials have actually taken. Some of the government’s choices so far also seem incongruous with the region’s needs. For instance, funneling so many early-response resources into Ebola-treatment units—which are extremely expensive—makes “absolutely no sense,” Courtney Blake, who helped lead the USAID response to the Ebola outbreak that began in 2014, told me. Treatment units, although important, represent a late line of defense, Blake said, because they do little to halt the virus’s spread. Top officials in Uganda’s Ministry of Health have also expressed confusion about the American contribution to the outbreak response, at one point last week saying that the ministry hadn’t communicated with the U.S. about treatment centers at all.

Those resources, Blake and others told me, could be better focused on efforts that would directly slow viral transmission—including PPE dispersal, testing, quarantining, and community engagement. And this morning, the State Department did announce that some of its allocated funds would help its “implementing partners” with “PPE procurement and delivery, border screening and surveillance, contact tracing, and diagnostics supplies.” Several experts also emphasized the importance of local communication: In the past week, two Ebola-treatment centers have been set on fire by protesters, in at least one case because family members of a man suspected to have died from the virus had been prohibited from retrieving his body. (The Ugandan Ministry of Health and Congo Ministry of Public Health did not respond to requests for comment.)

But what the U.S. actions add up to hasn’t been apparent to the experts I spoke with. “Is there a big-picture strategy?” Mohammad Karamouzian, an infectious-disease epidemiologist at the University of Toronto’s Dalla Lana School of Public Health, told me. “Or are they just trying to show they are doing something?” Any attempt to limit the virus’s spread is now more difficult, too, because “arguably the biggest implementation force on the ground in the region is gone,” Karamouzian said—namely, USAID.

The State Department has started to reconstitute some of the humanitarian resources that the Trump administration previously rendered defunct, Paul Spiegel, the director of the Johns Hopkins Center for Humanitarian Health, told me, by setting up a Bureau of Disaster and Humanitarian Response in the State Department and reassembling disaster-assistance-response teams. The department’s spokesperson argued that USAID reform has not undermined the country’s Ebola response and said that the U.S. responded faster to this outbreak than USAID did to similar outbreaks in 2014 and 2018. (Blake pointed out that although international emergencies were declared later for those epidemics—which grew more slowly than this one—USAID officials were already in the region, available to mount a local response, when those outbreaks began.)

Ultimately, though, the U.S. has been very clear about where its priorities lie—with its own interests. At a recent press conference, Secretary of State Marco Rubio said, “We don’t want anyone dying or being affected by Ebola, but our No. 1 priority will always be making sure it doesn’t come to the United States.” The Trump administration has put in place multiple travel restrictions aimed at keeping Ebola out of the U.S. And although in the past, Americans caught up in dangerous outbreaks have been flown home to be monitored and treated, during this Ebola epidemic, the U.S. has instead evacuated ill and exposed Americans to Germany and the Czech Republic and is standing up a makeshift quarantine center—for Americans specifically—in Kenya.

This “America First” stance has stoked anger among some communities in Congo, Leslie Roberts, an epidemiologist at Columbia University, told me, and where Americans were once welcomed as public-health allies, they are now seen as enemies. If a main part of the U.S. strategy is to coordinate directly with national health ministries, this depends on those ministries wanting to coordinate—which is not always possible in countries that have poor diplomatic relations with the U.S.

In the past, such as during a 2023 outbreak of Marburg virus in Equatorial Guinea, the U.S. depended on WHO relationships with other countries’ ministry of health, Beth Cameron, a former global-health-security adviser for USAID, told me. Coordinating through the WHO means that individual nations don’t need to scramble to remake such connections, or forge them anew, to confront each challenge. Inevitably, in some future outbreak, the U.S. will find that its isolation has left it unprepared to protect even itself.

Words of War

2026-05-28T13:41:19-04:00

Decades ago, it was a truism that the 24/7 news cycle exercised a malign influence on policy making. It kept senior leaders fixated on a flickering television screen when their time would have been better spent weighing evidence, debating alternatives, and considering opposing views. All true. But today we contend with 24/7 commentary, which is so ubiquitous that we barely notice it, even as it causes a kind of dry rot of our good judgment.

Supporters of the Trump administration’s war against Iran periodically complain that much of the criticism the administration faces is as ludicrous as denouncing Franklin D. Roosevelt’s war leadership in April 1942 would have been, before Midway, Guadalcanal, and the North Africa landings. They have no record of extending that sort of charity to previous administrations, but that does not invalidate the larger point.

The 24/7 commentary treadmill means that certain simplifying words get used over and over. But in war, above all things, realities are almost invariably complex. Take the very word war. Advocates and critics of the Iran conflict assume, without question, that this is a war that began on February 28, and that it was launched by President Trump and Israeli Prime Minister Benjamin Netanyahu.

[Eliot A. Cohen: Send the frigates]

That is arguably the biggest strategic mistake of all: not knowing when the war you are in began, or even who started it. The past few months of bombing, blockading, and missile and drone strikes are but the latest campaign in a war that began at the inception of the Islamic Republic. American service personnel have died for nearly five decades at the hands of Iranian mines, IEDs, and missiles. The speeches of Iran’s leaders leave little doubt that they believe that they have always been at war with the United States and Israel. Their unprovoked missile attacks on Israel and acts of terrorism in the past few years alone—including the attempted assassination of Trump during the Biden administration—suggest that we should concede the possibility that they may be right.

Americans hate long wars, to the point that they frequently refuse to acknowledge their existence. Yet World War II did not begin at Pearl Harbor: Arguably it began in 1937, when Japan began its major onslaught against China. The Vietnam War did not begin in 1965, with the shift of American forces to conventional combat rather than advice and support; it had begun by 1946, and perhaps earlier. And although Islamists of various stripes think the Crusades are an interesting and important model to study, Americans blanch at the idea of a war lasting centuries, and fought over religious issues, no less. This current bout of fighting looks different, however, if one frames it as merely a particularly violent episode in a much longer conflict.

The words victory and defeat are often misleading. Even wars that seem exceptionally clear-cut in their outcomes can be ambiguous. The Japanese were vanquished by American naval and air power in World War II, but they achieved a major war aim, shattering permanently the European empires of East Asia. Hitler perished in the bunker in Berlin, but achieved much of his most vital war aim, the destruction of European Jewry. And although Britain was, in one sense, a victor in that war, it lost its vitality, empire, and sense of world power.

In some wars, everyone loses. In others, both sides may reasonably claim victory. At the end of the War of 1812 the British believed, correctly, that they had administered a thorough drubbing to the Americans, saved Canada from conquest, and demonstrated the supremacy of British naval power. The Americans, for their part, believed (equally correctly) that Britain could no longer project power into the North American heartland to block America’s westward expansion, and that they had taught the Royal Navy a healthy respect for America’s naval potential. Canadians celebrate today the formation of a distinct identity based on the cooperation of their varied peoples in fighting off American invaders. Native Americans, who were not formally parties to the war, were in fact its only real losers.

Sometimes the words winning and losing make little sense. Wars are composed, as Churchill once said, “of trends and episodes,” by which he meant the long-term pressures applied by such measures as blockade and bombing, and the sharp fighting of battles with a defined beginning and end. In the present case, is Iran winning by closing the Strait of Hormuz? In some ways, yes, but then again, its oil exports are equally strangled, and it has suffered a battering by the two most advanced air forces in the world, using the world’s most advanced munitions, guided by exceptional intelligence. Maybe winning the narrative counts as victory, but that does not make good hundreds of billions of dollars of damage. In the current war, both sides have had successes and failures; better to accept that this will not resemble a basketball game, with a single outcome based on points.

The most overused word is quagmire, easily found in foreign-policy periodicals, politicians’ speeches, and pundits’ sound bites. It is a lazy word. When you go into a quagmire, you are sunk, and will either die there or come out exhausted and filthy. It is a word that, like much of the commentary surrounding war, assumes away not only variable outcomes but the importance of operational choices, individual personalities, accident, fortune, and contingency—in short, the stuff of any real war.

Quagmire became particularly prevalent in American usage to describe the Vietnam War, and that is the implicit comparison lurking behind its use today, now compounded by the experience of Iraq and Afghanistan. Applying it to a war in which the United States has not sent (and is very unlikely to send) large expeditionary forces to fight a protracted insurgency, but rather is using air power and a naval blockade against a state, is ludicrous.

[Robert Kagan: Checkmate in Iran]

For decades after the Civil War, Republican politicians could guarantee reelection by “waving the bloody shirt,” reminding voters that a lot of Democrats had either been southerners or sympathized with the southern cause. It was a good way to avoid having to come up with solutions to the problems the country faced. So, too, waving the bloody shirt of Iraq, or harkening back to the jungles of Vietnam, fails to aid in understanding the particular problem of Iran, which all U.S. administrations have had to face since 1979, and none successfully.

None of this excuses the blundering and incompetence of the Trump administration’s entry into the war, and quite likely its conduct of it. The administration did not begin, as it should have, by occupying key islands around the strait, deploying such mine-hunting assets as the Navy possesses to the theater, devising schemes for war insurance, or, above all, securing the support of allies rather than spewing venom at them.

But it does suggest that we should be wary of lazy pronouncements, festooned with questionable analogies and tired catchphrases. That the administration is often simpleminded in its reasoning and outrageous in its rhetoric does not excuse any of its critics for behaving in the same way. Worse still, it is all too easy to slide from slipshod thinking into prophecies to which one clings despite disconfirming evidence. From there it is not that far to begin rooting, in effect, for your country’s enemies to prove you right.

The TACO Equilibrium

2026-05-28T13:27:00-04:00

The whole world expects President Trump to end the Iran war any day now. Trump keeps insisting that he’s in no rush to do so. Through it all, the oil markets remain surprisingly calm. These facts are all related.

When the war broke out, experts warned that if the Strait of Hormuz remained closed for more than a few weeks, oil prices would spike to $150 or $200 a barrel. The strait has now been closed for three months. Yet the price for a barrel of the most heavily traded type of crude oil sits at about $94, not so far from where it was in early March, shortly after the war broke out. Even after Trump’s latest declaration, in yesterday’s Cabinet meeting, that he felt no pressure to reach a peace deal (“I don’t care about the midterms,” he said), crude prices jumped by only 2 percent. “The math just doesn’t add up,” Rory Johnston, an oil-markets analyst who writes the widely cited Commodity Context newsletter, told me. “For people like myself who spend all day analyzing this stuff, we’re looking at prices wondering: Am I going insane? What is happening?”

Part of the answer is that the United States and other countries have dipped into their oil reserves to make up for some of the lost supply. But that can’t fully explain why oil prices have remained as low as they are. The more important reason has to do with investor psychology. The price of a barrel of oil reflects not just physical realities today, but expectations about what the market will look like in the near future. For the past three months, the global oil market seems to have been operating under the assumption that, before too long, the Strait of Hormuz will reopen and oil will start flowing again.

That assumption is rooted in a deeper underlying belief: that Trump will inevitably back down once the economic pain gets high enough. This is the so-called TACO theory of Trump’s decision making, as in “Trump Always Chickens Out.” “The market has correctly realized there’s an audience of one who will determine the outcome of this, and that’s Trump,” Arnab Datta, a managing director at the think tank Employ America who specializes in energy markets, told me. “Among traders, the assumption is that the pain can only get so high before Trump retreats.”

[David A. Graham: The TACO presidency]

That logic turns out to be dangerously circular. Prices are low because investors expect Trump to end the war before prices get too high; but because prices are low, Trump faces less pressure to end the war. In fact, the president seems to have figured out that he can calm the oil markets simply by gesturing at the prospect of a peace deal every so often. Of course, a peace deal or a new cease-fire could still be announced at any moment. But the dynamic between Trump and the markets—call it the TACO equilibrium—is what has kept the war going longer than almost anyone expected.

As a general matter, the belief that Trump will back down in the face of economic disaster isn’t unfounded. After Trump announced his “Liberation Day” tariffs in April 2025, the stock market lost trillions of dollars in value in a few days. Then, as the tariffs went into effect, bond investors began selling off U.S. Treasuries, sending interest rates soaring. A mere 13 hours into his new trade policy, Trump backed off and announced a 90-day pause on the tariffs, citing the fact that the markets had gotten “yippy.” Interest rates fell and the stock market experienced its largest one-day rally of the year. Investors who had bet that Trump would blink made a lot of money.

In the following months, a version of this dynamic would play out again and again. A new consensus view emerged on Wall Street that investors should respond to Trump’s threats not by selling, but by “buying the dip” and profiting when he inevitably backed down. This tactic became known as the “TACO trade.”

Then, on February 28, the U.S. and Israel struck Iran. The price of a barrel of Brent crude, generally considered the global benchmark for oil, spiked from about $70 to almost $120 in a little more than a week. But then, on March 9, Trump announced that the conflict was “very complete” and that the strait had been reopened. (It had not.) Oil prices fell below $90 a barrel. The TACO theory appeared to have been vindicated once again.

Except Trump didn’t actually follow through this time. The war dragged on. Oil prices began creeping up again, eventually topping $110. On cue, Trump announced that his administration had had “very good and productive conversations” with Iran toward ending the war. And once again, the price of oil dropped, this time down to about $95 a barrel. This pattern has played out repeatedly: rising oil prices, followed by an announcement of an imminent peace deal, followed by falling oil prices, followed by the war not ending.

[Read: No way to make a deal]

The TACO theory turned out to have two core limitations. First, it can become self-negating. Oil traders assume that higher oil prices will force Trump to end the war. But that assumption is what keeps the price of oil lower than it otherwise would be. Second, the theory can be easily gamed. Trump probably understands that the markets expect him to chicken out. So as soon as prices start rising, he can act like he’s about to yield, and everything will cool down. When prices fall, the traders who bet against TACO will lose big. They’ll think twice about making the same bet again next time. “So we end up in this endless merry-go-round,” Johnston told me. “Prices rise, Trump talks about a deal, prices fall, and then Trump suddenly feels like he doesn’t actually need to make the deal.”

This strategy, analysts assured me, can’t last forever. Markets are starting to catch on. Johnston pointed out that the impact of Trump’s peace announcements on oil prices has been diminishing over time, as traders begin to recognize the pattern. Even more important, the law of supply and demand will eventually become unavoidable: Countries are running through their stockpiled oil reserves quickly and could begin to exhaust them over the next month. At that point, there won’t be enough barrels to go around, and buyers will start bidding up the price of the remaining ones. “It’s a ticking clock,” Gregory Brew, the Eurasia Group’s senior analyst for Iran and energy, told me. “We’re losing 13 million barrels of oil every single day. Eventually that reality is going to set in. And when it does, prices are going to rise very very fast.”

Trump could also start feeling other forms of economic pressure. The April U.S. inflation report showed that prices were rising at their fastest pace since mid-2023, thanks largely to the fallout of the Iran war. This made bond investors nervous. Higher inflation is extremely risky for the holders of government bonds because it creates the risk that any debt paid back in the future will be worth much less than it is today. Many bondholders therefore responded to the report by selling their U.S. Treasuries. This in turn caused the interest rate on government bonds to rise to their highest levels in nearly 20 years. (A bond sell-off lowers the price of bonds, which mathematically causes their “yield”—the interest rate they pay out—to increase.)

“Equity investors have been ingesting large amounts of hopium that any day now Trump will come to his senses and call the war off,” Jared Bernstein, the former chair of Joe Biden’s Council of Economic Advisers, told me. “But bond traders generally don’t have patience for hopium. They are much more likely to respond to what’s actually happening right now in the economy.”

Even chaos in the bond market, however, doesn’t seem to be affecting Trump as much this time around. The same week that Treasuries hit a record high, Trump declared that he had ordered his staff “not to rush” to reach a settlement with Iran. For now, the TACO equilibrium continues to hold.

The King of Queens

2026-05-28T12:00:00-04:00

President Trump delights in playing what he calls “the gay national anthem” whenever he wants to rev up a crowd. He’s obsessed with Elton John, was once friendly with Liza Minnelli, and has a Liberace-esque flair for gilded interiors. One of his favorite sports to watch—mixed martial arts—is basically sweaty, semi-naked dudes. And he is a deep and vocal admirer of the physique of fellow men, often announcing which ones he would cast in a movie: “They’re perfect specimens,” he said last year of the military pilots who had visited him in the Oval Office; “He looks like the Marlboro Man,” he cooed about a former Iowa state senator; “Young, handsome guy. It’s always nice to be young and handsome,” he complimented the president of Paraguay.

Some of Trump’s allies note that years before gay marriage was legalized, Trump had gay friends, took pro-gay stances, and allowed gay people to join his private club in Palm Beach starting in the mid-1990s. Ric Grenell became the first openly gay person to hold a Cabinet position when Trump appointed him acting director of national intelligence. Grenell, who is now the president’s envoy for special missions, once called Trump “the most pro-gay president in American history,” a title that Trump said he was honored to have.

To be clear: Trump says he is attracted only to women and, in fact, has been married to three of them. He once hosted the Miss Universe pageant, was caught on tape saying that he loves to grab women “by the pussy,” and was found civilly liable for sexually abusing a woman. Loads more have accused him of sexual misconduct. (Trump has denied the accusations.) “Women—I like. Men—no, I don’t have any interest,” Trump affirmed at a Board of Peace meeting earlier this year.

But there’s also little doubt that Trump has unabashedly embraced the aesthetic—the je ne sais quoi—of a certain kind of gay man. Some who are sympathetic to the president have gone even further. Blaze Media, a conservative outlet started by the talk-radio host Glenn Beck, ran a story in 2024 headlined “Donald Trump: Our First Gay President,” much in the way people talked about Bill Clinton as having been the first Black one. The story notes, in a section titled “Queen of Queens”: “He blows kisses to Hulk Hogan, weighs in on Fashion Week (‘used to be so glamorous and exciting! No stars, no fun—just boring’), and his rivalry with lesbian Rosie O’Donnell remains a gem of the catty naughties social feuds.” Pod Save America, a liberal podcast started by former aides to President Obama, declared that Trump would be a gay icon, if only he had “liberal social values.” The president, the episode’s title observes, “DEMANDS a Ballroom at the White House, Loves Musicals, & Wears Make-up.”

James Kirchick, the author of Secret City: The Hidden History of Gay Washington, told me that Trump’s personal story, a guy from Queens making it big in Manhattan, tracks with the “typical gay story” of men of his era. In another life, he continued, the 79-year-old could be a classic aging gay, “living in Wilton Manors, sitting at a bar, making bitchy comments to everyone who comes in.” (Of course, Trump’s perch from the Oval Office confers much more power than a bar stool does, and his comments have moved markets and sent allies reeling.) “It’s a gay man frozen in amber in the late 1970s and early 1980s, before AIDS,” Kirchick said, referring to the type of gay man he believes Trump would embody. “It’s a certain age and a certain era. It’s very campy.”

The comedian and podcaster Caleb Hearon deemed Trump to be of the “old-school-gay” era, “because, you know, gay guys used to be mean before media training,” he said in an interview with Ziwe Fumudoh on her YouTube comedy show. The president, Hearon continued, should have become “a red-carpet fashion adviser,” the sort who would say things like: “That dress, honey. I don’t think so!” “That would have been amazing. I would have watched every night,” he said. “Instead, he ran for office on a platform of mass deportation, so that’s where things got tricky, obviously.”

[Read: The surprising reason for the new homophobia]

People close to Trump say he has long been gay-friendly in his actions as a private citizen. In the early days of his career as a developer, Trump was mentored by Roy Cohn, the legendary and ruthless New York lawyer and political fixer, who was gay. During Studio 54’s heyday, Trump relished making cameos. In 2024, Trump quietly allowed a gay wedding at Mar-a-Lago, although he didn’t attend.

But Trump has also been willing to vilify transgender individuals, especially athletes, for political gain. The ACLU has issued a scathing assessment of Trump’s record on LGBTQ rights, and the Trevor Project, which supports LGBTQ youth, said that outreach to its crisis hotline skyrocketed—a 700 percent increase—the day after he was elected a second time. Jonathan Lovitz, a senior vice president at Human Rights Campaign, wrote to me in an email that LGBTQ+ people helped profoundly shape the culture that Trump experienced while coming of age in New York City. That’s why, he continued, many queer people are offended when Trump engages in certain forms of camp: “Not because it’s tacky (which it is), but because it underscores a deeper contradiction: he wants the benefits of a country and culture that queer people helped create, while advancing policies that make those same people less safe every day.”

Trump’s continued patter about men’s bodies has also drawn attention. As my colleague Marie-Rose Sheinerman and I dug into examples of these corporeal appraisals, we were surprised by their sheer quantity and just how much Trump seems to delight in complimenting other men. He has given the compliment of “handsome” at least 68 times so far in his second term—or 69 times, if we count the two Thanksgiving turkeys he also collectively described as such. He is unapologetic in his preference for Cabinet members and administration officials who seem to come out of “central casting”; he praised Treasury Secretary Scott Bessent, who is gay, for his Hollywood-worthy bona fides, before appreciatively noting that “under that beautiful exterior is a killer.”

He can almost never resist commenting on the physique of brawny men: “Look at the muscles on this guy!” he said, gazing upon a young cadet while delivering the commencement address at the U.S. Coast Guard Academy last week. Two days later, he took pains to praise the New York Giants quarterback Jaxson Dart, calling him a “beautiful guy” and waxing poetic about his “legs like tree trunks.” And speaking about the golfer Arnold Palmer in 2024, Trump managed to both reassert his preference for women while also remarking on the legend’s masculinity: “I love women, but this guy—this guy—this is a guy that was all man.” (He also noted Palmer’s powerful swing with “stiff-shafted clubs,” and his, um, alleged other assets: “When he took showers with the other pros, they came out of there—they said, Oh my God, that’s unbelievable.”)

[Read: The question that the lawyers representing trans athletes didn’t answer]

Unsurprisingly, late-night hosts and comedians have been eager to tango with Trump’s inner gay. Bransen Gates, an actor and a social-media personality, has become known for his Instagram videos in which he takes snippets of Trump’s speeches and vampishly lip-synchs them—mouth pursed, eyes wide yet coy, finger wagging—under archetypes such as “The straight man speaking at graduation who is ‘definitely not gay’” and “When you have a crush on a guy named Stephen” (Miller, in Trump’s case). In perhaps his best-known video, aptly titled “Tr*mp was born to be a gay man,” Gates reprises Trump’s comments at an October 2020 campaign rally. “I’ll kiss every guy—man and woman, man and woman,” Gates-as-Trump says, complete with sexually suggestive winks, eye rolls, and light shimmies. “Look at that guy, how handsome he is. I’ll kiss him, not—not with a lot of enjoyment, but that’s okay.”

In a March Fox News interview, Trump was asked about the sexuality of Iran’s leader, the sort of highly sensitive question that nearly any other president would have handled with utmost care. Instead, Trump somehow pivoted to how “the Palestinian regime” is bad for gays—“Who are the gays for Palestine?” he mused—and later laughingly noted that one of his rally songs, “Y.M.C.A.,” by the Village People, is considered “the gay national anthem.”

“I did very well with the gay vote, okay?” he told the hosts. (The “gay vote” is a difficult thing to measure, although a variety of polls found that in both the 2020 and 2024 elections, Trump did have some gay support. However, a majority of voters who identified as LGBT preferred his Democratic opponents.)

Paul Baker, the author of Camp!: The Story of the Attitude That Conquered the World, told me over email that when it comes to Trump, making the distinction between camp and campy is important. The latter is the more self-conscious, ironic adoption of camp. But Trump is “the original, pure form—it’s when someone’s behaviour is outrageous, excessive, subversive and unintentionally funny,” he said. “The person doesn’t realise they’re funny or that they’re camp. They’re just being themselves.”

The risk, he continued, is when camp becomes a distraction from the president’s actual policies, such as executive orders and actions that could negatively affect LGBTQ health. Upon returning to office, for instance, Trump rescinded nondiscrimination protections for LGBTQ youth in school, which advocates say could worsen their mental health. “Laugh at him on Instagram all you like, but don’t let that take away oxygen from crucial topics like electoral reform, protecting democracy, gun control, immigration, healthcare and access to education in the US,” Baker concluded in his email to me.

[Read: The YOLO presidency]

Kirchick’s husband, Josef Palermo, was the Kennedy Center’s first curator of visual arts, until he was laid off after Trump took control of the cultural institution. (Palermo forwent a severance agreement to be able to publicly share—including in an essay for The Atlantic—his observations about the decimation of the Kennedy Center under Trump’s leadership.) Before Palermo lost his job last year, the two attended the Kennedy Center Honors, which Trump hosted, and Kirchick discovered that he prefers Trump more as a gala emcee than as a political leader. Kirchick said that Trump was “great” in the role, describing him as “a combination of Joan Rivers and Don Rickles.” He added wistfully: “I wish he could just do that.”

Marie-Rose Sheinerman contributed to this report.

*Illustration sources: Roberto Schmidt / Getty; Christian Rose / Roger Viollet / Getty; Echoes / Redferns / Getty; Jack Robinsonv / Hulton Archive / Getty

The Blue-State Delusion Over Unions

2026-05-28T09:45:26-04:00

What are public-sector unions for, exactly? What problem are they supposed to solve? That’s the question I found myself asking earlier this month, when the best-paid railroad workers in America went on strike for three days.

To be clear, I get what the unions understand their purpose to be. It’s to get the best deal for their members. That’s what they’re designed to do, and they do it well.

Salaries at the Long Island Rail Road—a commuter-train system that connects suburban residents to New York City—now average $121,646, which is 50 percent more than the median household income in New York City ($80,483). Work rules entitle engineers to double or even triple pay when they drive different types of trains on the same day or when they deliver a train to the maintenance yard after driving passengers. Last year, more than 300 LIRR workers each earned $100,000 in overtime—in addition to their base pay. Those extra wages in turn inflate their pensions, which they can take at the age of 55 after 30 years of service.

All of this is as good for union members as it is unimaginable for most American workers. But taxpayers and commuters are the ones who pay for those generous compensation packages, and it’s reasonable to wonder whether they are getting a fair deal.

To her credit, Governor Kathy Hochul pushed back on the LIRR unions. But she quickly settled the strike on still-to-be-disclosed terms that will keep in place massive overtime payments, expensive work rules, and bloated pensions. That’s business as usual in blue states and blue cities, where public-sector unions wield fearsome political power.

[Michael Podhorzer: The paradox of the American labor movement]

None of this is inevitable. Strong unions persist because roughly 30 states have passed laws requiring collective bargaining with public workers. If this process advanced the common good, all would be well. But the available research suggests that it doesn’t. To the contrary, unions routinely insist on pay packages and work rules that degrade the efficiency and effectiveness of the public sector.

Our laws aren’t doing a good job, in short, of aligning union incentives with the public interest. That’s a big problem, especially as our most vibrant cities struggle to provide good schools, effective policing, and high-quality transit. Reform is long overdue. Thankfully, it’s also achievable.

For many union members, it’s completely obvious why we have collective-bargaining laws. “The training process for this job is over a year long,” explained one LIRR engineer on the picket line. “It consists of multiple examinations. Some of the written ones are incredibly difficult. We are very qualified. And, you know, frankly we deserve this money.”

We deserve this money. What should the public make of this argument?

In a market economy, compensation isn’t normally keyed to what a worker deserves in the abstract. It’s linked, instead, to what an employer has to pay to attract high-quality workers. An employer that pays too little will find itself with too few workers or workers who are bad at their jobs. An employer that pays too much risks being driven out of business by more cost-conscious rivals.

There’s nothing intrinsically fair about the resulting wage distribution. Because, from an employer’s perspective, the goal isn’t fairness. It’s running a successful business.

In the private sector, unions temper that unfairness by pushing corporate owners to split profits with workers. But private-sector unions can push only so hard: If they insist on compensation packages and work rules that make the business go bust, they could find themselves out of a job.

Matters are different in the public sector. The Long Island Rail Road, for example, is owned and operated by the government, much like public schools and police departments. As a result, the unions representing public workers aren’t constrained by the possibility of corporate bankruptcy. They’re constrained instead by politics.

Which means that politicians have to decide how to compensate government workers. One approach, favored by unions, is to depart from the baseline set by the market and pay workers what they deserve. It’s an appealing idea. Public workers do crucial work and ought to be compensated fairly for it.

The trouble, of course, is that there’s no end to claims about deservingness. Pretty much everyone thinks they’re underpaid and underappreciated. Sometimes they’re right; sometimes they’re not. But I don’t know what a teacher or a cop or a railroad engineer “deserves,” nor does anyone else.

Giving public-sector workers what they think they deserve, moreover, clashes with how everyone else in the economy gets paid. Is it fair for one group to get special consideration just because they happen to work for the government? Especially when taxpayers—working people themselves—are picking up the tab?

During negotiations with the railroad union, Hochul suggested that the answer is no: “Workers deserve to be paid fairly for their work,” she said. “But at the same time, we must be responsible with public funds and the fares paid by Long Island residents.”

That’s the right approach. When the government supplies public services, its goal should be to supply those public services as efficiently as possible—not run a tax-and-transfer system to aid the relatively small number of people lucky enough to be union members.

There is a better argument for public-sector unions, which is that unions have the leverage to demand compensation packages and work rules that are necessary to attract excellent public workers. Here’s Randi Weingarten, the long-standing head of the American Federation of Teachers: “If we want to recruit and retain high-quality teachers, it starts with a fair wage, adequate working conditions, and the resources and support to succeed.”

There’s a lot to this. The public sector, like the private sector, is only as good as its workforce. If unions help attract better teachers and cops, collective bargaining might improve the quality of public services. We should be happy, on this view, that unions are fighting for government workers. We’re all better off as a result.

Except that’s not what the research shows.

Start with schools. Two comprehensive reviews of the available evidence, one from 2025 and one from 2015, find that teachers’ unions reliably increase school spending, especially on salaries for veteran teachers. In general, however, they do not appear to help kids. “Most often,” the 2025 review says, “teachers’ unions have no impact or a slight negative impact on performance.”

Recent experience in Wisconsin is revealing. In 2011, Republicans passed a law, Act 10, that curtailed collective-bargaining rights for teachers. In the immediate aftermath, student outcomes suffered, mainly because of a sharp increase in teacher turnover. But that dip was short-lived.

Since then, a series of studies have suggested that Act 10 has improved student performance. Barbara Biasi, an economics professor at Yale, found that test scores rose when districts ditched seniority-based pay in favor of a more flexible approach. Morgan Foy of the University of Illinois found similar gains in test scores and attendance even in districts that didn’t adopt a flexible pay scale—because, he suspects, teachers worked harder when unions couldn’t protect them from discipline. And E. Jason Baron at Duke has shown that the promise of higher entry-level wages enticed more young Wisconsinites to get a teaching degree, which has improved the talent pool.

Now consider policing. In 2003, sheriffs’ deputies in Florida secured collective-bargaining rights because of an unanticipated court decision. Researchers at the University of Chicago Law School took advantage of that natural experiment by comparing sheriffs’ offices with municipal police departments that were unaffected by the court decision. Collective bargaining, they found, caused a roughly 40 percent increase in violent misconduct in sheriffs’ offices relative to police departments.

That’s the opposite of what you’d expect to see if public-sector unions made public services better. But it’s consistent with the general run of the evidence about policing. One forthcoming study, for example, finds that the extension of collective-bargaining rights significantly increased the number of civilians killed by police, especially nonwhite civilians, and “can explain 14 percent of all non-white civilian deaths by legal intervention between 1959 and 1988.”

To put it mildly, these results are hard to square with the claim that public-sector unions improve the public sector. At least three factors seem to be driving those results.

First, unions often push for job protections that frustrate workplace accountability. In the study of Florida sheriffs’ deputies, for example, collective bargaining appeared to cause a rise in violent misconduct, because of “a reduction in expected sanctions.” In other words, sheriffs’ deputies knew they could get away with it.

Second, unions push to equalize pay among their members based on seniority and credentials, not on quality of performance. That makes recruiting talented young people difficult, and rewarding good workers impossible. The Wisconsin reforms, for example, “led younger and less credentialed teachers to earn more on average, and older, more experienced teachers to earn less.” That’s bad for aging union members, but good for students.

Third, public-sector unions avidly negotiate for compensation in the form of pensions, not wages. But pensions are a poor recruitment tool: Starting wages matter much more to young people than pensions that will be paid out decades down the line. When unions use their power to boost pension payments, they aren’t working to attract talented young people. They’re working to reward their members.

If we want unions that actually improve the quality of public services, we’re going to have to reform our collective-bargaining laws.

[Jonathan Chait: The wrong way to win back the working class]

As matters stand, those laws require state and local governments to negotiate with unions. But they also establish what those unions are entitled to negotiate over—what is “bargainable.” And a very wide range of terms and conditions of employment are typically bargainable. That’s how you get demands for job protections, pay equalization, and hefty pensions.

None of that is graven in stone. The laws could be amended to limit the scope of what’s bargainable. Overtime, pensions, work rules, salary schedules—all of those would be off-limits. Unions would be left to negotiate over the one thing that is most likely to attract high-quality workers: base wages.

In that world, unions would still be powerful. They would still serve as a counterweight to local governments that might try to balance their budgets on the backs of middle-class workers. Their members would still receive job protections under civil-service laws. The unions just wouldn’t be allowed to make demands that frustrate the delivery of high-quality, cost-effective public services.

Reformed collective-bargaining laws would bring what unions want into better alignment with the public interest. Otherwise, we’re left with the LIRR engineer’s argument about what the unions are for: We deserve this money. The engineer may be right about what he deserves. Surely we all deserve better in this fallen world. But it’s no way to run a railroad.