If President Trump has one true skill, it is conjuring up alternate realities and getting his supporters to accept them. The examples are everywhere: dismissing stories he doesn’t like as “fake news” from corrupt media, casting the January 6 rioters as victims, election denial. After he baselessly insisted that there had been rampant mail-in voter fraud in the 2020 presidential election, 11 states made mail-voting access more restrictive. His repeated false claims in 2024 that Haitians in Springfield, Ohio, were eating pets led to real threats against the community. Trump builds a world, and by virtue of his power and influence, all of us end up living in it in some way or another.
That era may be coming to a close, in large part because of the way he and his administration have bungled the war in Iran. Trump seems unable to spin this conflict, divert sufficient attention away from it with a new controversy, or end it outright. His social-media posts, which once acted as a de facto assignment editor for the mainstream media, no longer command the same attention or respect. Now, interspersed with AI slop and inexplicable middle-of-the-night reposting binges, Trump’s Truth Social screeds sound more like the fevered ramblings of a radicalized Facebook uncle. For the past five months, he’s been trying to convince the American public (or at least his Truth Social followers) that the war is over, or ending, or a rousing success. His administration is even retroactively changing on the Pentagon’s website the number of American service members who have died in the conflict. But no one seems to be falling for it. His posts don’t have the juice. His alternate universe is no match for the harsh reality of war.
In the Trumpian vision of the world, the U.S. invasion of Iran was not only necessary but also urgent: In February, Trump alleged that Iran was attempting to rebuild its nuclear-weapons program and had continued developing long-range missiles that were threatening U.S. allies and could reach the United States “soon.” But Trump’s claim is dubious, at least according to a March 2025 U.S. intelligence assessment, which noted that “Iran is not building a nuclear weapon.” Experts also cast doubt on the strength of Iran’s missiles: “There is little evidence that Iran could build missiles that reach the United States in the near future,” Emma Sandifer of the Center for Arms Control and Non-Proliferation told FactCheck.org in March.
Justifications aside, the attacks, which were not authorized by Congress, were immediately unpopular with Trump’s base, which elected him in part for his isolationist tendencies. (On Election Night 2024, Trump even remarked, “I’m not going to start a war. I’m going to stop wars.”) Iran’s predictable retaliation, closing the Strait of Hormuz and drastically cutting off crucial supply routes for oil and other natural resources, subsequently drove up prices for goods around the world—immediately exacerbating the unpopularity of the conflict and shocking the markets. The fallout appeared to take Trump by surprise, and in the early weeks of the war, he used his bully pulpit for damage control: On March 9, with Brent-crude futures spiking to nearly $120 a barrel, he gave an interview to CBS News suggesting that the “war is very complete, pretty much.” Oil futures swiftly plunged to less than $90 a barrel.
Trump used this tactic repeatedly. On March 20, he posted on Truth Social that the U.S. was “close to meeting our objectives” in Iran. Oil futures dropped. On March 23, he signaled that discussions between the U.S. and Iran were productive, leading Brent crude futures to plummet by 13 percent. Each time, Trump offered no real proof of his claims, yet markets responded to his version of the story. An NBC News analysis found 32 examples of the president suggesting that the war was coming to a close. “We don’t need people that join Wars after we’ve already won!” Trump posted on March 7, in response to the United Kingdom considering sending military support. In late March, Trump posted that operations were “winding down”; that same day, he rejected a suggestion from Pope Leo that there should be a cease-fire. In early April, he posted that Iran had asked for a cease-fire; later, Iran called his claims “false and baseless.”
Tenuous agreements between the U.S. and Iran have been broken repeatedly. On July 12, the U.S. resumed heavy air strikes on Iran following an Iranian attack on a container ship in the region. As my colleagues Nancy A. Youssef and Jonathan Lemire recently reported, “The Trump administration is back at war. It just doesn’t want to admit it.” The commander in chief has become the boy who cried “Mission accomplished.”
As a result, Trump’s posts have become more erratic. On June 23, in response to rising gas prices, Trump posted on Truth Social, in all caps, to nobody in particular, “EVERYONE, KEEP OIL PRICES DOWN. I’M WATCHING! YOU’RE PLAYING RIGHT INTO THE HANDS OF THE ENEMY. DON’T DO IT!” And his posts no longer seem as able to influence the market. A recent report from Axiosclaimed that bond-market traders have stopped paying strict attention to Trump’s social-media feeds to read the economic tea leaves: “Researchers from JPMorgan show that for the most part, his social media posts are noise, with very little ability to move the market.”
And yet, Trump is attempting to double down. Last week, Truth Social announced that it plans to sell early access to posts from certain accounts (presumably including the president’s) and is considering charging as much as $100,000 a month. The clear implication is that Trump or his close associates will share government news that will move markets, giving paying subscribers a distinct first-mover advantage. Trump currently owns 41 percent of the stock through a family-run trust in Trump Media and Technology Group, the owner of Truth Social. Thus, the decision to charge access is akin to the president’s family profiting off his position as leader of the free world. (Trump Media and Technology Group did not respond to a request for comment.)
Trump is president, so obviously, any statement he makes attracts attention. But simply being president has never been Trump’s superpower. His chief skill, first as a mogul with an uneven history and then as a politician, is molding his image in a way that reflects how he sees himself: as the genius, adored, yet constantly victimized center of the universe. This, of course, is delusional. Yet somehow, he’s managed to make that fantasy at least partially real: He has, after all, been the main, divisive figure in American life for more than a decade.
With Trump, it can feel as if there are no limits—to his shamelessness, his lies, his graft, or his ability to sow chaos, lose interest, and move on to the next outrage. But it’s possible that we’re finally glimpsing the limitations of his ability to control the narrative. For too long, Trump has been the force that made reality feel flimsy, pliable. But in the Iran conflict, he’s met an immovable object. Now reality may just be catching up.
Editor’s note: This work is part of AI Watchdog, The Atlantic’s ongoing investigation into the generative-AI industry.
Amid tech companies’ ongoing efforts to vacuum up as much data as possible to train AI models, Meta has taken public posts from its own platforms, scraped massive amounts of content from the rest of the internet, and pirated millions of books to train its AI models. Now the company is being accused of downloading much more illicit material: huge volumes of porn, nonconsensual celebrity nudes, blueprints for 3-D-printable handguns, and millions of passwords acquired by hackers.
Strike 3, the parent company of Vixen Media Group, a prolific producer of adult films, is suing Meta for allegedly downloading 2,973 of its copyrighted videos. But in its efforts to collect evidence about these purported downloads, Strike 3 also captured information relating to many other files that Meta may have acquired over the course of two years: They include images from “Celebgate,” a 2014 hack that resulted in the leak of private photos belonging to Jennifer Lawrence, Kirsten Dunst, and others, plus several collections of deepfake celebrity porn featuring the faces of Natalie Portman, Scarlett Johansson, Elizabeth Olsen, Gal Gadot, and others.
Strike 3’s file-transfer lists include ordinary movies and TV shows, alongside dozens of videos from GirlsDoPorn, which was shut down after six people associated with the site were charged with sex trafficking. Meta allegedly downloaded several individual episodes from the website and two “GirlsDoPorn MegaPack” archives. Strike 3 also presents evidence that Meta made some content publicly available in addition to downloading it.
When I reached out to Meta, a spokesperson told me via email that “these claims are bogus.” Meta also noted in a court filing that Strike 3, which regularly files lawsuits against people who download its work illegally, “has been labeled by some as a ‘copyright troll.’”
This is a complicated case. Piracy can be hard to track with precision, and both parties are clearly conflicted: Strike 3, in seeking damages, wants Meta to look bad, and Meta would prefer not to be associated with this kind of content. Yet if Strike 3’s data are accurate, they would show that participating in mass piracy is a by-product of modern tech development—whether or not any of the material collected is explicitly used to engineer AI models or other programs.
Although Meta acknowledged the possibility that employees, contractors, or even visitors to Meta’s offices may have used the company’s networks to download the porn in question, its defense in court was that the material was accessed for “personal consumption” rather than as part of a company project. But neither the sheer volume nor the pattern of downloads looks like personal consumption. The file-transfer logs produced by Strike 3 show sequences of files that are unrelated except for a certain keyword or concept. For example, two files that were transferred consecutively on June 2, 2023—“Bajillion Dollar Properties S01E07” and “First Time Home Buyer Anal Fantasy”—are both loosely themed around real estate, but the similarities likely end there. Judge Eumi Lee, who denied Meta’s motion to dismiss the case, cited other such juxtapositions, such as consecutive downloads of “Teenage Mutant Ninja Turtles (1987-1996)” and a video labeled “Teen Sex Sessions 2 (2012).”
Strike 3 has argued that the content could be used for AI training. A Meta spokesperson told me, “We don’t want this type of content, and we take deliberate steps to avoid training on this kind of material.” In its legal defense, Meta pointed out that its terms of service prohibit users from using its AI products to generate images containing pornography. Even if Meta isn’t building a porn-generating bot, however, tech companies can use “unsafe” content to test guardrails for their systems—effectively telling the software what not to generate. (Lee noted that Meta’s terms of service are irrelevant to the question of whether it wants to acquire porn.)
It’s also possible that Meta is compiling an archive of material for no precise purpose, just as Anthropic acquired millions of books that it says it had no intention of using for AI training but wanted to keep for its “research library.” As AI-training techniques advance, there is no telling what kind of content a company might find useful in the future.
Nineteen of the allegedly downloaded files contain models of functional handguns that can be produced with 3-D printers. A file labeled “5.7 million passwords list” purportedly contains 5,718,107 passwords from accounts that were hacked from 2015 to 2019. The downloads also include password-cracking tools that can be used by hackers to break into people’s accounts. Other files include the movies BlacKkKlansman, The Banshees of Inisherin, and everything made by Studio Ghibli from 1979 to 2020; music by Dua Lipa, Elton John, and Paul McCartney; radio shows and podcasts such as The Howard Stern Show and 1619; and software including Microsoft Windows, Adobe Photoshop, Ableton Live, and NBA 2K23.
The files were downloaded through BitTorrent, a system that allows people to share files among themselves. When you download files via BitTorrent, you typically share those files with other people at the same time; if nobody is sharing the files, then nobody can download them. According to Strike 3’s data, which lawyers from the company told me it regularly gathers in an effort to track the illegal distribution of its intellectual property, devices on Meta’s networks also made some of Strike 3’s videos available to others—a practice known as “seeding”—for as long as three months after fully downloading them.
Meta additionally used BitTorrent to acquire large quantities of copyrighted books around the same time, as shown in testimony from another lawsuit. Some Meta employees were uncomfortable with this activity. According to court documents, Jelmer van der Linde, a former Meta employee who was asked to download books to train the company’s AI models in 2024, told his supervisor in a work chat message that it seemed like a “very very dark grey area. Legally, ethically …” He added that torrenting “copyrighted material is not very legal either.” He was transferred to a different project, and another developer downloaded the books instead. He has since left Meta and now works at the Ellison Institute of Technology in Oxford.
To verify Strike 3’s data, I reached out to Tom Chothia, a professor of cybersecurity at the University of Birmingham who has studied BitTorrent. He reviewed a technical description of Strike 3’s data-collection methods and told me that, assuming the records weren’t tampered with, “they have established that people at Meta were uploading” Strike 3’s videos.
One thing that complicates the case is that most of the torrenting did not occur over Meta’s corporate networks. Strike 3’s data show sequences of related file transfers across several networks, including Meta’s. Strike 3 argues that the other networks were used remotely in a coordinated way, perhaps to make the activity harder to trace. Meta has called the patterns a coincidence. But Lee found them compelling, writing that Meta’s coincidence theory “strains belief” and that the activity is likely indicative of “algorithmically coordinated behavior.”
Nearly 300 of the transfers were traced to a private residence in Mountain View, California. Strike 3 claims that this was the home of a Meta contractor’s father, and that the file transfers stopped when the contractor stopped working for Meta. In response, Meta argued that these downloads were “plainly indicative of personal consumption.” According to Strike 3’s logs, the files downloaded to the residence include, in addition to porn, Russian-language books, Chinese movies, and a wide variety of cracked software, among other things.
Unfortunately, in the AI era, illegal mass-downloading by tech companies is not uncommon. Other lawsuits have revealed that OpenAI and Anthropic have also used BitTorrent to download copyrighted books. Whether they have uploaded those books as well, or what other content they may have downloaded through BitTorrent, is not yet clear.
BitTorrent is a strange world in which to find respectable megacorporations. It is dominated by media piracy, and very little legitimate file-sharing occurs. But if AI models are intended to replace human creators, then they have to be trained on as much human-created content as possible. BitTorrent is perhaps the fastest and cheapest way to download high-quality work in large quantities. AI companies have claimed that their models can create original work, but so far, the development of artificial intelligence has involved a lot of theft of other people’s intelligence.
Nothing captures the decline of Tesla better than the Cybertruck. When the stainless-steel pickup hit the road in 2023, Tesla was the world’s biggest electric-car company. At the time, Elon Musk predicted that he’d eventually sell 250,000 Cybertrucks a year. In the first half of 2026, according to one estimate, Tesla sold roughly 7,000.
Tesla has faced several self-inflicted wounds. Musk’s affiliation with Donald Trump and his tenure at DOGE was a turnoff for electric-car buyers who lean to the left—which is most of them. A promotional event with the president himself on the White House lawn didn’t seem to do much good. Tesla’s worldwide sales fell 9 percent in 2025. The problem is not just a matter of Musk’s politics: He has gotten pretty bored with selling cars—or at least ones with steering wheels. Besides the ill-fated Cybertruck, Tesla hasn’t released a brand-new passenger model since 2020 as Musk has shifted the company’s focus to robotaxis and robots.
It can be easy to write off Tesla, or even forget about it entirely. All eyes are now on Musk’s other trillion-dollar venture, SpaceX, with its grandiose goals of launching data centers into orbit and colonizing Mars. But Tesla has become more important—not less—to the future of electric vehicles in the United States. Of the roughly 463,000 EVs that Americans bought in the first six months of this year, more than half were Teslas, according to figures from Cox Automotive—a jump of nearly 8 percentage points from this time last year. The Tesla Model Y alone claimed more than a third of EV sales.
In some ways, Tesla’s comeback has more to do with the failures of other car companies than its own success. Last year, Congress nixed the EV tax credits meant to encourage Americans to go electric. The Trump administration followed suit by rolling back regulations on tailpipe emissions. The entire automotive industry has been scrambling as a result. EV sales have cratered by about 24 percent so far this year, though they are showing some signs of recovery.
Tesla isn’t immune to any of this. In the U.S., its sales slumped by an estimated 11 percent through June of this year. But the broader downturn makes Tesla that much more crucial to keeping the electric dream alive. In recent months, numerous automakers have yanked EVs from the market, delayed new arrivals, and canceled their plans to debut new EVs. Last year, Ford axed the F-150 Lightning, a battery-powered pickup that was supposed to anchor Detroit’s electric future. (Ford is hyping up a cheaper electric pickup truck—but it’s still in development and won’t come out until next year.) Just last week, Honda pulled the plug on its last remaining fully electric option, the Prologue SUV, after killing off a slew of upcoming models.
Then there’s the matter of price. The cheapest Tesla will run you a little under $39,000, which is hardly a bargain, but is thousands below the going price for a new car these days. It’s all the more remarkable considering that Tesla makes money from the cars it sells—a feat that has eluded much of the industry. Automakers continue to grapple with the high cost of batteries and of designing new kinds of cars. Consider that from 2022 through 2025, Ford lost $16 billion selling EVs. The company expects its electric-car business to simply break even in 2029. That’s partly why so many automakers sprinted for the haven of high-margin gas-powered vehicles once the going got tough.
As sales of new EVs flounder, used ones are picking up the slack: Americans gobbled up an estimated 128,000 of them last quarter, smashing records. Since Tesla has sold EVs in large numbers for so long—its first mainstream hit was the Model 3 sedan in 2017—it carries the market. One of the cheapest paths to an EV that can run 250 to 300 miles or more in a single charge is a used Tesla.
Musk’s car company is crucial to the future of EVs in another way. It dominates the infrastructure of the electric era: plugs. One of the biggest speed bumps to wider electric-car adoption is the challenge of where to recharge. Tesla has been blanketing the country with its Superchargers since its early days in 2012, and now operates about half of all fast-charging plugs in the United States. This is great marketing for Tesla, of course. But it’s also a win for almost anybody looking to give up gas. In recent years, Tesla has opened up most of its formerly exclusive stations to other EVs.
No matter what Musk says about Tesla’s AI future, he may have no choice but to pay attention to his car business. This week, Tesla reported that despite surging revenue, its operating profits in the second quarter of 2026 were the lowest in six years, in part because of massive expenditures on AI. Musk is also hemorrhaging cash at SpaceX. All of that money has to come from somewhere. Beyond America’s troubled EV industry, Tesla is doing more than just surviving. A spectacular second quarter of deliveries worldwide stunned Wall Street analysts and points to the beginnings of a bigger rebound. (Musk did not respond to my request for comment.)
The EV market would be in far worse shape right now without Tesla. But the real shame is that Tesla could be doing so much more to get drivers to go electric. The automaker has managed to hang on in the U.S. by serving up little more than the microwaved leftovers of its limited, aging lineup. While Tesla’s automotive innovation has stalled, Chinese companies are releasing one affordable, high-tech model after another. Perhaps Tesla could roll out a truly budget EV. You know, something like the $25,000 car Musk once touted onstage but reportedly shelved in favor of the Cybercab, a robotaxi without a steering wheel. A less polarizing, smaller pickup could probably do a whole lot better than the Cybertruck has. At this point, almost anything would clear the bar.
The air is bad this summer. Really, it’s historically disgusting. Smoke from wildfires has gusted over people’s homes, spreading haze and yellowing the sky. Last week, four midwestern cities—Chicago, Detroit, Milwaukee, and Toledo—reached the highest levels of particulate pollution they’ve recorded in two decades. For more than 100 million Americans across at least 20 states, the weather was grim: fumy with a chance of scattered lung conditions.
The air was bad like this two years ago, during that summer’s set of conflagrations—and let’s not forget the epic plumes of 2021, which turned the full moon over New York City a ghastly shade of red. Welcome to the new America: the land of burning throats.
In some respects, our skies have gotten cleaner in the past few decades; average levels of fine particulate matter have been cut nearly in half since 2000. But wildfires have been getting bigger and more frequent, and sending off plumes of air pollution that come and go in noxious episodes. The climate trend goes in only one direction, so this will be our future. That means we’ll need to have a better sense of when it’s safe to be outside, when it’s time to skip our morning stroll, and when we ought to keep our children home from camp. We’ll need to learn to live with smoke.
Enter the air-quality index: a weather metric for the age of wildfires. Perhaps you have never heard of the AQI or noticed it only last week, in news reports and weather-app alerts, when the smoke was getting thick. Chances are that you’ll be looking at it quite a bit in years ahead. But the AQI’s color-coded warnings, which run from wary orange (“unhealthy for sensitive groups”) to Grim Reaper maroon (“hazardous”), are useful only in the broadest sense: They may call attention to the risks of breathing smoke and remind you to be careful. As a tool, the AQI is inconsistent overall—its forecasts can be slow and imprecise, and its particulars don’t always have a bearing on direct experience. If wildfire events will get only more common, then this lousy measure must be fixed.
Globally, there’s no agreement on what constitutes unhealthy air—and no standard index of air quality. America’s version of the AQI was first created 50 years ago, as a way to standardize recording and reporting of pollutants and to clarify communications to the public. The system works like this: The concentrations of five pollutants—particulate matter, ozone, sulfur dioxide, nitrogen dioxide, and carbon monoxide—are measured throughout the day at thousands of monitoring stations across the country; the readings are then combined and measured against safety thresholds that are based on regulatory limits. The AQI score is assigned such that a value of 100—which is supposed to mark the line between safe conditions and potential hazards—corresponds to standards set by the federal government.
Even just that fact can be a problem, Kevin Cromar, the director of the Health, Environment, and Policy program at NYU’s Marron Institute of Urban Management, told me. The regulatory limits for pollutants are, in theory, a reflection of those pollutants’ risk to human health. But that connection may be indirect, and the thresholds certainly aren’t calibrated perfectly for use by members of the public, who might use the AQI to figure out whether it would be safe to take a walk.
Another problem stems from the distribution of air-pollution sensors around the country. Where these are abundant, as in major cities, the reliability of any given reading might be high. But other parts of the country are more like sensor deserts, where AQI scores are interpolated from readings made in far-off places, and precision is reduced. The variation in uncertainty is not communicated to the public. AQI reports tend to look the same—a single number—no matter where you are.
In some countries, air-quality scores are calibrated to reflect whatever mix of pollutants happens to be airborne on any given day. The United States is more single-minded: It focuses only on whichever pollutant has a concentration that most exceeds the standards, and disregards the rest. If, say, there was enough nitrogen dioxide on Monday for an AQI score of 105 and enough ozone for a score of 103, our air-quality reports would give only the former number: 105. Now imagine that on Tuesday, the nitrogen dioxide held at 105, while ozone levels dropped to 30. The resulting AQI report would look the same as it had the day before—again, 105—even if the air felt much better in the lungs of a person with asthma. Given this design, Cromar said, the meaning of the index may even be inverted from time to time, with a higher AQI corresponding to a lower risk of health effects.
This sort of issue would be easy to ignore under conditions like last week’s, when a single factor—particulate matter from wildfire smoke—dominated readings. But the mixed-pollutant problem points to a deeper issue with AQI: It can be hard to know what the numbers really mean. AQI reporting isn’t standardized across weather apps and news reports. In some places, you might see the measure as a total average for the day; in others, you’ll get a moving average (known officially as NowCast) that tries to show you how air quality is changing by the hour; still others, such as the basic weather app on my Google phone, offer up a mysterious hybrid of the two. Cromar told me that even people in his field—which is to say, the most sophisticated consumers of this sort of information—may be flummoxed by the range of AQI reports. “It’s a mess,” he said. “There’s just a lot of different information out there, and it’s unreasonable to expect the public to somehow make sense of all that.”
Then there’s the fact that different people may react in very different ways to the same concentrations of particulate matter, or the same readings on the index. In a smokier era, we’ll each have to learn for ourselves how the AQI numbers that show up on our phone relate to our sensations and experience. Cromar said that we’ll have to form new types of weather intuitions: Oh, the AQI is 135; I know what that feels like—I’ll put off my shopping ’til tomorrow.
Or maybe we could just get rid of AQI completely and have our apps provide us with raw measurements instead: a daily digest of pollutant concentrations, laid out in micrograms per cubic meter. At first, these data would be cryptic—who besides the biggest air-pollution nerd would know the difference between 35 μg/m3 of particulate matter (yellow-level risk, per the AQI) and 60 μg/m3 (red-level risk)? But then again, none of us was born with the understanding that 35 degrees Fahrenheit is cold and 60 degrees is pretty nice. I also had to learn, at some point in my adult life, that 35 percent relative humidity is very dry for where I live, whereas 60 is about average. This is true for every number in our weather apps: the daily highs and lows, the chance of rain, the UV scores, the pollen counts. You have to gain a feel for what they mean.
We’ve already formed internal models when it comes to measured temperatures, wind-chill factors, and the heat index. Now the weather’s turning, and it’s time to do the same for smoke.
Yesterday, OpenAI made an alarming disclosure: An assortment of its most advanced AI models, including one that has not yet been released, had autonomously broken out of the company’s internal systems and hacked into the databases of another tech firm, Hugging Face, to steal some information. OpenAI’s report seemed to augur the very sort of disaster that IT professionals have been warning about since last year, when Anthropic’s top models began demonstrating the ability to orchestrate and automate severe cyberattacks.
OpenAI’s models were not exactly trying to take over the world. The company says that it was running some routine evaluations in what is known as a “sandbox”—a walled-off environment that limits internet access, lest the bots simply search for the test’s answers. But these AI systems, including GPT-5.6 Sol, which is available to consumers, broke out of the sandbox and accessed the open web by exploiting a previously undetected vulnerability. The models discovered that the answers to this particular evaluation were on Hugging Face, a company that hosts and publishes many AI models and papers, and then broke in to steal them.
“We consider this incident to be an unprecedented cyber incident,” OpenAI wrote in a blog post, which also noted that the company is partnering with Hugging Face to investigate and address the issue. An OpenAI spokesperson pointed me to the blog post in response to a request for comment, and Hugging Face’s CEO, Clement Delangue, pointed me to an X post in which he wrote that the companies are collaborating and that “we strongly believe there was no malicious intent on their part.”
The first-order implications of the incident are the same that have arisen from countless AI-driven cyberattacks: Advanced agents are capable of more and more skilled hacking, and the internet is full of rickety and vulnerable code. Lax digital security may have been fine when hacking was a rare skill. But now the speed, scale, and sophistication of AI hacks mean that everything is vulnerable—tech companies, hospitals, banks, electrical grids, the military.
When Hugging Face first reported that it had been hacked last week, it noted that “autonomous, AI-driven offensive tooling is no longer theoretical.” That analysis stemmed from the company’s own investigation into the hack, for which it used a Chinese-developed AI model called GLM-5.2. That Hugging Face used this particular model actually underscores one of the major risks of this moment: GLM-5.2 is free to download, giving cyber capabilities to anybody with an internet connection and technical know-how. Defenses are not keeping pace with this new reality, and neither are AI companies: Prior to the release of GPT-5.6 Sol, a U.K. government agency tasked with evaluating advanced AI systems found universal jailbreaks—ways to hack the AI itself so it can take, for instance, criminal actions—during multiple rounds of testing. OpenAI said that it had mitigated those vulnerabilities, but the U.K. agency said that it expected future tests “to surface similar jailbreaks.”
But jailbreaks are clearly not even necessary for things to go haywire. The Hugging Face hack showed advanced AI systems deciding upon and then orchestrating a high-grade cyberattack on their own—the models working against, rather than according to, human instruction. Models are ruthless in their pursuit of a goal, sometimes taking disastrous shortcuts. Imagine, for example, a bot tasked with reducing the federal deficit that institutes an elaborate and hard-to-detect accounting gimmick, or a customer-service agent that issues absurd discounts in pursuit of customer-satisfaction ratings. In the case of GPT-5.6 Sol and the unreleased model, “all evidence suggests that the models were hyperfocused on finding a solution” for an evaluation, OpenAI wrote in its blog, “going to extreme lengths to achieve a rather narrow testing goal.”
This unwanted behavior is a predictable and alarming result of how the entire AI industry is developing its models. The past year’s advances in AI coding and agents—that is, all the frenzy about Anthropic’s Claude Code and OpenAI’s Codex—have been the result of an overriding emphasis in AI training through a process called “reinforcement learning.” This involves giving AI models lots of hard problems—math proofs, coding challenges, what have you—and then providing positive feedback for correct answers and negative feedback for incorrect ones.
In many reinforcement-learning paradigms, the ultimate aim is just for the AI to arrive at the solution—it doesn’t matter how it does so. This is the AI equivalent of Mark Zuckerberg’s infamous dictum to “Move fast and break things” in Facebook’s early years. Last year, while I was working on a profile about Anthropic, one of the company’s researchers, Joshua Batson, told me that the new reinforcement-learning-trained models are getting “really good at solving coding problems, but they end up learning to solve them at all costs,” making them “bloody-minded.” More recently, Anthropic reported that its own advanced cyber model, Claude Mythos Preview, had taken “reckless” behaviors to accomplish various tasks and also, in one instance, had broken out of a testing sandbox. In that case, Mythos had been asked to do so—but the model then posted details about the exploit, unbidden, on the internet.
To be clear, this sort of AI “reward hacking” is a known problem that tech companies are putting lots of effort into addressing. But these incidents keep cropping up regardless; if anything, research suggests that they may become more common. Meanwhile, OpenAI, Anthropic, and Google DeepMind are under tremendous economic pressure to make their models more and more capable, which means that the bloody-minded reinforcement learning is all but certain to accelerate.
Anthropic has poached such an array of high-profile professors that it has become a punch line in academia. “‘I’m joining Anthropic’ is the new meme right now,” Subbarao Kambhampati, a computer-science professor at Arizona State University (who has not joined Anthropic), told us. This month, the AI company hired the chair of UC Berkeley’s department of electrical engineering and computer science, presumably to help build more capable bots. Perhaps more surprisingly, Anthropic has in recent weeks also picked up a Stanford economist, a theoretical physicist from the University of Maryland, and an analytic philosopher from UT Austin.
AI companies are turning into something like mini-universities in their own right. OpenAI employs top mathematicians and physicists—including one who studies black holes, and another who specializes in string theory. At least three computer-science professors joined Meta’s AI lab in late June. DeepMind, like Anthropic, is home to a crew of philosophers. And Anthropic’s recent job postings indicate an interest in hiring legal scholars and political scientists. It’s unclear exactly how many current and former professors are working at AI companies. Across these four firms, we found more than 80—the majority of whom are computer scientists. Some have left academia entirely; others are still working part-time at a university. That number is likely a significant undercount, because we don’t have access to internal data; it also doesn’t include the many professors who have started their own companies, those at other AI start-ups, and those working with the industry in a less formal capacity.
Particularly for AI researchers, these companies have a strong gravitational pull. “Much of the important research is being done in industry now,” Humphrey Shi, a computer-science professor at Georgia Tech who joined Nvidia as a vice president last fall, told us. As he sees it, “If you want to do something that really, truly matters, you probably want to join one of those entities.” Tech firms are making offers—including very enticing salaries—that are hard for academics to refuse. In the process, research that previously would have happened in the open is getting locked up behind closed doors.
For decades, universities were the center of AI research. The field itself officially began at a gathering of researchers at Dartmouth in 1956, and federal funding provided much of the field’s early support. In the early 2010s, Silicon Valley executives began to take AI’s commercial potential more seriously, and set out to hire the best researchers. In 2013, Google paid $44 million to acquire a start-up run by a trio of AI researchers from the University of Toronto. Facebook then hired Yann LeCun, an NYU professor, to establish the company’s AI-research lab; Uber poached some 40 Carnegie Mellon researchers to work on driverless cars. But for the most part, even as more work was being done inside of private companies, many of the newly hired academics at tech companies retained their professorships and established a culture of open research. “Researchers will be strongly encouraged to publish their work,” OpenAI wrote in its founding announcement. (Note the organization’s name.) This norm helped lead to the current AI boom: In 2017, scientists at Google published a research paper that was immediately of interest to OpenAI. Google’s innovation, called the “transformer,” is what the T in ChatGPT stands for.
As AI has taken center stage, Silicon Valley has intensified its efforts to recruit star researchers—and looked beyond computer-science departments. Philosophers help train tech companies’ bots to better interact with humans, and economists study the labor-market implications of AI. Compensation is only part of the draw. The current era of AI research requires massive amounts of computing power. Universities have only a fraction of the resources that Silicon Valley can provide, and the chasm has widened as the Trump administration has cut back on scientific funding. Anca Dragan, a UC Berkeley computer scientist who heads DeepMind’s AI-safety-and-alignment department, wrote to us that she was partly motivated to take the job to acquire “the data, compute, and budget access to make progress on safety at the frontier.” Some professors who remain in academia are also forming partnerships with frontier labs or starting their own AI companies, in part so that they can pursue their research without resource constraints. And in some cases, would-be star Ph.D. students are dropping out of their graduate programs—or skipping them altogether—to pursue careers in AI instead.
Although plenty of research is still happening within universities, many professors told us, the result is a flywheel: As more academics get sucked up by industry, the center of AI research moves further from academia, thus increasing the incentive for remaining researchers to leave. The exodus could have some advantages. Many of the professors currently at the labs are on temporary leave; some will likely return to academia full-time. When they do so, they’ll bring new knowledge about frontier research to their institutions. (Dragan is currently preparing to teach a small Ph.D. seminar on AI safety.) Scientists at AI labs might also innovate at a faster pace. After all, they don’t have to worry about applying for grants or waiting years for an article to go through peer review. “If some of the best research is being done in these companies, then I kind of want colleagues and faculty members to be there,” Chris Gregg, a Stanford computer scientist, told us. “You go where the best research is being done, and if that happens to be at a company, so be it.”
But when academics decamp to AI labs or spend more of their time working on their own companies, universities are left with fewer professors to teach the next generation of researchers. (Gregg said that students occasionally ask why courses are no longer offered. The answer is sometimes that the professor is on leave at an AI company.) When star professors leave universities, it becomes harder for students to do work that helps them stand out to prospective employers, Shi said. To differentiate themselves from other applicants, it helps to have worked with mentors who are doing consequential or cutting-edge research. Jennifer Chayes, the dean of the College of Computing, Data Science, and Society at UC Berkeley, told us she fears that as research concentrates in labs with proprietary models, it will be harder for people outside the labs to use AI models to advance science. “Computer-science departments at universities will survive this,” she said. “I don’t know if our innovation economy will.”
As the AI race intensifies, companies now decline to release much of their research for fear of giving away their competitive advantage. Both Gregg and Chayes said they hear from researchers who have gone to AI labs that they can’t publish the work they want to. “The top few AI companies share some things that are at the cutting edge, but it’s a very narrow slice of the potential AI literature that could exist,” Nathan Lambert, an independent AI researcher, told us. According to one study, when AI experts permanently transition from universities to companies, they publish roughly 65 percent fewer papers a year. Papers that do get published tend to emphasize progress on safety efforts and not frontier AI capabilities; corporate research can help fuel marketing hype. And with competition among AI companies escalating, it’s not only the research that is getting locked down, but AI models too. Last month, when Anthropic released a new powerful model called Fable, the company announced that it would invisibly degrade the model’s ability to do certain AI research. The move was positioned as a safety decision, but the resulting outcry from academia was strong. (Anthropic apologized and changed the policy.)
Academics inside of the tech companies are more sanguine. Shi, the Georgia Tech professor who works at Nvidia, said the company is committed to open research, which is partly why he took the job. Meanwhile, Dragan, the Berkeley computer scientist at DeepMind, explained that publishing isn’t everything. “I find a much bigger priority than publishing is engaging in policy and regulation,” she wrote to us. We also reached out to spokespeople at OpenAI, Meta, and Anthropic for comment. Anthropic declined to comment; OpenAI and Meta did not respond.
The implications are potentially profound, not just for AI research, but for science more generally. In Silicon Valley, there’s a popular view that leading AI labs will end up as the primary drivers of science. “AGI has the potential to be the ultimate tool for advancing science and medicine,” DeepMind CEO Demis Hassabis wrote last week. Hassabis has previously said that an inspiration for his company is Bell Labs, the R&D arm of AT&T, which employed some 1,200 Ph.D.s at its height in the 1960s. Scientists at the lab earned at least 10 Nobel Prizes, and invented the transistor and the modern solar cell. Already, DeepMind has done Nobel Prize–winning research on protein-structure prediction, and the top models have already demonstrated impressive capabilities in mathematics research. On Sunday, a mathematician who works at Anthropic posted that he had used the company’s Fable model during the World Cup final to resolve a nearly 90-year-old mathematical problem.
But if scientific talent and computing power becomes concentrated inside private firms, Silicon Valley could also end up as a gatekeeper of science. Some are raising concerns: A group of academics recently published a declaration warning about “the increasing involvement of technology companies in mathematical research.” If left unchecked, they argued, the incursion of tech companies into research could affect “the scope and depth of mathematical research itself.” Instead of accelerating science, as the leaders of AI companies claim they will, they might end up suffocating it.
Physical credit cards should be a thing of the past. Over the past few years, touchless services such as Apple Pay have driven a boom in the digital-payment industry: Digital wallets now account for about half of global online purchases and one-third of point-of-sale payments. You would think that credit-card companies might abandon investing in cards that weigh down your wallet—but you’d be wrong.
Take, for example, the Platinum Card from Robinhood, introduced in March. Coated in 99.9 percent pure platinum, it weighs 27 grams, or more than half the weight of an average golf ball. “If you throw it on a dinner table, it's gonna make a ‘clang,’” Deepak Rao, a senior vice president at Robinhood, told me. The Platinum Card is just the latest addition to an ever-growing heap of lavish cards that speak to America’s appetite for flashy, metallic, and loud ways to pay.
American Express last year released a mirrored card so reflective that one card holder I spoke with told me that she uses it whenever she needs to touch up her lip liner in a fix. A similarly mirrored credit card made by the fintech company Bilt comes with its very own polishing cloth. Apple, meanwhile, provides a cleaning guide for its titanium card that recommends avoiding “abrasive objects.” Revolut, an online bank, offers a steel credit card that comes in a range of colors, including lavender, space gray, and rose gold. And JPMorgan Chase’s cards come in boxes that are themselves so intricate and high-quality that Brian Dean, a finance influencer, told me he keeps one on display on his shelf.
In 2022, the journalist Amanda Mull wrote in The Atlantic that metal payment cards were on the path to becoming luxury accessories. Today, that transformation seems complete. Opulent cards, marketed for years as status symbols, have been embraced by an entire commercial ecosystem of banks and fintechs trying to one-up each other, eager consumers, and influencers extolling cards’ virtues and ranking their weights with microscales.
Institutions are now designing cards with virality in mind: Some cards omit names and numbers so they can be filmed more easily, and others are proudly marketed by grammage. A card’s weight informs its so-called plunk factor, akin to the leather grade of a handbag or the sparkle of a diamond. Lewis Rubovitz, the chief strategy and marketing officer at CompoSecure, the leading manufacturer of metal cards in the United States, told me that most providers want their credit cards to be “heavy, heavy, and heavy.”
Such weight doesn’t come cheap: Industry experts told me that plastic cards cost issuers about $1.50 each to buy from manufacturers, whereas metal cards run companies about $15 on average. That price can surge into the hundreds or even thousands of dollars if those cards are made of metals like palladium, platinum, or gold. As cards get fancier, they typically become heavier. Whereas a plastic card weighs approximately five grams, cards made with stainless steel—the most popular material for metal cards—can weigh more than two or three times as much. A solid-gold card can weigh up to 36 grams, the equivalent of three AAA batteries.
Barclays makes a card plated in 24-karat gold, an inactive version of which is listed on eBay for $199. Robinhood-card holders can upgrade the company’s steel Gold Card to a version made entirely of 10-karat gold. Those cards can cost thousands of dollars to produce, depending on gold prices, according to Rao. American Express, meanwhile, has twice released limited-edition runs of cards made from melted-down Boeing aircraft, and Revolut coated a recent release in titanium of the grade used in an Audi F1 race car.
It’s all starting to get more than a little absurd. Leanne Fremar, JPMorgan Chase’s global chief brand officer, told me that the company has experimented both with different ways to incorporate metal into its cards, such as manufacturing versions with a metal core or veneer, and with finish techniques that make cards textured, ombré, or polished. Some efforts are years in the making. Robert Myrsater, the head of design at Bilt, told me he had dreamed of making a credit card with a reflective, chromelike finish since 2021, but he ran into a serious issue ahead of the January release of Bilt’s mirrored Palladium Card: It retained finger smudges. He tested out numerous layering treatments before finally just including the microfiber cloth that owners can use to keep the card looking fresh.
If your credit-card company doesn’t offer metal versions, third-party vendors have stepped in to do the job for them. Metal-CreditCard.com will “upgrade” plastic credit cards by extracting their chips and placing them into steel cards. Engraven Card, another company that turns plastic cards into metal, finds particular success in custom designs that make explicit references to wealth. “I’ve upgraded welfare EBT cards to say Millionaires Club,” Travis Jean, Engraven Card’s founder, told me.
The metal-card industry’s sales pitch is that a credit card’s weight carries genuine social currency. Online videos rank the various sounds metal cards make, ranging from “mid” to “like a flex,” when they’re plunked down on a table. “That matters to people: how it feels when it goes down at that restaurant bill,” Jessica Ling, the executive vice president of global advertising at American Express, told me. Metal-CreditCard.com’s founder, Vincent Torres, told me something similar: “The plunk factor of dropping a full metal card on the table, it changes people’s perspective of just who you are.” (Torres recently wrote a book titled How a Metal Credit Card Transformed My Life.)
The message appears to be working. Spencer Johnson, a YouTuber whose unboxings of cards have amassed hundreds of thousands of views, told me that videos on metal credit cards are “high-dopamine content.” Viewers “want a little taste of that just by watching somebody else open theirs, even if it’s not attainable for them,” he said. Rachel Tyers, another credit-card influencer, first made it big on Instagram when she posted a video explaining how to apply for the mirrored card from American Express. The video has 3.6 million views, and she’s since made videos of herself using scales to compare the weight of credit cards and testing which are scratch-resistant or show fingerprints.
Whether you really should be investing any of your sense of meaning and self worth into a three-inch slab of metal is, of course, a question all its own. But card sellers are tapping into a surge of interest in luxury goods. Demand for metal credit cards is on the rise alongside that for high-end jewelry, expensive homes, and premium airplane seats. Several people I spoke with speculated that the ubiquity of digital payment has only helped credit cards stake out new territory in the realm of the physically ostentatious. “It is a much more dramatic gesture to be at a table with a bunch of friends, and you’re plunking down a metal card over beeping your phone into the payment system,” Clint Henderson, a managing editor at The Points Guy, told me. Digital-payment services might be more convenient, but they can’t be held, flashed, or thrown down on a restaurant table. They certainly won’t score well on a drop test.
And so, the designs keep coming. Cards that light up, are made entirely of glass, and even contain jewels have already been manufactured for clients by CompoSecure. Revolut is hoping to soon bring its Ultra Card, which arrives encased in a transparent picture frame, to the United States. As for Robinhood, Rao has an idea for how to outdo his past products: a card in Vantablack, one of the world’s blackest coatings. “It’s going to be epic, but I haven’t figured out how to do it yet,” he told me.
There are worse things you could do than drive to a Parisian airport and mess around with a temperature sensor on the airfield—for only a few minutes! But why do that at all?
To make money, maybe. This is the scheme that some mysterious person (or persons) reportedly carried out in April to win a bet on the prediction-market platform Polymarket, which allows people to wager on the highest temperature in Paris on any given day, not unlike weather derivatives. The popular version of the story says that the shadowy figure warmed the sensor with a hair dryer—a photo showing this turned out to be AI generated—but it could have been anything. A lighter, a hand warmer.
By artificially boosting the number on the sensor, this person won about $20,000. The French police are investigating because you can’t trespass at the airport or tamper with its equipment, but Polymarket apparently didn’t take the winnings back. (It did change the airport that it relies on for Paris-temperature readings; the French government, however, blocked access to the platform this week.)
Over the past couple of years, plenty of concern has been expressed about how prediction markets may enable—and create the incentive for—insider trading. Some of the many people who have wagered on markets corresponding to when a war would start, when a regime would change, or even who would be at Taylor Swift’s wedding may actually have known the outcome before it happened. “Mention markets,” which allow participants to bet on words that a public figure might use in a speech, or how long the speech will be, can obviously be gamed by people close to the figure in question—just last week, ABC News reported that a teleprompter operator at the White House has made more than $100,000 by using Kalshi to wager on President Trump’s speeches. (“The White House has strict ethics guidelines that we expect all staffers and officials to follow,” a White House spokesperson told the outlet, while confirming that the employee had been placed on leave.)
The issues with insider trading are clear: It’s illegal, for starters, and the platforms ban it, anyway. (Although insider trading also arguably serves the purpose of prediction markets, which is to channel the “wisdom of the crowd” to forecast the event most likely to happen.) People can also easily identify and avoid the markets most susceptible to it. But what about the hair-dryer sort of trading, when an apparent outsider changes the course of events to cash in on a given outcome? “I think of it as probably a bigger issue or problem than insider trading would be,” Vincent Grégoire, a professor of finance at HEC Montréal, told me. The Parisian-airport incident didn’t hurt anybody, but future ones might. (You can imagine a bunch of ways in which someone might create a dangerous situation while, for example, attempting to delay an airplane.)
Some experts call this type of fraud “manipulation of the physical world”—a riff on market manipulation. I asked others for catchier ideas. Joshua Mitts, a professor at Columbia Law School, wasn’t sure that manipulation was the right word, because it has a specific definition for financial markets, but he still kept saying it accidentally as we spoke. I agree that the term is a little lacking. Manipulating the physical world is what I do every moment of every day as I breathe and walk. I would call this type of fraud “outsider trading.”
Outsider trading is also against platform rules. Polymarket and Kalshi both prohibit any type of market manipulation. When I reached out to Polymarket to ask about what had happened in Paris, a spokesperson relayed a statement to me from Olivia Chalos, its deputy chief legal officer, who wrote that Polymarket “maintains monitoring, review, enforcement, and reporting processes to address potential violations.”
But it’s obviously a bit of a whack-a-mole situation. The U.S. Commodity Futures Trading Commission recently proposed more federal rules for prediction-market platforms to ease the problem. One of the big suggestions was to prohibit the listing of markets that would “create perverse financial incentives”—for instance, markets on wildfires or a market on whether a famous ape named Little Joe would escape from a zoo (for a second time). Polymarket includes a note on that market explaining which law a person would be breaking if they attempted to free the gorilla themselves, and that they could receive a life sentence in prison for doing so. That may have been the first time a prediction-market platform has featured such a warning, the Bloomberg columnist Matt Levine noted in a May newsletter, “but I have not done a comprehensive survey and I doubt it’ll be the last.”
The experts I spoke with said that the most appealing markets for this kind of trickery would rely on a single number or simple, discrete outcome that a normal person could cheaply affect—the temperature reading in Paris, for example. Caleb Davies, an internet-famous prediction-market trader, highlighted another when he wrote about his experience losing thousands of dollars betting on Spotify’s streaming charts, which scammers had apparently figured out how to game by artificially juicing how many times a given song was played. Davies says the experience led him to an epiphany. He’s no longer just betting on the data, he said: “I’m also betting that nobody is going to mess with it.” And that’s a bad bet.
Davies posted a screenshot to X to show that he’d removed about $250,000 from his Kalshi account, and he says he’s stopped using it entirely. “I love prediction markets,” he told me. “I’d love to keep playing in them. But right now, both platforms are garbage, I think.” When I contacted the company for comment, the Kalshi spokesperson Jacki McGavick said that Kalshi “takes market manipulation or fraudulent activity as seriously as any federally regulated financial exchange” and is investigating the Spotify incident.
For the sake of conversation, Davies scrolled through Kalshi while we were on the phone, looking for opportunities for manipulation. Spotify charts were a good one, he knew. Rotten Tomatoes scores would be better (more money), though a little harder to game. (Nevertheless, New York magazine’s Matt Stieb figured out how to do it.) I pointed out to him that a market was open for the timing of this year’s first human case of screwworm in the United States. You could expose yourself to some infected livestock. But he wasn’t interested. “I think I’d just rather not make money than do that,” he said. “I’ve read about that—it doesn’t sound very appealing.”
Right now, the platforms—which are very much in advertising mode—carry a lot of markets that don’t have a ton of trading volume. They want as many as they can to attract new users and generate revenue. Also, a lot of them are funny: Most people probably recognize that it’s a waste of money to wager on whether Little Joe the gorilla will escape, but it’s fun to talk about. As more outsider-trading stories surface, though, people will likely be more hesitant to participate in markets that seem easy to rig. This problem, even more than insider trading, may start to limit what Polymarket and Kalshi can offer.
There are plenty of things the platforms could do to intervene in outsider trading, Joshua Della Vedova, an associate professor of finance at the University of San Diego’s business school, told me. They could limit the size of bets in markets in which the result would be easy to rig. They could look for suspicious trades using some of the same methods that detect insider trading. His preferred method looks for “excess accuracy.” If someone bets that an Olivia Rodrigo song is going to hit No. 1, that makes sense—but if they also specifically bet “no” on Drake and Charli XCX and Bad Bunny, then they’re demonstrating a suspicious amount of confidence. That confidence could come from having an inside line, or it could come from using bots to spike streams artificially. Platforms could also disallow markets that are resolved by checking a single number from a single source, he suggested. That would have solved the issue at the Paris airport, for example, and it’s not as if Polymarket couldn’t have used an aggregation of weather data, rather than just one reading.
Without some change, the platforms may continue to send ripples through reality—beyond just the goofy markets that they choose to list. A Wall Street Journal report found that Polymarket had been encouraging people to share videos with faked versions of their Polymarket dashboards, showing made-up earnings on imaginary bets that never happened. Weirder to me were the staged viral clips through which Polymarket and Kalshi looped themselves into the New York Knicks’ run to their first NBA Championship in more than 50 years.
In one of those clips, a purported Knicks fan outside Madison Square Garden yelled, apparently improvisationally, “My mayor Muslim, my bagels Jewish, my Christian Dior, Knicks in four.” Later, when it turned out that he hadn’t made this up at all—someone had posted it as a TikTok comment weeks before and parts of the rhyme were even older—and that the mic he’d yelled into had been Kalshi branded, the whole thing a Kalshi stunt, nobody seemed to care. The chant became real. It was everywhere.
The platforms pitch themselves as truth machines, which could mean a couple of different things. Are they finding it or making it? At some point, it may be difficult to identify the difference between reflecting our reality and creating a new one—or doing both.
The AI boom has showered some of the nation’s most prominent companies in market value. OpenAI and Anthropic are now the two most valuable private companies in the world. Google, Microsoft, and Nvidia have become larger than ever. But among the biggest winners has been Caterpillar, a purveyor of yellow trucks and cranes. Caterpillar’s stock has more than doubled in value over the past year, making the company worth six times as much as Nike. Its ascent has little to do with the construction vehicles (both real and toy) that it is known for. Instead, Caterpillar’s success comes from its giant gas-powered engines that are helping power the nation’s data-center build-out.
For all the attention given to improving model capabilities, the most important inputs to AI are not bits but atoms and electrons. To train and run its models, the AI industry needs to build. Data centers are incredibly complex technological and industrial operations—they require tech companies to erect power plants, build wastewater facilities, and set up all manner of electrical equipment, power lines, and advanced-cooling equipment. That calls for concrete, steel, silicon, glass, copper, and liquefied natural gas.
This data-center construction has long been frenzied, but now it is fast approaching an inflection point that is both awesome and alarming. The major players—Amazon, Google, Microsoft, Meta, and Oracle—are on track to spend more on data centers by the end of the year than they bring in from their operations. That means tech giants whose non-chatbot businesses bring in tens of billions of dollars every year will likely need to take on debt to keep paying for their AI ambitions. From the launch of ChatGPT, in late 2022, through the end of last year, these firms’ capital expenditures—most of which go toward data centers—well exceeded half a trillion dollars. The companies intend to spend a similar amount in 2026 alone. And next year, their AI investments will likely exceed $1.1 trillion, according to a recent report from J.P. Morgan.
Silicon Valley has long enjoyed a straightforward advantage: Its products and services cost relatively little to make but generate tremendous revenues. But operating a data center is in many ways more like running a steel mill than a smartphone app. In particular, data centers require stunning amounts of electricity, especially as AI models have gotten more powerful. Five years ago, a standard data center might require 10 or 50 megawatts of power, enough for tens of thousands of homes. A year ago, data centers demanding a gigawatt or two of power were considered gargantuan. This week, Meta announced that it is more than doubling the size of its flagship AI data center, which will increase its peak power needs to five gigawatts. A data center proposed in Utah, if completed, will demand nine gigawatts of power. That’s several large cities’ worth of energy pulsing through a few big warehouses, all dedicated to bots.
Electrical grids typically don’t have enough energy to supply these megaprojects, and so the fastest way to bring a data center online is to build your own power plant. That’s where Caterpillar comes in: The company has an enormous number of back orders for its power equipment, as does every major natural-gas-turbine manufacturer in the world. In May, Elon Musk reportedly spent at least $1 billion to buy an energy company with a fleet of combustion turbines, likely to power his AI model, Grok.
Power is only one part of the challenge. A data center requires wiring together dozens of cutting-edge AI chips onto a refrigerator-size rack, then linking thousands of those racks in a building. Those chips run hot, up to 200 degrees Fahrenheit, and are typically cooled by some combination of water and industrial fans; get too close to one of the racks, and you may break a sweat.
OpenAI’s self-proclaimed goal of making “intelligence cheaper, better, and more abundant over time” really amounts to powering and cooling data centers as efficiently as possible—in other words, getting really good at HVAC. And then there’s the complex global supply chain of mirrors and lasers and rare-earth minerals that go into the chips used to train and run AI models.
As the AI industry pours more money than ever before into data centers, executives have begun sounding like they are in the construction business. OpenAI CEO Sam Altman has said that the biggest constraint on his company is “electrons”—that is, power. Musk, Microsoft CEO Satya Nadella, former Google CEO Eric Schmidt, and others have said the same. Meanwhile, Nvidia CEO Jensen Huang recently said that a lack of skilled “plumbers and electricians” will be the AI boom’s “hardest” bottleneck to overcome.
Still, Silicon Valley seems to have only just begun reckoning with what it means for its primary business operations to be in the material realm. You cannot vibe-code a data center. In perhaps the clearest sign of the AI industry’s desperation to keep building, Anthropic, the most safety-conscious firm, is spending $1 billion a month to rent a data center from Musk, who just a few months ago deemed Anthropic “evil.” Anthropic CEO Dario Amodei, for his part, has accused Musk of “disturbing negligence.” Meanwhile, the tech industry is struggling to quell a mounting AI backlash, which has largely taken the form of opposition to data centers. A few days ago, New York became the first state to institute a moratorium on new “hyperscale” data-center construction.
If anything, AI is poised to become even more like drilling an oil well. All the copper, silicon, electricity, and labor that go into building and running a data center are mustered to process and spit out “tokens”—in essence, words—that are the basic unit of AI. The State Department recently announced an international pact to bolster the AI supply chain, called Pax Silica. “If the 20th century ran on oil and steel, the 21st century runs on compute and the minerals that feed it,” Jacob Helberg, the undersecretary for economic affairs, said in the announcement. Financial markets and investors are already beginning to treat tokens as the raw inputs to modern civilization: Various firms are launching such markets for compute, akin to how you can trade oil, steel, and corn on a stock exchange.
By dominating the digital world, Silicon Valley has already exerted indirect control over the physical one. Today’s data-center build-out is a whole new beast. From the most concrete aspects—cooling, power, land—to the most abstract financial instruments, AI has morphed into a heavy industry.
There are two competing schools of thought over just how water-intensive AI is. In one, the technology is horribly thirsty. Data centers will exacerbate droughts throughout the country and “drain the great lakes,” as one popular Reddit thread puts it. The former Representative Marjorie Taylor Greene has taken up the issue, sarcastically posting at one point: “How dare the peasants complain about data centers stealing their water and driving the cost of electricity!!”
The other school, populated by Silicon Valley types, holds that “The Data Center Water Crisis Isn’t Real,” as an article in the right-wing publication Pirate Wires argued in December. The issue with water “is totally fake,” “insane,” and has “no connection to reality,” OpenAI CEO Sam Altman said a few months ago.
As with so many debates around AI, the arguments are not always presented in good faith. But unlike, say, the ambiguities over how much work a chatbot can or should be trusted to do, water use seems like it should be measurable, with clear facts supporting one side or the other. That is not the case. Does AI waste a ton of water? Not necessarily. Sometimes. It really depends. “We don’t have the real ground-truth numbers on data-center water use to put everyone on the same page,” Eric Masanet, a sustainability researcher at UC Santa Barbara who studies data centers, told me. Neither data-center operators nor utilities provide much insight; what public information exists can be contextualized, recontextualized, and distorted to suggest basically anything.
Yet the water issue is tremendously consequential—not only for the well-being of residents in all of the counties where new data centers are popping up but also for the future development and regulation of AI itself. Just this week, when New York Governor Kathy Hochul signed the nation’s first statewide data-center moratorium, her office cited the “massive amounts” of water that hyper-scale data centers could require. In a drought-stricken county or during a heat wave, a data center may seriously burden local water supplies. Where water is abundant and utilities can afford new infrastructure, data centers may pose little issue.
Outside of water-stressed regions, “it’s not really a major concern for now,” Fengqi You, an energy-systems expert at Cornell, told me. “But as the sector grows, who knows?” Turning any particular case into a sweeping pronouncement produces little but heightened emotions and confusion. As ever, the truth is far more nuanced and conditional than activists or tech executives admit.
The problems start with the framing. How much water do data centers use? is not a particularly useful question. On a national scale, these buildings do not use an alarming amount of water. In 2023, data centers used slightly more than 17 billion gallons of water for cooling, as much as about half a million Americans collectively use in a year. This may seem like a lot out of context, but it’s less than a tenth of a percent of the total amount of water used by U.S. farms that year.
Things can look different at the county level. A Meta data-center campus under construction in Lebanon, Indiana, could, at its peak, demand 8 million gallons of water a day: This would be a drop in the bucket in New York City, but it’s more than double the peak demand of the town. Every data center demands a distinct evaluation, and the picture can quickly shift. Water use depends on the local climate, the local water supply, and the makeup of the regional electrical grid, as well as on the design of the actual data center. For Meta’s Indiana data center, city officials have said that planned upgrades mean there is “nothing to indicate” that they will run out of water. Meanwhile, in Newton County, Georgia, a local official has said that “we just don’t have the water” for all of the data centers trying to move in and that the county must “race” to improve local water-recycling facilities.
Rather than worrying about the quantity of water, it’s best to ask how a particular data center uses water. In general, water is always used for the same thing—to keep the computer chips inside these buildings cool—but there are many different methods for doing so. Broadly speaking, data centers can either expel the heat those chips generate with cooling towers, which evaporate water, or with air-cooled chillers or similar technologies that use electricity to vent away hot air. Think of cooling towers as human sweat, which passively cools us down as it evaporates, and chillers as car radiators, which use a chemical coolant and then blow air to physically push out the heat.
The AI industry has settled on the car-radiator approach for many of its biggest, most controversial data centers. They run water directly into the servers to pull away heat, cycle the water down through an air chiller, and then—crucially—reuse it. That’s how OpenAI’s Stargate data center in Texas, Meta’s Hyperion data center in Louisiana, Microsoft’s Fairwater data centers across the country, and many other enormous facilities operate for most of the year. Such closed-loop systems don’t lose water to evaporation, which is likely what Altman was referencing when he called concerns over water “insane”: Once these data centers have water, they don’t need more of it, in theory, because it’s never used up.
The real incentive for closed-loop systems is convenience. Building out local water infrastructure or tapping into a reservoir can take years, using groundwater is controversial (groundwater doesn’t replenish quickly, and using it can pollute the environment), and you can’t realistically bring water from Lake Michigan to your data center in dusty Arizona. Water availability can be a serious bottleneck to new data-center construction. A closed-loop system with air chillers bypasses these problems.
So why doesn’t every data center just use chillers? For starters, air chillers require substantially more electricity compared with cooling towers to expel the same amount of heat. “Can data centers use zero water? Yes, that’s very easy,” Shaolei Ren, an AI and sustainability researcher at UC Riverside, told me. “But the downside, the cost they have to pay,” is using more electricity—10 to 65 percent more, by various estimates. One nine-gigawatt facility being considered in Utah would require nearly as much power as New York City.
Data centers’ almost surreal power demands are largely being met by combustion turbines, many of which are built by the tech companies themselves (to avoid waiting on upgrades to the local grid). And generating electricity—including from these private power plants, which burn fossil fuels—needs water too. In 2024, Meta’s “indirect” water consumption—referring to the water used during power generation—was 19 billion gallons. That’s 23 times its direct water consumption, and most of it went to data centers.
Few industries consistently report indirect water use, in part because many of the estimates are flawed. Indirect use comes from averaging the water footprint of all the power sources on a regional grid—which includes hydroelectric plants that consume huge amounts of water due to evaporation from dams and lakes. In other words, indirect water-consumption figures can arguably under- or overestimate any individual plant’s contributions. Paradoxically, the AI industry’s decision to avoid using water to dissipate heat from inside its data centers might expand some facilities’ overall water footprint. “If you use water to make your cooling more efficient on-site, you will use less electricity,” Jonathan Koomey, a data-center and sustainability researcher, told me. “It’s not a simple matter of water use bad on-site.”
It might be possible to design data centers that use both less electricity and less water. In cooler climates, for instance, data centers can pump in outdoor air; this is what Amazon does at its mega data center in northern Indiana. In summer, spraying small amounts of water into the outside air can bring down the temperature enough to do the same. Data centers are also beginning to experiment with allowing the chips to run at higher temperatures, which reduces the overall cooling demands. And, of course, the rather obvious way to build data centers that put less strain on the environment is to opt for renewables—which consume almost no water—instead of natural-gas turbines.
Until these approaches are the norm, people will understandably worry about the effects these sprawling facilities will have on their communities. Assessing how any data centers will affect nearby reservoirs ultimately requires the scarcest resource of all when it comes to these facilities: accurate and precise information. Tech companies want data centers up and running quickly, so they contract with shell companies that negotiate behind closed doors and under nondisclosure agreements. Residents might find out about new construction only after shovels are in the ground.
Already, rising temperatures and prolonged droughts are out of any individual’s control; now a big industrial facility arrives, under the cover of night, that may or may not exacerbate many problems caused by the climate crisis. This sense of confusion goes a long way toward explaining why data centers’ environmental impacts have become one of the key battlegrounds in the AI backlash—the primary issue, in many cases, is not lack of water so much as lack of control.
This article originally referred to a facility under construction in Utah. In fact, that proposal is still being considered, and construction has not begun.
In an impressive challenge to the idea that no publicity is bad publicity, Anthropic is now promoting itself with images of disaster, death, and human suffering. The AI company’s latest ad, which debuted during a World Cup match last week, is a perplexing 90-second artifact. The commercial opens with footage of a house engulfed in flames and the crackling sound of burning wood. From there, a series of dark images—rows of gravestones in a cemetery, a homeless couple sleeping on the street, laborers working in a West African granite quarry—flashes across the screen. These stills are accompanied by a series of voice-overs: “Can AI be trusted?” “Who’s going to hit the brakes if we need to?” “How do we really ensure that what we’re aiming to achieve really does benefit the majority of people?” The questions are left unanswered.
Halfway through, the ad moves from dystopian film trailer to feel-good corporate ad. We see a photo of a mother sitting beside her daughter as a voice speculates that AI will allow her to be a better mom. Later, a medical worker uses a laptop covered in Anthropic stickers. “Maybe it’ll cure some great things,” a man says. A nature montage plays—there’s a humpback whale, grassy hills, and people hugging.
Finally, text flashes on the screen: “There’s hope in hard questions.” In the background, we see the ocean. And then Anthropic’s tagline: “Keep thinking.”
The commercial is tone-deaf. It’s one thing to mention the possibility of hitting the brakes on AI progress; it’s another to overlay the question on an image of dead Americans. The video is also just confusing. Does Anthropic think that AI is going to set the world on fire and leave us jobless, or that its chatbot, Claude, will lead us to a more human future where we spend our days watching whales by the sea? The company’s answer seems to be: We don’t know. The video has been mocked, obviously. Some people likened it to hypothetical ads by oil companies professing concerns about the environment. The head of one tech-media company posted, “I actually like the tagline but we don’t need the shots of *checks notes* burning houses and graves!” Meanwhile, OpenAI’s Sam Altman—who has a self-interest in undercutting Anthropic—chimed in to call the ad “satire.”
This is not the first time that a company has botched its advertising around AI: OpenAI has had its own share of embarrassingads. By now, it seems like a rite of passage. Two years ago, Apple ran a commercial depicting musical instruments and art supplies being crushed by a hydraulic press. The ad was supposed to show off just how much creativity could be compacted into an iPad, but instead it seemed to symbolize generative AI’s bludgeoning of the arts. Apple subsequently apologized. During the Olympics that summer, Google ran an ad in which a young girl wants to write fan mail to the track star Sydney McLaughlin-Levrone—so her dad fires up Gemini and asks the bot for help. The ad was highly unpopular, and Google pulled it from TV. If anything, Anthropic has had a comparatively good run until now. During this year’s Super Bowl, the company aired a commercial dunking on OpenAI for putting ads in user conversations, which even Altman said made him laugh (despite his qualms with how the ad presented OpenAI).
For all the criticism, Anthropic’s latest commercial captures the company’s ambivalence about its own existence. Here is a leading AI company worth nearly $1 trillion arguing that the very technology it is racing to commercialize may well imperil the world. “We founded Anthropic because we believe the impact of AI might be comparable to that of the industrial and scientific revolutions, but we aren’t confident it will go well,” the company wrote in 2023. In recent years, Anthropic has made a show of positioning itself as the ethical AI company. Earlier this year, it attracted the ire of the Trump administration when it pushed back against the Pentagon’s demands for unrestricted access to its AI models over concerns that the technology could be misused.
The new advertisement is part of a broader campaign Anthropic is running to address the “hard questions” raised by AI’s progress. In an email, an Anthropic spokesperson pointed me to a series of blogposts about the initiative: This spring, the company traveled across the country to better understand the public’s views about AI. In the video’s description on YouTube, Anthropic says that every question in the ad comes from a person that the company spoke with. (The spokesperson declined to comment further.)
The idea that Anthropic is worried about AI’s negative impacts and wants to hear from the public is certainly better than the alternative. The trouble, though, is that as Anthropic has matured, the company’s constant messaging about its anxiety over the future of AI has become less palatable and, occasionally, downright maddening. “Anthropic is at war with itself,” my colleague Matteo Wong wrote in January. Since then, the company’s valuation has soared. As it develops more advanced AI models and races toward a public offering, the contradictions are becoming more difficult to square. However sincere Anthropic may be—Wong noted after several visits to the company that its employees can be “almost pathologically earnest”—it’s no surprise that people are wary of the world’s most valuable private company asking the “hard questions” about AI.
The same tension is at play across the industry. OpenAI was founded as a nonprofit with a mission to “ensure that artificial general intelligence benefits all of humanity.” In the years since ChatGPT was first released, the company seemingly has prioritized financial gain instead. By now, it feels pointless for Anthropic and the rest of the industry to simply pose questions about AI ethics and safety. These questions are already consuming the public’s imagination amid a mounting AI backlash. What matters is the answers. And on that front, AI companies have less to offer. Perhaps the commercial’s tagline provides a helpful clue: For now, Anthropic will keep thinking.
Editor’s note: This work is part of AI Watchdog, The Atlantic’s ongoing investigation into the generative-AI industry.
As they scramble to keep their systems online, AI companies are making things expensive for the rest of us. Large language models such as ChatGPT and Claude are so resource-hungry that tech companies may be purchasing 70 percent of the world’s supply of high-end computer memory, causing a shortage. As a result, the prices of computer memory and storage are skyrocketing: Hard drives that I bought for my reporting two years ago for $350 each were $800 when I checked two weeks ago, and are now out of stock. The prices of some laptops have gone up as much as 50 percent, and low-cost computers are being hit the hardest. Affordable entry-level computers may “disappear by 2028” according to one forecast. And the memory shortage is expected to continue for years.
The memory is being put into data centers, which tech firms are expanding at incredible speed. They are planning to multiply total U.S.-data-center capacity by a factor of eight over the next few years. The demand for electricity at these sites is already so great that some companies are repurposing jet engines to power them.
The problem is not simply that AI is being deployed so widely or quickly. Other computer technologies have seen similarly massive growth without triggering such a large spike in electricity or a shortage of computer components: Video and music are now streamed around the globe, accounting for many terabytes of internet traffic daily; the smartphone boom required the manufacturing of billions of devices that are now transferring huge amounts of data; billions of household devices are also now part of the Internet of Things; and whole industries have moved their operations to cloud software, which is hosted not in the sky but in, yes, data centers.
The problem with generative AI, in the industry’s own jargon, is that it does not scale. The cost of growing from, say, a thousand users to a million is a key factor that venture capitalists examine when they evaluate start-ups. They want to see that the cost of adding each new user decreases over time, so that the company can support millions of users and make increasing profits. This is achieved partly through the careful engineering of computer systems that can efficiently handle more users who want to post photos, hail Ubers, or stream music.
With generative AI, the work of building efficient, scalable systems has not been done. And the problem is exacerbated by the ever-larger generative-AI models, which have grown from 175 billion parameters in 2020 to more than 1 trillion today, according to independent estimates (the actual sizes of the models powering products such as Claude and ChatGPT are secret). The large in large language model should not be a selling point. But the industry’s observation that bigger models tend to outperform smaller ones has given rise to a totemic belief in “scaling laws” that suggest any problem can be solved by simply making models bigger. “Maybe with 10 gigawatts of compute, AI can figure out how to cure cancer,” OpenAI CEO Sam Altman wrote on his blog in September.
Yet the returns are diminishing. The bigger an AI model is, the less it improves with each added parameter, and so it must be made bigger at a faster rate just to sustain steady progress. I asked a few AI researchers whether they could name any other real-world software that scales so poorly. None of them could think of any. Even outside the world of software, it’s hard to find a comparable example, given that economy of scale is the principle that has made light bulbs, cars, and clothing so affordable. By economic and engineering measures, generative AI might be the worst technology ever deployed.
But with the massive investment behind the current bloated approach, there may not be much will to change. Ilya Sutskever, a co-founder and former chief scientist at OpenAI, said in a November interview that companies take the brute-force approach “because it gives you a very low-risk way of investing your resources.” It’s harder, he argued, to invest in research that would reengineer a product currently accruing trillion-dollar valuations. Those who suspect we are in an AI-driven bubble economy have pointed out that the profitability of these companies remains an open question, largely because of the high cost and inefficiency of the technology.
Efficiency is a core principle of computer science. One of the first things undergraduates learn is that writing a program that sorts a list of 50 words is easy. But if you give that program 50 million words, it will likely run out of memory or take hours to finish. Much of computer science is learning the clever coding techniques that prevent this from happening. Many of these techniques take advantage of repeating patterns in the data so that as the program receives more input, it takes less time or memory to process each additional bit. Such efficiency is one reason that modern smartphones and computers are so capable and affordable. This is called logarithmic scaling, and it looks like this when you graph it:
Large language models do not scale logarithmically. When they’re given more words to process, they get slower and use more memory—the time and resources increase faster as the input grows. In technical terms, LLMs scale quadratically. Any computer-science student knows that this is very bad.
Epoch AI, an organization that tries to determine the costs of operating AI models, published a graph last year, which is reproduced here with permission. It shows the exponentially increasing costs of serving more “tokens”—the words users type to chatbots—with several public AI models.
AI does not have to be built this way. Traditionally, the goal of AI was to solve problems in ways that simulated human mental processes. Researchers observed their own thinking and tried to implement their mental habits in code. This approach has mostly been abandoned, partly due to the difficulty of discerning and articulating the rules of human thought, but it did have the benefit of consuming far fewer resources and data.
Today’s approach to AI doesn’t try to describe the rules of human thought; instead, it gives a computer millions of examples to imitate. The huge quantity of examples is one reason that large models can perform better than small ones when generating language, images, and music—they have more material to draw from. Some researchers want to bring back the old, more efficient approach and combine it with the modern approach, but so far these projects have not drawn nearly as much attention or funding as the models that power chatbots.
Chatbot companies are aware that their products are inefficient. Some have found techniques for improving performance, but they have not yielded significant gains. Occasionally, companies claim to have made breakthroughs—Anthropic CEO Dario Amodei has called them “compute multipliers”—but they are usually described in vague terms, and there is no evidence that the basic problems of quadratic scaling and exploding model size have been overcome. (Anthropic declined to comment on the record when I reached out to ask about this.)
Some researchers are working on extremely small models that require fewer data and less computing power. I spoke with Alexia Jolicoeur-Martineau, an AI researcher at Microsoft who independently designed one of these, and asked her about the industry’s brute-force approach. “It’s a bit insane,” she told me. “At some point you have to learn to be a bit more efficient.”
Last year, Jolicoeur-Martineau won a $50,000 prize for her paper on a “tiny recursive model” that doesn’t consume huge quantities of computing resources. “The idea that one must rely on massive foundational models trained for millions of dollars by some big corporation in order to achieve success on hard tasks is a trap,” she wrote. Her model isn’t a substitute for an LLM—it’s designed to solve logic problems in fields such as biology and electrical engineering, rather than generate language—but it can do some of the tasks that much larger AI models are currently being used for.
Yet we seem stuck with LLMs, perhaps because they have been so aggressively marketed. They are now being added to everything, whether you want them or not. In 2024 and 2025, they were integrated into both Windows and MacOS, which means that more computing power is now needed to run a basic personal computer. Smartphones are also being sold with upgraded hardware as companies anticipate new AI features. Inefficient AI is also being added to common programs such as Adobe Photoshop and Microsoft Word, meaning that computers need to be more powerful to run this software.
This is all especially bad because computers are no longer improving at the rate they used to. Since the 1950s, manufacturers have learned to make microchips steadily faster, smaller, and cheaper, a trend known colloquially as Moore’s Law. But in the past few years, components have gotten so small that manufacturers have run into molecular-level limitations on shrinking them any further, which has slowed progress significantly.
Instead of shrinking components, manufacturers have been focused on developing new hardware that is tailored to AI. This has provided occasional incremental improvements in performance, but none has come close to keeping up with the exponential curve of AI’s increasing demands.
Ultimately, inefficiency may be of little concern to the people within the tech industry who believe that they are replicating intelligence itself. There is an almost-religious conviction among many in Silicon Valley that something mindlike could arise from LLMs, which are ultimately just statistical language-generating software—this, despite the software’s inability to recall basic facts, its lack of common sense, and its complete dissimilarity to a biological brain. Even Yann LeCun, one of AI’s “godfathers,” toldTheNew York Times recently that “LLMs are not a path to superintelligence or even human-level intelligence.” But the mythological lure of AI is so strong that many engineers believe that nothing should stand in their way. Not even the basic task of writing efficient software.
If Julius Caesar had debuted this year, William Shakespeare might have been accused of writing it with AI. A certain suspicious rhetorical device appears again and again in the play. It’s in Act I, Scene ii: “The fault, dear Brutus, is not in our stars, but in ourselves.” In Act III, Scene ii: “Not that I loved Caesar less, but that I loved Rome more.” And later in that same scene: “I come to bury Caesar, not to praise him.”
These famous lines include what has become perhaps the best-known tic of AI writing—a sentence that tells you what the subject isn’t as well as what it is: It’s not X; it’s Y. Once you start noticing the construction, you see it all over the place. In one version, the Y is additive: It focuses, intensifies, or expands on the X. An annual review by Citizens Financial Group reported that growth in its private-banking division was “not just a win for the private bank—it’s a win for the entire enterprise.” In another variant, the Y supplants the X as the preferred descriptor. “The target was never a man. The target was the truth,” Michael Flynn, a former Donald Trump adviser, wrote in a March X post.
Then there are constructions like No A, no B, just C, which especially seem to crop up in AI-generated fiction. Lines such as “No bag, no things, no armor, just me” helped to fuel accusations of AI writing in the horror novel Shy Girl, which was pulled by its publisher this year. (The book’s author denied using AI to write it. Citizens Financial Group has previously said that its communications team “leverages the technology in a number of areas.” Flynn did not respond to a request for comment.)
The prevalence of this device isn’t just anecdotal—it’s measurable. (Sorry.) Barron’s reported that its appearance in corporate communications more than quadrupled from 2023 to 2025. Researchers at Pangram, which makes an AI-detection tool, estimate that Not just X but Y sentences appear three times as often in AI writing as they do in human writing. Elyas Masrour, a founding engineer at Pangram, told me that all of the major chatbots—including ChatGPT, Claude, Gemini, and various open-source models—rely on it to varying degrees.
Many other well-known chatbot tells—such as the usage of delve—have come and gone as AI companies honed their models and worked out kinks. Last fall, ChatGPT became obsessed with goblins and gremlins, prompting another intervention: OpenAI retired ChatGPT’s “nerdy” personality, whose affinity for mythical creatures had apparently infected its other models. Yet It’s not X; it’s Y has shown no signs of abating.
Before ChatGPT came along, the construction was obscure enough that it didn’t really have an agreed-upon name. Now there’s a scramble for what to call it. Terms from academia, such as antithesis and metalinguistic negation, capture some forms of the construction but not others. In an email, Laurentia Romaniuk, a product manager for model behavior at OpenAI, referred to it as “contrastive phrasing.” Despite its clunkiness, the most popular name I’ve seen is “negative parallelism.”
When deployed judiciously, negative parallelism can be punchy. But ChatGPT turns to it too often, Romaniuk acknowledged, which can feel formulaic. So the company is working on ways to broaden the chatbot’s repertoire. In the meantime, she added, users can try giving ChatGPT “custom instructions.” On a Reddit forum about AI writing, users trade tips for scrubbing negative parallelism from chatbots’ writing. One suggested pasting Claude’s output into another AI chatbot and telling it to act as a copy editor who has a strict ban on “negative pairings” such as “it wasn’t X, it was Y.”
One obstacle to a more comprehensive fix is that no one seems to know for certain why AI models are so enamored with negative parallelism in the first place—maybe not even the companies that created them. (Anthropic and Google did not respond to my requests for an interview.)
The simplest theory is that humans trained them that way. Large language models are built by first identifying patterns in unfathomable quantities of human-written text: books, academic papers, patent filings, and especially the internet. Negative parallelism was, of course, present in the initial training data. Shakespeare aside, there are lots of famous examples: In the 1960s, the legendary football coach Vince Lombardi popularized the saying that “winning isn’t everything; it’s the only thing.” In the 1990s, a frozen-pizza brand’s commercials insisted: “It’s not delivery. It’s DiGiorno.”
But the training data also included lots of bad writing that AI companies don’t want their chatbots to mimic, Tuhin Chakrabarty, a computer-science professor at Stony Brook University who studies AI writing, told me. So they also undergo “reinforcement learning,” a process by which human reviewers grade the models on their responses. Through trial and error, chatbots are guided away from inappropriate responses (making stuff up, giving illegal advice, insulting the user) and toward those rated helpful. Chakrabarty said that it’s plausible that human reviewers tended to give high marks to responses that included It’s not X; it’s Y. That could be because negative parallelism gives the impression of nuance and insight: The AI seems to be reasoning its way from a subpar descriptor to a more apt one.
That still may not be enough to explain just how prevalent the construction seems to be across the major AI models. Several experts I talked with pointed me to another, even weirder explanation.
Although chatbots have advanced dramatically in their research and reasoning capacities, they are still fundamentally text-prediction machines. They generate answers one “token”—or chunk of text—at a time, based on what has come before. Each successive word choice factors in both the statistical likelihood of that word coming next in a sequence, based on patterns in the original training data, and the likelihood that it will lead to a highly rated response overall. In other words, the models are always seeking a balance between the clever word choice and the obvious one.
When a chatbot uses negative parallelism, according to this theory, it’s essentially hedging between the two. Once it has started a sentence whose function is to characterize something, the path of least resistance is to say first what the thing isn’t (X), and only then what the thing is (Y). Put another way: For a sentence that begins with “This is,” following it with “not just” is both more likely and safer than the many options for how to directly characterize its subject. And after “This is not just,” the rest of the sentence gets easier too. The next word can be X—the boring, obvious descriptor that gets negated—which in turn sets up the final choice of Y, the somewhat punchier descriptor.
Even if researchers could figure out exactly why chatbots embrace negative parallelism, there’s another factor that could make it very hard to fix: “When something gets into these models, it’s very hard to pull it out,” Masrour, the Pangram engineer, said. That’s because one of the main ways that AI models have continued to evolve is by training on text generated by other bots. That AI text is presumably replete with negative parallelism, which further bakes it into the newer model. Now consider that a growing share of the writing on the internet is also AI-generated. This, too, becomes training data for future generations of AI.
On top of that, some AI labs are also using AI instead of, or in addition to, human reviewers in the post-training process, Chakrabarty said. Without intervention, there’s a risk of “model collapse,” in which AI reinforces its own biases to the extent that it loses touch with the human data that were meant to ground it. “It’s a very vicious loop,” Chakrabarty said. “There’s already negative parallelism in the text, and then AI is preferencing negative parallelism—it comes to a point where it just cannot write without that.” AI language is eating its own tail.
Chatbot clichés might be grating, but there’s an upside to them: They make AI writing easier to distinguish from the human variety. Masrour said that although the AI writing’s specific markers keep changing, it isn’t actually getting any more difficult for Pangram’s software to detect. The stubborn persistence of constructions such as negative parallelism may be one reason.
The trade-off, for human writers, is that a once-potent rhetorical device is now a cliché that makes you sound like a bot. That has put some people in the awkward position of insisting that they’re not using AI—that’s just how they write. Before you mock them for it, consider that you too might soon find yourself talking and writing more like a machine: A recent study by researchers in Germany suggested that AI’s writing tics are now cropping up more in spontaneous human conversation. If that continues, maybe negative parallelism will eventually lose its status as an AI-writing tell after all. The fault, dear readers, will be not in our chatbots, but in ourselves.
This summer’s lineup of sporting events has been an embarrassment of riches. This morning, the No. 1 men’s tennis player in the world (Jannik Sinner) and the winningest men’s tennis player of all time (Novak Djokovic) played in the semifinal of the biggest tennis tournament in the world (Wimbledon). In the afternoon, Spain and Belgium are kicking off their World Cup quarterfinal match. If that wasn’t enough, the evening brings a full slate of Major League Baseball games, plus NBA Summer League debuts for half the league’s rookies. All of this comes after a packed June in which, at one point, the World Cup, the NBA finals, and the NHL finals all briefly overlapped.
If it feels like there are more sports to watch than ever before, that’s because there are. This year’s World Cup is the biggest ever, the tournament having jumped from 32 teams to 48. It may come around only every four years, but the sports calendar no longer stops. The MLB, NHL, and NBA have all added games over the past few years. In 2020, the NFL tacked on two extra playoff games; the following year, the league added an extra regular-season game for the first time in nearly 50 years. Since then, it has colonized ever more calendar territory, rescheduling games from its standard Sunday slate to the Wednesday before Thanksgiving, Black Friday, Christmas Day, and a number of late-season Saturdays. Every major league has expanded over the past few years, Stephen Master, an adjunct professor of sports media at NYU and the former global head of sports at Nielsen, told me: “There’s not a league that hasn’t been touched by it.” And while this isn’t the first time these leagues have grown, the recent across-the-board expansion surge is unusual, the experts I spoke with all said.
If you’re a sports fan like me, more games to watch can seem like a good thing. You don’t get the tiny African island nation of Cape Verde holding out for a miraculous draw against the mighty Spain without a 48-team World Cup. When leagues expand their playoffs—as, say, the MLB did in 2022—they give more teams a chance to qualify and more fans something to root for. But the glut of games can also be overwhelming, even if you aren’t someone who binges three different leagues. There’s simply too much.
The feeling of excess has to do in part with changes to the sports-media ecosystem. Before streaming, most viewers had access only to local, in-market games and a handful of national broadcasts each week. Now die-hard fans can watch everything (although doing so might require three separate streaming subscriptions on top of a cable package). If you want to consume all 570 hours of top-flight Spanish soccer over the course of a season, that is an option available to you.
But a good deal of blame falls on sports leagues themselves. More games means more money: more ticket sales, more corporate sponsorship opportunities, and more valuable media-rights packages. Master suspects that last factor in particular may go a long way in explaining the ballooning schedules. In 2005, 14 of the top 100 live TV broadcasts were sports, according to Nielsen; in 2025, they made up 95 of the top 100. As the bottom has fallen out of virtually all other live TV programming, networks have come to rely on sports—and they’ll pay through the teeth to keep them. At same time, they’re now competing for media rights against deep-pocketed streaming behemoths such as Netflix, Apple, and Amazon.
The more the value of TV deals swells, Master said, the more money leagues leave on the table by not expanding. In 2024, the NFL sold the rights to televise a few games to Netflix for $75 million apiece—this fall, Netflix will broadcast the first-ever NFL game on Thanksgiving Eve.
In aggregate, the expansions contribute to a feeling of oversaturation, of too-muchness. For a fanatic, either you’re always missing something or sports take over your entire life. How are you supposed to have a regular, socially and familially acceptable Christmas Day while also keeping tabs on five prime NBA matchups and three NFL games? For a casual viewer, the proliferation of games can detract from the experience of watching any individual one.
Consider the NBA. The league introduced a pre-playoffs play-in tournament in 2021 and another in-season tournament in 2023. Yet for many years now, fans have complained that the 82-game season is too long, as have such leading lights as LeBron James and the Golden State Warriors coach Steve Kerr. To preserve their bodies, players end up taking semi-regular off nights (“load management,” to use the term of art), making it next-to-impossible for fans to predict whether the player they’re buying a ticket to see will even suit up that night. The length of the season and inclusiveness of the postseason (two-thirds of teams get in) ensure that most games have little meaning.
The nature and extent of the problem vary league to league. But the drive to grow is universal, and so is the resistance. The move to expand the World Cup generated serious backlash, in part because the group stage of the tournament has become lower stakes and hopelessly arcane. And seemingly no one favors the College Football Playoff’s planned expansion from 12 to 24 teams, because it will render regular-season games less consequential and water down the quality of the playoff itself. A number of MLB stars, too, have called for a shorter season, because 162 games is just flat-out a lot. Unlike basketball or baseball, pro football is far from the point where regular-season contests become meaningless, but that hasn’t stopped its fans from debating whether the NFL is sapping some of the magic that comes from football’s scarcity.
Whatever fans think of these expansions, chances are they’ll tune in. Sure, they’ll complain about how the extra games are diluting the product, but “everyone says that,” Master said. “And then the ratings come out and they’re incredible.” This is not a matter of fickleness; it’s a fundamental misfiring of the free market. Generally, if a business offers a product and it sells, that means consumers want it. But when sports leagues add games, the experience for fans degrades in a way that doesn’t always show up in the dollars and cents. Just because a fan watches the 82nd game of the NBA regular season doesn’t mean they wouldn’t prefer, say, a 70-game season in which every game was a little more competitive.
There’s no going back. FIFA is considering further expanding the World Cup to 64 teams, and the generally strong showings from first-time participants at this year’s tournament will only bolster that push. Despite having tripled just a couple of years ago, plans to double the size of the College Football Playoff already seem to be in motion. And NFL owners are pushing hard for another week of games. “In the NFL, there’s a sense of inevitability that it’ll go to 18 games,” Scott Rosner, a sports-management professor at Columbia University, told me. “There’s almost too much money at stake for it not to.” Owners aren’t going to walk away from an extra $1 billion in revenue, and many players aren’t going to walk away from the extra salary. The experts I spoke with couldn’t think of a single example of that happening.
Sports have always been a business, but it can feel these days as though we’re moving into a hard-knuckled, wring-out-every-penny era. The ever-multiplying micro-sponsorships, the split-screen commercial breaks interrupting play, the carving up of media rights, the expanded schedules—it all chips away, little by little, at the viewing experience. Add all of that to the way betting is reshaping the landscape, and in the process something intangible about sports fandom is being lost. Ultimately, every business has one destiny: to grow.
In late March, I started receiving daily texts from the federal government about AI. “🇺🇸AI is changing how we work and live,” one message read. “You might feel curious, skeptical, or unsure—that’s normal.” I had enrolled in an AI-literacy course from the Labor Department created to help workers succeed in the ChatGPT economy. The weeklong program, created in partnership with an AI start-up and delivered by text message, was supposed to equip Americans with “foundational AI skills,” according to an agency press release. But the government’s texts were not reassuring. One message encouraged me to ask a bot for “side hustle ideas.” Another suggested that I brush up on my AI skills by doodling: “Grab a friend and see whose drawing of a hippo AI can recognize … and whose it mistakes for a lumpy potato. 🥔.”
When it comes to AI’s threat to jobs, America has acted like a deer in headlights. For all the catastrophic messaging about looming layoffs, efforts to prepare for a major transition have lagged. After speaking with a roster of the nation’s top business and political leaders, my colleague Josh Tyrangiel concluded earlier this year that no one seemed to have a plan for how to proceed. Government efforts to shore up the labor market have been largely underwhelming. AI-literacy courses like the one I took are hardly a sufficient response if you believe that mass job loss is forthcoming; the Trump administration announced a goal last year to expand apprenticeship programs, but progress on that front has been middling. Meanwhile, the tech industry has barreled ahead in its efforts to develop more capable models. Earlier this year, Anthropic’s Dario Amodei warned that “AI isn’t a substitute for specific human jobs but rather a general labor substitute for humans.”
But over the past several weeks, politicians and tech executives have started to make a show of more seriously preparing the country for a future of AI layoffs. As promising as these new efforts are, many of them also end up serving the interests of Silicon Valley.
Consider a flashy proposal for state ownership of AI companies. Last month, Senator Bernie Sanders proposed the AI Sovereign Wealth Fund Act, which would give the federal government a 50 percent stake in major AI companies including Anthropic and OpenAI. The act would also give an appointed commission voting shares to help “stop bad decisions that will reap massive job loss,” Sanders said on a press call. The legislation, which calls for the tech industry to hand over trillions of dollars worth of stock, is unsurprising coming from Sanders, who regularly takes aim at “Big Tech oligarchs.” What’s surprising is the fact that at least one of the alleged Big Tech oligarchs is on board with a watered-down version of Sanders’s proposal.
After Sanders first announced his plan, Sam Altman requested to meet with the senator. The OpenAI CEO reportedly told Sanders that although he does not endorse the exact plan, he supports the idea of public ownership of AI companies. Altman has been flirting with such ideas since even before ChatGPT was launched, and in April, OpenAI endorsed the creation of a “public wealth fund” that would give every citizen “a stake in AI-driven economic growth.” More recently, according to the Financial Times, Altman has had recent conversations with the Trump administration about public ownership of OpenAI and its competitors; the company is reportedly floating giving up a 5 percent stake.
Still, it’s difficult to imagine a future in which our dysfunctional Congress passes a law to take a stake in AI companies, and then the government seamlessly redistributes the earnings to the rest of us. And regardless, such arrangements are unlikely to save a flailing labor market. Sanders’s aggressive wealth fund, for example, aims to start by paying roughly $1,000 to each American annually—which is not trivial, but is also far from a solution to mass unemployment. And private companies don’t give away equity out of altruism. If the government were to seize control of AI companies, the country could end up responsible for bailouts (although Sanders’s bill prohibits use of the fund for such purposes), and public ownership could dilute the demand for more onerous regulation. Perhaps that’s why Gwynne Shotwell, the president of SpaceX, home to Elon Musk’s AI efforts, announced this week that she and her husband would give stock to Trump Accounts, the savings accounts for children that just launched. Anthropic has also said that it would like to invest in researching “AI sovereign wealth funds” as a possible response to extreme unemployment.
AI companies are eager to be seen as supporting other initiatives too. In May, OpenAI’s nonprofit foundation, which owns roughly a quarter of the AI company, announced that it would spend $250 million on grants and partnerships to support the forthcoming economic transition, while also building its own team to seed new ideas. And last month, Anthropic committed $350 million toward similar efforts, much of which will fund “major research trials and program evaluation on promising public policies.” Specifics are still being worked out, but both organizations have noted an interest in, for example, investing in efforts to better measure how AI is affecting labor markets, as well as unemployment-insurance reform. (Anthropic declined to comment; a representative for the OpenAI Foundation pointed me to the company’s April policy paper.)
Economists, for their part, remain unsure about what types of policy interventions could best mitigate future problems, or really how big those problems are likely to be. AI has not yet led to widespread job losses, and many economists aren’t convinced that a jobs apocalypse is inevitable. But even absent the extreme, if moderate labor displacement occurs quickly, America could be in for a real shock. In that light, one new nonprofit, Raise Us, launched last month with the intention of testing out different policies to see what works. Led by Gina Raimondo, Joe Biden’s commerce secretary, and Eric Holcomb, a former Republican governor of Indiana, the organization has already raised more than $500 million from private companies and philanthropies (including the Rockefeller Foundation; Melinda French Gates’s group, Pivotal; and Emerson Collective, which is the majority owner of The Atlantic). Some of that funding comes straight from Silicon Valley: Anthropic, Amazon, Microsoft, and the OpenAI Foundation have contributed cash to the effort and are among the organization’s founding partners.
Raimondo, who worked extensively on AI policy during the Biden administration, told me that she hopes that by doing smaller-scale experiments now at the state level, the organization will gather data on which interventions are most effective, helping inform federal policy efforts down the line. (A set of both red and blue states—Utah, Connecticut, Arkansas, and Maryland—have signed on as initial government partners.) For example, the group is interested in testing wage insurance. Under such a policy, if someone loses a job and accepts work at lower pay, wage insurance would aim to temporarily fill some of the gap. There is “some positive evidence” in favor of the policy, David Autor, an MIT economist who serves as an unpaid adviser to Raise Us, told me, but it has “never been done at any significant scale in the United States.” The group also plans to invest in a slew of other interventions, including apprenticeship programs and career-navigation tools. Overall, the group’s efforts will likely be useful in generating real-world policy evidence that is “undersupplied” and “extremely valuable,” Susan Athey, a Stanford economist who is not affiliated with Raise Us, told me.
At the same time, the tech industry’s association with Raise Us is also a play to boost its flagging reputation. Brad Smith, the vice chair and president of Microsoft, told me that the Raise Us coalition is similar to the public-private partnerships that formed in response to the pandemic. It’s a compelling comparison, but a flawed one: Imagine if the vaccine companies involved with Operation Warp Speed were also engineering more sophisticated viruses. Silicon Valley has wedged itself into a strange position. AI companies are racing to automate work while simultaneously preparing to parachute in as saviors.
Of course, shepherding a smooth labor transition is in Silicon Valley’s self-interest. An AI backlash is brewing, and tech companies’ reputations will only further degrade if layoffs mount. The industry also has reason to worry about the federal government taking far more dramatic action against it. But all this hullabaloo around wealth funds and public-private partnerships is about much more than just optics. AI companies are leveraging the current anxiety over layoffs to induce greater dependence on their bots. Several industry-supported efforts to ward off a jobs crisis involve training people to better use AI. In Maryland, for example, Raise Us is supporting the launch of a start-up-accelerator program, which will provide displaced workers with financial assistance—and access to AI tools to help them start businesses. And almost half of Anthropic’s $350 million commitment will go toward creating a national fellowship called Claude Corps. Through the program, Anthropic will pay early-career workers $85,000 each to spend a year working to integrate Claude across at least 400 nonprofits. The solution to Claude’s threatening entry-level workers’ jobs: Hire young people to further evangelize Claude.
When the iPod Nano was first released, in 2005, it cost $199 and was sold on the promise of limitlessness. This year, Celeste Stange bought her magenta 8-gigabyte model on eBay for $69 and the exact opposite reason. Every time she’d pick a song on Spotify to stream, she’d think about the millions of other songs she could listen to instead and get paralyzed by musical FOMO, she told me. “Now I only have what’s on here.”
Stange, who is 29, is part of a recent “analog” movement in which people—usually those in Gen Z—opt for less distracting alternatives to their “everything” device. Those alternatives are not always, in fact, analog, but are in many cases older digital devices: an iPod, a digital Canon PowerShot, a DVD player. Tiffany Ng, the 25-year-old author of the newsletter Cyber Celibate, now has a first-generation iPhone, an 11-year-old iPod, two CD players, a Walkman radio, and a 1986 Macintosh Plus that takes 45 seconds to load the “Welcome to Macintosh” screen.
Ng lives in Williamsburg, Brooklyn, where 20 years ago, the word analog would’ve conjured hipsters getting into vinyl. The argument then was that a record player produced real, textured music in a way an iPod’s digital files could not. But today, people are “going analog,” as they call it, to escape devices that offer endless options, in favor of those that offer a modicum of constraint.
The iPod’s original tagline—“1,000 songs in your pocket”—suggested that you could dance like the carefree silhouettes in the ads because you didn’t have to lug around CDs. This was the era when technology was “convenient, but not connected,” Tony Fadell, a former senior vice president at Apple who is known as the “father of the iPod,” told me; then the iPod became the iPhone, which soon had an app store, and convenience and connectivity became inextricably linked. People had their library of songs, but now they also had app notifications and Angry Birds and Instagram posts from that girl they went to middle school with. Eventually, streamable Apple Music replaced iTunes, and tens of thousands of songs became 100 million songs. “I can’t go through all of it,” Stange said.
Stange could still carve out her own corner of Spotify through playlists, but on her Nano, she truly has only music she chooses, with artist photos she uploads manually. For many people, that little bit of added friction is part of the appeal. Jess Fisher, a 29-year-old actor in Los Angeles, now buys new CDs each month, which she puts on her Innioasis Y1, an iPod-esque MP3 player that the company started making in 2023, after users complained that their touch-screen MP3 devices were too distracting. “I am more excited about music because I am restricting myself from it,” she told me. When we spoke, she’d just gotten a new CD. “I’m going to sit down and listen to it and relish it, because I’ve been waiting a week to get it.” She also looks forward to the five free weekly MP3 downloads she gets with her library card, through a music site called Freegal. She has an alert set to repeat every Monday: “It’s Freegal Time!”
I spoke with nearly a dozen people trying out similar technology. Isaac Mosna, who creates YouTube videos about tech under the handle Canoopsy, started taking photographs on a first-generation iPhone because he was nostalgic for a time when “everything was a little more physical and real.” (The older camera also makes the pictures look warmer, he said.) Others told me about their Nintendo DSes and emulators they’d filled with vintage games, so they could play without distractions. These choices have become popular enough that they’re juicing secondhand markets for older tech. On eBay, people searched iPod 1,300 times an hour on average last year, and prices of certain models are up by more than 50 percent, according to data provided by the site. Back Market, a refurbished-technology company, started carrying iPods, Game Boys, and other retro tech after a limited run of Nokia 3310 phones (which can basically be used only to call, text, and play Snake) sold out much faster than expected, Thibaud Hug de Larauze, the company’s CEO, told me.
For at least some people buying into this trend, the logic goes something like: I get a DVD player, therefore I learn how to exist comfortably in the present moment. Ng has been replacing a different modern piece of technology with its older counterpart every month, an experiment she began after her iPhone broke. For the five or so hours that the Genius Bar was repairing it, she “felt so, so stupid.” She didn’t have a way to tell time. She didn’t know how to get around Manhattan without Google Maps. She decided then that she wanted to know she could live without modern technology. When she posted a newsletter titled “I chained my phone to a wall for a week,” people DMed her to ask where she’d bought the chain. (It was an old belt.)
Going “analog” is also one of the few ways to own an actual copy of a song, a movie, or a game these days. And buying those products means a person has to commit to their own taste more than they would by just pressing “Play” on “Fun Upbeat Mix” curated by Spotify. It has given Stange the ability to say, “I’m not going to pretend that Geese is the second coming of God” just because everyone else is listening to them. She’d gotten so used to streaming, though, that she wasn’t sure what to put on her iPod at first. She looked up a Rolling Stone “Greatest Albums” list and downloaded some of the recommendations. Fisher did the same.
A CD or DVD is more work to buy and can physically break down in a way a Spotify playlist cannot, but the people I spoke with generally framed those inconveniences as part of the experience. “If I want to listen to an album that I really love, it involves me spending maybe $20 on a CD, it involves me putting that CD on, it involves me using the physical remote,” Mosna said—it’s “an experience.” (Cleaning and maintaining his CD player is a bit of a pain, though.) Dan Cohen, a 24-year-old fashion designer, told me his professors couldn’t imagine why anyone would want the cross-body CD holder he’d made for his undergraduate thesis: They’d lived through the annoyance of CDs getting scratched, iPods crashing and deleting music, and DVDs getting stuck on their menu pages for eternity. And some of these devices were outmoded for a reason. Ng, the newsletter writer, got the 1986 Macintosh in part to play around with PageMaker and its word processor, but she’s still trying to figure out exactly how to use the computer.
After all, modern “analog” enthusiasts still have backups. Everyone I talked with had kept their smartphone. When Stange and I met, she took out her magenta Nano and her laptop, to show me how she downloaded her music onto the iPod. The Nano promptly died. She plugged it into the laptop to charge, but still, no dice. “I’m pretty sure it broke on the way here,” she said, and told me she’d need to figure out how to fix it. She sounded a little excited about the challenge. In the meantime, she still had her Spotify subscription.
When the 82-year-old psychologist Peter Gray describes the way he grew up, he punctuates the anecdotes by saying that modern parents would be arrested for letting a child have such fun. When he was 4 years old, he would walk to a store in Minneapolis to buy cigarettes for his grandmother. When he was 11, he would sometimes stay home from school in Hill City, Minnesota, to operate a newspaper printing press owned by his mother and stepfather.
His parents were not arrested, and that’s because the childhood they permitted him to have was basically normal at the time, even if his family did have a newspaper printing press in the house. As a boy, Peter was obsessed with fishing and baseball; neighborhood friends taught him how to ride his bike and catch grasshoppers. Although Gray’s career as a scientist would begin with laboratory studies of rat hormones, he eventually found his way to writing about his childhood, in a fashion. Over the course of his 30 years in the psychology department at Boston College, he mixed principles of biology and anthropology to put together an evolutionary theory of play.
Gray’s academic work defines play as a self-directed activity done only for its own sake. This, he came to believe, enables kids to figure out how to solve their own problems, nurture their own relationships, make their own rules, and manage their own disappointments. But he says that our society has spent the past 70 years or so interfering with that process. We’ve made it harder and harder for kids to do anything: They’re kept indoors for greater portions of the day and given less unstructured time; they play organized sports supervised by adults; they don’t go anywhere alone. Gray grew certain that this loss of independence has been harmful to their mental health.
Gray’s theory, which he laid out in a 2013 book called Free to Learn, quickly found a welcome audience. The book was celebrated by advocates of free-range parenting and won endorsement from academic luminaries such as Steven Pinker. When Gray’s fellow psychologist Jonathan Haidt and his co-author Greg Lukianoff published their 2018 best seller on the threat of safetyism, The Coddling of the American Mind, they used the title of Gray’s popular TEDx talk “The Decline of Play” as a chapter header. Haidt, who is an Atlantic contributor, told me that Gray was “the star academic” in the section of his book that deals with play. “I wish every school in America could hear a talk by Peter Gray,” he said.
Only recently has Gray expanded his idea in a way that is not quite so crowd-pleasing. Children’s need for unstructured play and exploration (guided by some safety rules and common sense) applies not just to vacant lots and city parks and backyards in the suburbs, he says, but to other settings too. It now extends to the wild spaces of the internet. “To grow up well, children have to be able to play in the world that they’re growing up in,” he told me when we spoke at his home in late winter. Kids should be free to play without their parents’ supervision, Gray insists, even when they go online.
Gray has always been a playful academic. As a graduate student at Rockefeller University during the late 1960s, he once filled every mailbox at the school with a note proclaiming that neckties were no longer required in the dining hall because they stifled thought by cutting off circulation to the brain. But the turning point in his career came a decade later, when his son, Scott, took up the mantle of rebelling at his school. Gray and his now-late wife moved him to a nontraditional school in Framingham, Massachusetts, where children received no formal coursework and directed their own education.
Scott thrived in his new environment, and Gray, who saw how much happier and more engaged his son became, pivoted from doing lab experiments on rats to making more philosophical explorations of play and learning. He also studied his son’s new school, publishing survey data on the careers and lives of its alumni, as well as detailed observations of how its students played.
Eventually, Gray would have enough of structured academia himself. By 2002, he’d made sufficient money off of a well-regarded psychology textbook that he was able to resign from Boston College and live comfortably in the small town of Millis, Massachusetts. The back of the wood-paneled house that he shares with his second wife is made entirely of glass, providing a broad view of the Charles River. By his account, his retirement has been as idyllic as his childhood. Gray sometimes kayaks against the current, up the river, for exercise.
When I visited him on a frigid day in March, he was wearing the uniform of a practical person—simple shoes, navy trousers, light layers topped with a grandpa cardigan. David Sloan Wilson, an evolutionary biologist and friend of Gray’s who is a professor emeritus at Binghamton University, described him to me as an old man with a “very boyish look.” I would say he is more like a child’s drawing of a nice old man: pure-white hair, a reedy frame, smile lines.
Wilson also described his friend as a pied piper, apparently intending this as a compliment. And from his riverside retreat, Gray has been involved in leading a movement. In 2017, he joined up with Haidt and two others to found the nonprofit Let Grow. The organization, which raised about $2 million in contributions in 2024, encourages parents and teachers to stop watching kids so intently, and tries to fight “neglect” laws that frame a lack of child supervision as criminal or reckless behavior. Let Grow has also developed a program called Play Club, through which schools can offer age-mixed free-play time when the kids (mostly) supervise themselves.
Gray and Haidt both served on the group’s board of directors. They had a collegial relationship, even though there were some minor differences in approach and worldview. The first of those was on the subject of kids and video games: Gray thought that video games provided kids with fruitful ways to play without adult control and was adamant about their value; Haidt respected that position but wasn’t so sure. The two psychologists also disagreed, from time to time, on whether kids should be using social media. But these seemed, at first, like secondary issues.
Then, in 2023, Haidt sent Gray a prepublication copy of his next book. The Anxious Generation: How the Great Rewiring of Childhood Is Causing an Epidemic of Mental Illness opens with an extended metaphor comparing smartphone and tablet use to a spine-melting trip to Mars, and goes on to make the case that personal technology has caused a full-scale youth-mental-health crisis. Its argument seemed to fit the moment: Children were reporting that they didn’t like how much time they spent on social media and didn’t feel like they were in control of their habits; some were truly suffering. The Anxious Generation gave them—and their parents—language to describe what was happening.
Haidt’s book went on to spend more than two years on the New York Times best-seller list. It was translated into dozens of languages and excerpted in this magazine. And the policies that it suggests, to remove smartphones from schools and bar kids younger than 16 from social media, quickly came to seem like obvious solutions. (In the past few years, they’ve been taken up by legislators in the United States and abroad.)
But when Gray first read the manuscript in his lofted study overlooking the river, he was appalled. “The book frankly makes me mad,” he told me. “I have to say that. I think it’s unethical.” Although The Anxious Generation goes on at length about the benefits of free play, and even makes specific mention of Let Grow, its message overall was for Gray inimical to the nonprofit’s mission. As he sees it, the analysis improbably suggests that taking away phones would prompt kids to run around exploring as their grandparents had done. The few freedoms that kids had left were private communications with their friends and some independent movement, permitted only so long as they could be tracked and called. More than that, Haidt’s book implies that kids cannot be trusted to delve into the online world on their own, or even taught to do it safely. The internet is too dangerous for children—too full of scary strangers and powerful temptations—much in the same way that parents and pundits had decried the perils of the physical world generations before. Haidt had wanted feedback on the manuscript. Gray told him he disagreed with its premise.
Months later, after the book came out, Gray stepped down from the Let Grow board so that the other members wouldn’t feel caught in the middle of a conflict. Then he posted a critique to his Substack. Gray wrote that he took “no pleasure” in rebutting his colleague’s work. “I have tried to avoid it but no longer can,” he wrote. “As a society we have almost a knee-jerk reaction to believe that the solution to any problem experienced by kids is to deprive them of yet one more freedom, and this book is helping to jerk some of those knees even further.”
The two men haven’t spoken a word to each other since their falling-out. “Jonathan Haidt is a likable person,” Gray told me. “He’s polite. He’s generous.” Gray said he is a fan of Haidt’s previous work: “His strong point has always been kind of large, somewhat philosophical arguments based on general observations and a certain amount of evidence, and he’s very good at that.” But Gray viewed Haidt’s newest large, somewhat philosophical argument as not only incorrect but immoral.
The first time we spoke, Gray referred to social-media age minimums as a violation of human rights. When we met in person, I asked him whether that was truly his position. He repeated that it was. Before The Anxious Generation was published, Gray had been working on another book about the broad topic of play. But he scrapped that one after posting his rebuttal, and instead began to write a full-length counterpoint to Haidt’s ideas for an imprint at Penguin Random House. That book, titled Restoring Childhood: How to Set Kids Free in the Age of Anxiety, is due to publish in September.
The new book will argue that the mental-health crisis affecting children is real, but has nothing whatsoever to do with what Gray describes as the “moral panic” over smartphones and social media. The real problem, he says, has to do with schools—and in particular with the 2010 rollout of the Common Core standards, which narrowed teachers’ options for creative curricula and increased the amount of time that the average American student spent taking tests.
To make that case, Gray points to the American Psychological Association’s annual Stress in America study, which surveyed kids in 2009, just before Common Core was introduced, and in 2013, just after. In 2009, 43 percent of U.S. teenagers said that performing well in school was a source of stress in their lives. In 2013, this number jumped to 83 percent. Gray found analogous survey data from before and after similar educational reforms in Sweden and England were put in place.
School policies were not the only things that changed from 2009 to 2013; smartphone use, for example, skyrocketed across the same few years. But lots of evidence affirms the notion that in America, at least, children are more stressed out by school than by any other aspect of their lives. Young people often say that they hate school, and youth suicides are far more common during the school year. In 2024, 68 percent of U.S. teenagers surveyed by Pew Research Center said that they felt a great deal or fair amount of pressure to get good grades—significantly more than those who said they felt pressure to look good or fit in. Studies of teens in North America and parts of Europe also suggest that school pressure has increased more for girls than it has for boys over the past two decades, which is consistent with the fact that, in some respects, the teen-mental-health crisis has been more intense for girls.
Restoring Childhood goes on to make another, even more surprising argument: It says that computers and video games have actually been responsible for improving children’s mental health. Teen suicides rose in every decade from the 1950s through the 1980s, Gray writes, as restrictions were increased on children’s freedoms. Anxiety and depression seem to have been rising too. But that trend was temporarily reversed by the arrival of the digital world. Suddenly, kids had a new place where they could connect with one another, make their own rules, and solve their own problems. They were among the early adopters of the new technologies, and so became authorities in their households, giving them a chance to feel competent and helpful. Teen-suicide rates never went back down to the levels of the 1950s, but they did decline by about 40 percent from 1990 to 2010. “Everybody was ignoring that,” Gray said. “Nobody was writing about that.”
Just two years ago, Penguin Random House put out The Anxious Generation. Now the publisher seems confident that people are ready for a counternarrative: Gray says that he received a $500,000 advance for his newest book. Wilson, the Binghamton professor and Gray’s friend, told me that he read the book in manuscript form and found himself convinced. “I want it to have the exact same exposure and impact as Jonathan’s book,” he said. “I would like it to be as splashy as it could be and to set up conversations at every level as to which of these interpretations is correct.”
You will not be surprised to hear that Haidt has a very different reading of the evidence. When I asked him what he made of the claim that computers were actually a boon to children’s mental health from 1990 to 2010, he instead proposed that any improvement during that period might have resulted from the phasing out and banning of leaded gasoline. (Lead exposure has been tied to developmental disorders and mental-health problems.) As for Gray’s critique of his smartphone-and-social-media hypothesis, Haidt said that it was overly reliant on the dissenting opinion of what he characterized as a minority of researchers. In particular, he mentioned Candice Odgers, a psychology professor at UC Irvine, and Christopher Ferguson, a psychology professor at Stetson University.
Gray does feature Odgers and Ferguson in Restoring Childhood, but these are not the only scholars who have taken issue with the science as presented in The Anxious Generation. I spoke with more than a dozen people who study technology and child development, and many expressed concern that Haidt overstates the strength of correlational findings and suggests causation where it hasn’t been proved. (A 2024 report on teenagers’ use of social media from the National Academies of Sciences, Engineering, and Medicine concludes that “the scientific literature on the health effects of social media use is mixed and inconclusive.”) Several of them also quibbled—as did Gray—with Haidt’s emphasis on controlled experiments that found people’s mental health improved after breaks from social media. They said these are flawed because participants consciously or subconsciously know the results that are expected of them. (Haidt, when I put this to him, replied: “If all you can say is, Well, maybe they could guess the hypothesis, then you basically are saying all psychological research is useless.”)
But if the quality of Haidt’s arguments and evidence is up for debate, the same is true of Gray’s. The dangers facing kids on social media, for instance—in an ecosystem that was specifically designed to capture their attention for profit—aren’t quite analogous to those they’d find offline. Or take the crucial data from the Stress in America surveys, which are presented in Restoring Childhood as a meaningful signal of the harmful effects of Common Core reforms. Reported levels of school stress did double from 2009 to 2013, according to that work, but the numbers roughly doubled for other common stressors, too, including family finances—so any school-related effect on stress does not appear to be unique. When I asked Gray about this issue, he said it had been “nagging” at him, and that he was going to add a footnote to the book about it before publication. I pointed out to him a few weeks later that the surveyors had also changed their methodology between 2009 and 2013. The phrasing of the questions had shifted; the first survey had asked respondents to select their top two stressors, but the second one asked them to assign scores to a list of possible stressors. Gray was caught off guard by this, and he went on to ask his publisher to adjust or remove any reference to the Stress in America surveys. That request arrived too late to change the print edition, but Gray said he plans to make the edit to the ebook and audiobook.
Restoring Childhood does offer other lines of evidence, though. “I think there’s a reasonable hypothesis here,” Ferguson told me when I asked about Gray’s book, which he had read in advance, and blurbed. He said he thinks that focusing on schools as the source of children’s mental-health decline makes sense, but added that he would like to see more evidence that the school environment had gotten that much worse. “My memory of schools in the ’70s and ’80s was: They also sucked,” he said.
Odgers, who also provided an endorsement for Restoring Childhood, told me that she appreciates the fact that the book is “broadening the conversation” about children’s mental health to include “a major stressor as reported by young people, that we actually see in the data.” But she expressed frustration at the logic used by Gray and Haidt alike. She noted their shared willingness to apply one explanation (video games, unleaded gas) to the mental-health improvements of the 1990s and 2000s, and then a completely different one (Common Core, social media) to a later section of the same trend line. No serious epidemiologist would reason in this way, she said. Both Gray and Haidt tended to downplay other obvious factors, such as the severe adult-mental-health crisis that unfolded during the same years. “Caregiver mental health is by far the strongest predictor of childhood mental health,” Odgers said.
Pete Etchells, a professor of psychology and science communication at Bath Spa University, also pushed back on the idea that Common Core is the primary cause of mental-health problems among American kids. “This is a common trap that we fall into when trying to figure out what’s going on with those declines,” he told me. “We try to answer the question of ‘What one big thing can explain this?’ The answer, in my view, is that there isn’t one big thing.” This was generally the thesis of Etchells’s own book Unlocked: The Real Science of Screen Time (And How to Spend It Better), which came out the week before The Anxious Generation did, sold only a few thousand copies, and received hardly any attention at all.
Come September, the fight over that “one big thing” will be renewed. Which change that happened 15 years ago was the real source of so much misery for children? “You can’t run experiments on history,” Haidt said, so we’ll never be able to prove that smartphones and social media caused the steep decline in youth mental health. “We just have to say which hypothesis is more plausible,” he said—and he’s yet to hear one more plausible than his own.
When I met with Gray, he said that he has total confidence in his idea: “The evidence is really very overwhelming.” Later on, I tried to press the issue: Couldn’t there be some other factors in the mix? What if Common Core had been just one of many causes of the problem? “It would be nice” if there really weren’t any one big thing, he said, but he simply didn’t feel that this was the case. “So far, I haven’t heard of any other possibility that has the same plausibility.” He’d been studying the numbers and considering the alternatives, and he didn’t see how his theory could be wrong.
Two hundred fifty years and two days into the American experiment, a 55-year-old bespectacled bald man from Liverpool enters a sterile hotel conference room in Atlanta, shaking his head. “It’s all gone to hell, hasn’t it?” he mutters.
In 14 hours, the United States men’s soccer team is scheduled to play Belgium in the World Cup Round of 16, a match that ought to be a celebration of the U.S.’s triumphant, and somewhat unexpected, run in the world’s biggest sporting event. But Folarin Balogun, Team USA’s star striker, was given a red card during the previous match with Bosnia-Herzegovina—he stepped on another player’s ankle—that made him ineligible to play. This led President Trump to petition FIFA to review the penalty, which, of course, triggered allegations of collusion and corruption: International lawyers were summoned to draft appeals, and a joyous sporting event started to look more like a legal drama.
For Roger Bennett, the scandal threatened to overshadow every value he has been working 16 years to promote. “This is not why we watch,” he told me just after 6 a.m., settling into a chair for what would be almost two straight hours of TV appearances. “It’s antithetical to the beauty we are trying to protect.”
If this kind of language about sports turns you off, you will likely not enjoy Bennett or the many hundreds of pieces of World Cup content that his company, the Men in Blazers Media Network, has put out over the past month. In the broadest terms, Bennett is a soccer podcaster and new-media executive. A better description of him might be: America’s chief soccer evangelist. Before we knew that the U.S. would get crushed by Belgium 4–1, before the team’s chances for World Cup glory died on the field, Bennett declared that Monday, July 6, would be the most important day in the country’s history. In theory, he was joking—but honestly, I wasn’t sure.
Andrew J. Clark / ISI Photos / Getty
Roger Bennett during a live show before the FIFA World Cup 2026
Ever since Bennett started his podcast, on a whim in 2010, he has been monomaniacally devoted to getting Americans not just to enjoy soccer but to become emotionally invested in it. For Bennett, the sport is a way to access the big, complex feelings and topics that typical sports fans don’t always talk about: community, love, agony, joy, Emily Dickinson’s poetry. A box of tissues emblazoned with the phrase Cry Proudly is prominently displayed in his podcast studio.
I first met Bennett in 2014, when his stint covering the World Cup for ESPN made him a niche celebrity among the set of weirdos in America (me) obsessing over soccer and waking up before dawn to cheer on clubs from English towns they’d never been to. Bennett was certain that the United States would soon embrace soccer—an intoxicating but tenuous idea that now reads as prescient. Millions of Americans have watched the men’s team make a run in the World Cup this year—the first time the country has hosted the men’s tournament since 1994. In fact, Monday’s game against Belgium reportedly pulled in more viewers than the 2025 World Series and Game 5 of the NBA Finals.
This is a moment of vindication for Men in Blazers, but it also has a greater meaning for Bennett, who became an American citizen in 2018. He routinely ends his podcast monologues with the phrase “I love America” and openly expresses his patriotism. “I grew up in the dying embers of the English class system,” he told me. “I felt very trapped. America always felt like the opposite, like opportunity. For me, that was life-affirming, life-saving.”
Maybe that all sounds cheesy. But it’s been hard to miss the collective enthusiasm and sense of American community emanating from the tournament: the initial delights of watching Scotland’s Tartan Army drink and bagpipe its way through New England, winning over steely Bostonians along the way; or the residents of Lawrence, Kansas, embracing the Algerian team with marching-band pep rallies. There was the viral pleasure and even pride in seeing foreigners encounter Buc-ee’s with reverential appreciation or giddily experience industrial quantities of Mountain Dew Baja Blast at Taco Bell.
Then there were the games themselves, seemingly broadcast on every available public television in America. Over the past month, I’ve seen strangers in the airport rapturously celebrating over last-second goals, gaggles of men and women kicking soccer balls on city streets while clad in a veritable United Nations of jerseys. People in my life who are normally allergic to sports have come over for marathon World Cup viewing sessions. At one point I found myself—a man from Cleveland—wearing a Senegal jersey, sweatily embracing a Senegalese man after a spectacular volley sent the ball to the back of the net. It felt like a dream.
Dreams always end, and in a way, that’s what happened on Monday. But Bennett’s larger ideology, that soccer is the “most important, least important thing,” may yet endure. There’s a reason he hired more than 100 people and retrofitted a coach bus into a mobile podcasting studio—a reason he spent the past month traveling across the country for a series of rowdy pre-match shows with fans. Moments like these forge bonds that don’t easily break.
Before the sun is up, I watch as Bennett goes on Morning Joe and CBS Mornings, and rips through a slew of radio and podcast hits. He calls nearly everyone he works with on these shows “a beautiful human being” and ends every interview with the same sign-off: “Big, big love to you.”
Bennett pops up from his table the instant his last interview ends. Within 30 seconds he’s on his phone, rifling off voice memos to people on his staff and agents of players he’s trying to book. “When’s Jon Stewart coming on?” he asks his producer, who pulls up a booking sheet a mile long. He completes seven phone calls in as many minutes just after 9 a.m., and tells each person on the other end that he loves them before hanging up. By 9:30, we’re in a production meeting for that day’s Men in Blazers live show; Bennett is Zooming in from his phone, because he’s outside picking up coffees for his team. A staffer on the call informs him that they’ve made a giant cardboard cutout of a lemon-pepper-wet chicken wing—an Atlanta delicacy—for somebody in the crowd to wave. “You could not be a more beautiful soul. You make the world a better place,” Bennett replies.
The day moves on. A detailed planning meeting is convened to discuss next summer’s Women’s World Cup. Men in Blazers is already pitching sponsors based on its online engagement numbers—according to Blinkfire, a marketing-analytics firm, the company has almost twice the engagement per follower across social media this tournament than the World Cup’s official broadcast partner, Fox Sports. Bennett out-hustles more established outlets by relentlessly blanketing the internet with instant-reaction clips, multiple daily podcasts, and memes; he then uses that attention to drive people to Men in Blazers’ frequent live shows, building an army of loyal obsessives. Listening to Bennett articulate this strategy and then jump into another TV interview, I’m struck by the realization that this whole operation resembles a political campaign as much as a media company. It’s hope and change and a bright new vision for America complete with stoppage time and hydration breaks.
Somewhere around coffee No. 5, Bennett and a few people from his team head to Busy Bee, a legendary soul-food café in Atlanta, to document some of the World Cup host city’s culture—another page from the campaign playbook, though Bennett’s excitement for the chicken and waffles is palpable. Outside next to the parking lot, sitting on a swelteringly hot picnic table, Bennett attempts to explain why he’s been averaging just three and a half hours of sleep a day for the past 26 days. He argues that what many people have been feeling over the past month, with each watch party or viral video of fans going crazy, is what the sociologist Émile Durkheim calls collective effervescence—a feeling of unity that emerges from a shared experience.
Courtesy of Charlie Warzel
Men in Blazers fans gather in Atlanta.
Bennett’s theory is that the World Cup, a rare monocultural event in today’s fragmented world, is one of the few reliable producers of such effervescent spectacles; it’s good for the internet, but also good for the soul. His media-distribution system ties it all together: “When Congo scores that goal, they need the immediate reaction, then they need the Congo fans across America; they need the scene in Kinshasa, the postgame on the field and the tears, the locker room, the Congo fans in Queens or Brooklyn,” he said. “It’s just incredible and it’s happening every single game.”
I can see why one would get swept up in what Bennett describes as “a very profound moment for the nation, which is so busy falling back in love with the world, and the world falling back in love with America.” The entire event feels at times like a summer vacation, a much-needed reprieve from harrowing world events. A cynic could see it as artificial or inauthentic—geopolitics, only without all the horrific stuff. And yet the moments are authentic, which may be why the White House intervention and FIFA’s arbitrary and unusual red-card suspension felt so jarring, even offensive, to die-hard fans of the tournament. It was a reminder that politics always lurk in the background, and that the euphoria is fragile. A U.S. loss would be a stress test.
About an hour into the game, with the U.S. down 2–1 to Belgium and looking positively shell-shocked, Bennett distracts himself by asking another Men in Blazers host, Becky Sauerbrunn, a former U.S. women’s team captain and two-time World Cup champion, what it feels like when you get scored on by your opponent immediately after you score. (Not good is the answer.) Later, when a flub by the U.S. goalkeeper puts Belgium up 3–1, Bennett yells at the television: “Not like this!” The agony is genuine. A close, competitive U.S. loss would hurt, but maybe it would be fine. This result, though—an impotent, scared team that is clearly not ready for the moment—feels like two steps forward and one very large step back. “This is not good for business,” Bennett said to me, cracking a weak smile.
What I’ll remember most about the game is how quickly confidence turned into insecurity. Shortly after the final whistle, announcers on Fox offered a sheepish plea to their massive broadcast audience: This doesn’t have to be the last game you watch. Sitting in a room with Bennett and Sauerbrunn and others like them, people who have devoted their lives to the sport, you could sense just how precarious the moment felt to them. How can you promise hope and change if the results stay the same?
Bennett was adamant that the Men in Blazers’ YouTube livestream start the minute the game ended, as if to avoid giving the audience too much time to think about what they’d seen. For almost two hours, Bennett and his co-hosts, plus special guests including the The Fault in Our Stars author John Green, dissected the match, which Bennett dubbed “a cup of tears and darkness, self-loathing, fear, and human agony.” As far as podcasting goes, the show should have been a grueling, dismal spectacle. But Bennett, who had been up early enough to text me at 4:35 a.m., was slaphappy, making jokes about Thomas the Tank Engine and quoting lines about pain from Dante’s Inferno. All the while, listeners poured into the stream, which became a strange, beautiful form of catharsis. “Football,” Green said one hour into the show, “is not about destinations; that’s the thing that we mistake all the time. It’s about journeys. Because if it was about destinations, no one would like it.”
After the show, Bennett rallied half a dozen people—many of whom had been up for almost a full 24 hours—for a midnight Waffle House excursion. “I’ll always remember this night and being with all of you,” he told the group. The pep talk reminded me of an anecdote Bennett had told me earlier. His mother-in-law had become obsessed with taking family photos at every get-together, no matter how inconvenient or annoying. “What she’s realized is that photographs and memories are inextricably connected,” he said. “I’m very aware of how memory is formed. And in this moment, this tournament, memory is being formed.” Suddenly, it clicked: Bennett is obsessively documenting the insanity, the joy, the collective effervescence so that we remember how this all felt. So we don’t lose it.
I had plans to meet Bennett early the next morning, but I slept through my alarm. When I texted to apologize, he confessed that he had almost missed a morning TV hit, having fallen asleep, laptop on his chest. Political campaigns are exhausting. But Bennett wasn’t defeated. Too much progress had been made. “As I speak to you, I am headed to Atlanta Stadium to watch Lionel Messi face Mo Salah,” he texted back. “The streets are packed. Football is here and thousands of people are ready to make memories again.”
In recent weeks, Silicon Valley has been fawning over an AI model released by a lab in China. The program, called GLM-5.2, has been called a “marvel,” “very good,” and a “step change.” The billionaire venture capitalist Marc Andreessen posted on X that “AI insiders are saying GLM-5.2 is the first Chinese AI model to match and often beat” the top public U.S. models. Guillermo Rauch, the CEO of the AI-coding platform Vercel, said that he is “genuinely impressed, almost shocked” by GLM-5.2’s coding abilities. Or as one AI founder at a San Francisco dinner party recently told me: “Praise GLM-5.2.”
In a sense, GLM-5.2 is China’s answer to Claude Code, Anthropic’s agentic assistant that has reshaped the AI boom. This has been the year of AI agents—tools that don’t just chat but promise to do stuff on your behalf, whether coding a website or booking a vacation. Chinese AI models have been steadily improving, but none previously proved capable or consistent enough to be used as agents. Now GLM-5.2, developed by the Chinese company Z.ai, rivals some of OpenAI’s and Anthropic’s top offerings and, by many measures, has leaped ahead of Google Gemini. And it is several times cheaper.
Despite Silicon Valley’s awe over GLM-5.2, an inexpensive competitor couldn’t have come at a worst time for America’s frontier AI labs. Having successfully persuaded corporate America to give their products a try, OpenAI, Anthropic, and Google are now struggling to prove that their tools are worth the money. These bots can be very expensive to use, running up bills into the thousands of dollars, per employee, per month. Uber reportedly spent its entire 2026 budget for Anthropic models in only a few months. Other Big Tech companies including Meta, Amazon, Tesla, and Adobe are also reportedly clamping down on employee AI usage. Citi at one point shut down employee access to OpenAI’s and Anthropic’s most expensive models, according to reporting from 404 Media (which Citi has contested).
Even though political leaders and tech executives have framed the U.S. as in the middle of a contentious technological race with China, Americans can still use Chinese models. While it’s too soon to know whether GLM-5.2 is really capable of replacing America’s top-tier AI agents, any firm or developer who is balking at the costs now might have an alternative. The arrival of GLM-5.2 poses a business dilemma for Silicon Valley—and possibly a national-security dilemma for the country as a whole.
In some ways, America’s AI industry has been here before. Many previous impressive Chinese AI models have not triggered a mass exodus of OpenAI and Anthropic customers, with one major exception: In January 2025, after DeepSeek launched a cheap AI model on par with America’s best, adoption of Chinese AI models leapt up. Within two months, the share of global web traffic to Chinese AI models jumped from roughly 3 to 13 percent, according to research from RAND. OpenAI, Anthropic, and Google quickly responded to DeepSeek with cheaper models of their own, and then the rise of AI agents early this year seemed to solidly put American labs back on top.
Even before Z.ai launched its new model, there have been quiet signs of a shift to cheaper, Chinese models amid mounting concerns about AI bills. Over the first five months of 2026, DeepSeek’s adoption among the 70,000 U.S. firms that use Ramp, a financial-operations platform, increased from 0.1 to 0.3 percent, Ara Kharazian, Ramp’s lead economist, told me. Kharazian added that 6 percent of Ramp customers that spend money on AI use third-party platforms that provide access to many different AI products. On OpenRouter, one such platform, the six most popular AI models are Chinese; in under a month, GLM-5.2 already ranks fifth. These data don’t capture all the software developers and companies directly downloading Chinese AI models, which are typically open source, and configuring them on their own computers—that is, largely scrappy start-ups and academics who don’t have the budget to use a fintech service like Ramp. Chinese models accounted for nearly half of all open-source AI downloads from February 2025 to 2026, according to data from the popular AI platform Hugging Face.
Meanwhile, OpenAI and Anthropic keep releasing more, rather than less, expensive products. And soon, they will likely have even more competition. The Claude Code and broader AI agent frenzy started seven months ago, which is also approximately how far behind Chinese AI firms have been lagging in model development. China’s other labs, including DeepSeek and Moonshot AI, will almost certainly release similarly capable and inexpensive AI agents before long. Coinbase, a popular crypto company, claims to have nearly cut its AI spending in half by defaulting to cheaper models including GLM-5.2 and Kimi, another popular Chinese bot. “The scenario to worry about is China has good-enough models at a quarter of the price,” Kyle Siler-Evans, an AI researcher at RAND, told me. “I think that is likely the future we’re headed toward.”
Kharazian cautioned against overstating this narrative. Whereas some tech-forward firms are pulling back on their AI spending, most American companies already spend very little on the technology: The median Ramp customer spends just $11 per employee on AI. “Not to say that the rest of the market won’t go in that direction,” Kharazian said, but Anthropic and OpenAI will have “ample time to respond” with competitive pricing—as they did with DeepSeek.
Perhaps the biggest obstacle to the ascendance of GLM-5.2, at least in the U.S., will have little to do with model capabilities or pricing. There are seriousconcerns about Chinese firms essentially using their AI to hoover up sensitive data and steal corporate secrets. The perceived risks of using Chinese technology could create a profound chilling effect among the customers of any U.S. firm—not to mention the uncertainty created by the possibility of federal regulation. Consider that Chinese electric vehicles are by all accounts better and cheaper than their Western counterparts, but they can’t be bought in the U.S.; one possibility is a near future in which Americans are functionally or legally barred from accessing the most cost-effective AI models.
The biggest implications of GLM-5.2 and the coming onslaught of cheap Chinese AI agents may also be geopolitical rather than economic. Whatever soft power the U.S. gains from its technology may wither as software developers in countries with less tense relationships with China turn to DeepSeek and GLM-5.2 in greater numbers (consider how the EU, U.K., and Canada are importing Chinese EVs). Meanwhile, despite Big Tech spending ever greater sums on advanced AI chips to train their models—chips that are banned from export to China—the gap between U.S. and Chinese models has not widened; if anything, it may be shrinking. That means any military, economic, cyberoffensive, or other advantage that AI can grant the nation could be vanishing. America’s lead in the AI race, for the first time since DeepSeek, is at real risk of slipping.
When I was 16, I did something I’m embarrassed to admit: I waited in a long line to buy a video game called Assassin’s Creed III. Over the past few days, though, that experience has become ever so slightly tinged with nostalgia. Last week, Sony announced that, starting in 2028, new PlayStation games will be available only as digital downloads rather than physical discs. Will kids ever get to embarrass themselves like this again?
The transition is indicative of a greater loss—one that I, and I suspect many people, have mixed feelings about. More and more, entertainment companies are ushering people into seamless digital spaces where convenience replaces ownership: games without discs, music on Spotify instead of vinyl, movies on Netflix instead of a theater.
Markets change, and entertainment changes with them; this has always been the case. Before the advent of at-home consoles, public-arcade games were built to be played in short, endlessly repeatable loops—the goal being to pull in more and more quarters with each new attempt. In the 1990s, some developers padded out the lengths of their games and in some cases made them artificially difficult, to prevent customers from completing them during rental periods instead of buying them.
And there’s a fairly obvious supply-and-demand explanation for why the music industry stopped hawking CDs; why DVD-bargain bins have disappeared from some American supermarkets; and why GameStop, once the central node of a booming video-game retail industry, has closed much of its brick-and-mortar empire. When the PlayStation 4 launched, in 2013, just 13 percent of Sony’s video games were being purchased digitally. Last year, that number reportedly approached 80 percent. In its announcement, Sony underscored that the transition from physical to digital is part of an effort to “align more closely” with how most gamers are already choosing to play.
Developers see it too: Last month, the publisher behind the upcoming Grand Theft Auto VI—which is expected to be one of the biggest entertainment launches of all time—revealed that the “physical” edition of the game would actually be a box with just a download code inside. Sony’s announcement “probably does move the markets in a big way towards digital,” Giorgo Paizanis, who leads the consulting firm BCG’s gaming work, told me. “But we’re already 80 percent there.” (PlayStation did not respond to a request for comment.)
The convenience of downloads may be an upside, though there are certainly real downsides in the transition away from physical media. When you buy a disc, you own it and can resell it or lend it out the old fashioned way—without online mediation. No corporate middleman was watching me hand my copy of Red Dead Redemption to my friend when I was done with it. Although physical games can be damaged or even decay over long periods of time, I could still loan that same copy out today if I wanted. Digital purchases only grant you a license for use, and that license can be revoked. At around the same time that Sony announced the digital transition for PlayStation, it also alerted customers that more than 500 movies and shows would be pulled from its online marketplace, removing them from the libraries of users who had purchased them. (As an Arts Technica headline put it, “We’re Reminded We Don’t Own What We Buy.”)
Physical games also function as markers of personal taste, just as sprawling record collections or personal libraries of dusty paperbacks do. A copy of Elden Ring likely conveys a different impression than a Criterion Collection Blu-ray, but video-game discs at least may one day radiate a kind of old-school cool. (Early NES cartridges may already possess this energy.) There’s a reason that people are still nostalgic for Blockbuster, and that the decline of physical-media sales slowed dramatically in 2025.
When it comes to vinyl records, at least, part of their appeal today lies in their physicality. Flipping and periodically cleaning a record ultimately personalizes the experience of listening to it. Up close, you can see how the rumblings of a bass guitar, or the vibrations of a clarinet’s reed, are transcribed in the grooves. The motion of playback rhymes with the mode of artistic production. And because you can fit only so many of these records in your house, you inevitably end up listening and relistening to the same ones. Connections form.
Video-game discs are not vinyl. Records produce their own sound, which might be distorted even by a hint of dust or moisture, whereas discs are used to load entirely virtual experiences. If they break, they break. I won’t miss that, even as I recall the feeling of popping that Assassin’s Creed III Blu-ray out of its box, the sound of the disc whirring in my console—small pleasures sacrificed in the name of progress.
Login
