What to expect from Google this week

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

When Google opens its doors tomorrow for its annual developer conference, I/O, it will do so as a clear third place in the foundation model race. A year ago, at Google I/O 2025, the situation looked very different: The company was still riding high from the launch of Gemini 2.5 Pro that March, and distinguishing among the top-tier large language models often felt like a subjective splitting of hairs. 

But a foundation model’s reputation these days rests largely on its coding capabilities, and for months Google’s coding tools have been outgunned by Anthropic’s Claude Code and OpenAI’s Codex. Those systems are so dramatically superior to Google’s own offerings that the company has reportedly had to allow some engineers at DeepMind, its AI division, to use Claude for their work—lest they fall farther behind.

So when I arrive at the conference in Mountain View, California tomorrow, I’ll certainly be on the lookout for any efforts Google is making to claw its way back into frontrunner position. But I’m also eager to see new developments in areas where Google shapes the cutting edge, such as AI for science. The company’s moves there might receive less attention, but they will be no less consequential. 

Here are three things I’ll be paying particular attention to over the next two days.

An attempted coding comeback

Google is taking its AI coding crisis seriously. According to reporting from The Information, there’s a new AI coding team at DeepMind. And the Los Angeles Times has reported that John Jumper, who shared a 2024 Nobel Prize in chemistry with DeepMind CEO Demis Hassabis for their work on the protein structure prediction software AlphaFold, is lending his talents to the efforts. I would be surprised if we don’t see a major new coding release at I/O, perhaps in the form of an update to the company’s Antigravity agentic coding platform.

That said, we shouldn’t expect anything transformative here. Googlers have access to models and products that are substantially ahead of those released to the public, yet they were still reportedly fighting over who got access to Claude Code last month. Unless the company has made astonishing progress since then, Google probably won’t make it back to the coding frontier in the next two days.

Science and health

Coding might be Google DeepMind’s weakness, but science is its conspicuous strength. It is the only frontier AI company to have earned a Nobel Prize. And as LLMs have come to dominate the AI-for-science landscape, Google has only solidified its lead. Last year, the company released multiple scientific AI tools, including the AI co-scientist, which formulates hypotheses and research plans in response to user questions and has been described as an “oracle” by one Stanford scientist, and AlphaEvolve, a system that iteratively discovers new solutions for mathematical and computational problems. If any new scientific tools are announced at I/O, they’ll be worth noting.

I’ll also be paying close attention to any moves Google makes in health and medicine. Google is doing some of the best research out there on LLM-based health tools, but OpenAI has defined the health AI conversation since the release of ChatGPT Health in January. Google has announced that it will be making its AI-powered Health Coach publicly available tomorrow, but promotional material suggests that the tool is geared more toward providing advice on topics such as fitness and diet than to addressing users’ medical concerns. Is this another area where Google has fallen behind, or is the company exercising appropriate caution in a high-stakes domain? 

The drama

While Google fans congregate down in Mountain View, roughly 30 miles north in Oakland the Elon Musk v. Sam Altman trial will be wrapping up. The past few months have seen more than their fair share of AI CEO drama—before the trial, the animosity between Altman and Anthropic CEO Dario Amodei took center stage as Anthropic and OpenAI worked to negotiate deals with the US Department of Defense. But DeepMind’s Hassabis has, for the most part, steered clear of such drama. He effectively presents himself as a Nobel Prize-winning nerd, and if he has written screeds about any of his peers, they haven’t been leaked to the press or appeared in legal discovery.

That’s not to say that Google is controversy free. Last month, a group of 600 employees, many of whom work for DeepMind, sent a letter to CEO Sundar Pichai protesting an impending DoD deal. Google signed that deal the next day. Hassabis, Pichai, and all the other big names will surely do their best to skirt these and other touchy subjects while on stage, but controversies will worm their way in regardless. It will be interesting to see whether Google can maintain its veneer of neutrality.

Three things in AI to watch, according to a Nobel-winning economist

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

A few months before he was awarded the Nobel Prize in economics in 2024, Daron Acemoglu published a paper that earned him few fans in Silicon Valley. Contrary to what Big Tech CEOs had been promising—an overhaul of all white-collar work—Acemoglu estimated that AI would give only a small boost to US productivity and would not obviate the need for human work. It’s okay at automating certain tasks, he wrote, but some jobs will be perfectly fine.

Two years later, Acemoglu’s measured take has not caught on. Chatter about an AI jobs apocalypse pops up everywhere from Senator Bernie Sanders’s rallies to conversations I overhear in line at the grocery store. Some previously skeptical economists have gotten more open to the idea that something seismic could be coming with AI. A California gubernatorial candidate said last week that he wants to tax corporate AI use and pay victims of “AI-driven layoffs.” 

On the one hand, the data is still on Acemoglu’s side; studies repeatedly find that AI is not affecting employment rates or layoffs. But the technology has advanced quite a bit since his cautious predictions. I spoke with him to understand if any of the latest developments in AI have changed his thesis, and to find out what does worry him these days if not imminent AGI.

AI agents

One of the biggest technical leaps in AI since Acemoglu’s paper has been agentic AI, or tools that can go beyond chatbots and operate on their own to complete the goal you give them. Because they can work independently rather than just answering questions, companies are increasingly pitching agents as a one-to-many replacement for human workers.

“I think that’s just a losing proposition,” Acemoglu says. He thinks agents are better thought of as tools to augment particular pieces of someone’s work than something malleable enough to handle a person’s whole job.

One reason has to do with all the various tasks that go into a job, something Acemoglu has been researching in his work on AI since 2018. For example, an x-ray technician juggles 30 different tasks, from taking down patient histories to organizing archives of mammogram images. A worker can naturally switch between formats, databases, and working styles to do this, Acemoglu says, but how many individual tools or protocols would an AI require to do the same?

Whether or not agents will supercharge AI’s impact on jobs will come down to whether they can eventually handle the orchestration between tasks that humans do naturally. AI companies are in heated competition to prove that their AI agents can work independently for ever longer periods without making mistakes, sometimes exaggerating the results—but Acemoglu says many jobs will be spared from an AI takeover if agents can’t fluidly switch between tasks.

The new hiring spree

For years Big Tech has been offering staggering salaries to recruit AI researchers. But I asked Acemoglu about a different hiring spree I’ve noticed: AI companies are all building in-house economics teams.

OpenAI hired Ronnie Chatterji from Duke University in 2024 to be its chief economist and announced last year that Chatterji will work with Jason Furman—Harvard economist and former advisor to Barack Obama—to research AI and jobs. Anthropic has convened a group of 10 leading economists to do similar work. And just last week, Google DeepMind announced it had hired Alex Imas, an economist from the University of Chicago, to be its “director of AGI economics.”

Acemoglu has noticed colleagues getting snatched up for these roles too. “It makes sense,” he says: AI companies are well aware that public skepticism about AI, in large part due to job concerns, is growing. And they have strong incentives to shape the economic narrative around their technology (consider OpenAI’s latest proposal for a new era of industrial policy).

“What I hope we won’t get,” Acemoglu says, “is that they’re interested in economists just to further their viewpoints or further the hype.” That tension hangs over the emerging field of “AI economics”; it’s concerning that some of the most influential research about AI’s impact on work may increasingly come from the companies with the most to gain from favorable conclusions.

AI apps

I don’t think of AI as hard to use; most of us interact with it via chatbots that use plain language. But Acemoglu says we should consider how it compares with the sort of software that kicked off earlier tech transformations, like PowerPoint for slide decks and Word for documents. 

“Anybody could install these on their computer and get them to do the things that they want them to do,” he says. They spread accordingly. 

“We have not seen the development of apps based on AI that have the same usability,” he says. Even if anyone can chat with an AI model, it tends to take a while for the average worker to get practical and productive use out of it. That’s part of the reason why AI has not yet shown any seismic impact on the job market or the economy. One of the key signals Acemoglu is watching, then, is the creation of apps that make AI easier to use. 

But he acknowledges that for a while, we’re going to see all sorts of conflicting evidence about AI: anecdotes that college grads are finding the job market worse and worse, but no noticeable effect of AI on productivity, for example. “There’s a huge amount of uncertainty,” he says. And that’s the most telling thing about the AI economy right now: the certainty of the rhetoric alongside the uncertainty of everything else.

Week one of the Musk v. Altman trial: What it was like in the room

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

Two of the most powerful people in AI—Sam Altman and Elon Musk—began their face-off in court in Oakland, California, last week. Musk is suing OpenAI, alleging that the millions he spent to fund it around a decade ago were meant for a nonprofit, not a corporation, and that the company has reneged on that mission since. 

The stakes are high—even a partial win for Musk could set OpenAI back as it reportedly plans to go public this year. But most of the attention comes from the spectacle of a feud on X now playing out in federal court. “Cringey texts, raw diary entries, and endless scheming behind the founding and growth of OpenAI are expected to come to light,” my colleague Michelle Kim wrote before it began. And the trial unfolds as the cultural backlash against AI swells; some of the signs held by protesters outside the courthouse suggest that to a significant number of people, whatever the outcome of Musk v. Altman, we all lose.  

Most of us have had to observe the trial from afar, but Michelle, who also happens to be a lawyer, has been in court each day. I caught up with her to learn what’s unfolded thus far and what might come next.

Can you give us the overview of what this case is actually about? What exactly is being decided, and who is favored right now?

Elon Musk is arguing that Sam Altman and OpenAI president Greg Brockman have breached the company’s charitable trust by effectively converting OpenAI into a for-profit company. Musk alleges that is not what they promised him in the company’s early days. He has asked for several remedies, like a crazy amount of damages and removing Sam Altman. But the main remedy he wants is unwinding OpenAI’s restructuring. [In October 2025 OpenAI struck deals with the attorneys general of California and Delaware that would essentially allow its nonprofit portion to have less day-to-day control of OpenAI. It’s a compromise from what OpenAI originally proposed, but Musk still wants to stop it.] 

OpenAI argues that Elon Musk actually agreed to have the company operate a for-profit arm, because he knew building AI is very expensive. So it’s about proving what Musk knew, what he didn’t know, and whether he really was deceived by Altman and Brockman.

There’s a big debate about when exactly Musk found out about this alleged misconduct. Musk founded OpenAI with Altman and Brockman in 2015, and he brought the suit in 2024. There’s a statute of limitations for charitable trust claims; you need to have brought a claim within three to four years after you find out about the alleged misconduct. So Musk tries to paint a picture that back in the day he was a little suspicious, but that it was really only in 2022 that he realized OpenAI was no longer committed to its original charitable mission, and that he had been scammed. It’s only the first week of trial, but I’m not sure Musk has proved this to the judge and jury.

What were some standout moments thus far?

At one point one of Elon Musk’s lawyers said, “We could all die as a result of AI.” I think a lot of the people in the room were really shaken by this comment, and the judge told Musk’s lawyer: You talk about all these safety risks that OpenAI has when building AI, but Musk is also creating a company that’s in the same exact space. She basically said, I’m sure there’s plenty of people who also don’t want to put the future of humanity in Elon Musk’s hands. 

And then the lawyers just kept going on and on about the catastrophic risks of AI and whether Elon Musk or OpenAI was in the better position to steward AI safety. And the judge sort of snapped. She said very sternly that this trial was not about whether or not artificial intelligence has damaged humanity. And I thought that was a really striking standout moment of the trial that pointed at how even though it is technically just about whether Elon Musk was really deceived by OpenAI, it’s also become a huge discussion about AI safety and some of the practices that the labs are engaging in when building AI. 

Can you give us a look behind the curtain at how getting into this trial works?

There are tons of reporters. This is a very high-profile suit, so I have to wake up around 4:30 a.m. and show up to the Oakland courthouse at 6 a.m. sharp to get in line. And on some days, even 6 a.m. doesn’t get you into the courtroom. There are lots of photographers in front of the courthouse, especially on days when you know Musk or Altman and Brockman are present. And there’s also some concerned citizens who want to watch the trial. I usually have to wait, like, two hours in line to get in to be one of the 30 people who claim the unreserved seats in the courtroom. 

What has it felt like to see Elon Musk testify? How would you describe his demeanor?

He shows up in a crisp black suit. He can be this inflammatory person on X, but in the courtroom, he is calm, cool, collected, and looks very comfortable. He has been in a lot of lawsuits. He knows how to talk to the jury and how to present himself in front of them and the judge. He’s also cracking jokes with his lawyer and even the opposing party’s lawyer and the judge. 

And he can be witty. There was this one moment when OpenAI’s lawyer was asking Musk a question and sort of fed him an answer. And Musk said “That’s not a leading question, that’s a leading answer.” The judge intervened and said, “You’re not a lawyer, Elon.” And then he was like, “Well, I did take Law 101.”

That said, he does get flustered and uncomfortable when OpenAI’s lawyer asks tough, piercing questions. Which he’s been doing.

What are the biggest things we’ve learned that weren’t clear in the earlier phases of this case?

On the fourth day of the trial, Musk admitted during cross-examination that xAI distills OpenAI’s models to train its own models, which was shocking. Musk followed up by saying that this is standard practice among all the labs now and that xAI wasn’t doing anything beyond what others were already doing. But a lot of the journalists started typing away at their laptops as soon as Musk made this comment. 

I also learned that there’s just so much scheming among Big Tech executives. You know about it vaguely, but to hear firsthand accounts and read their emails and text messages is fascinating. 

For example, there was a text message between Musk and Mark Zuckerberg of Meta, where they’re kind of teaming up to stop OpenAI’s restructuring. They’re even trying to make a bid to buy all the assets of OpenAI’s nonprofit. The level of scheming that goes on among these executives is mind-blowing.

What happens next?

OpenAI’s president, Greg Brockman, who was meticulously taking notes during some of Elon Musk’s testimony, is expected to testify next week. And Stuart Russell, a computer scientist at UC Berkeley, will testify about AI safety. I’m expecting that to open the floodgates to this crazy discussion about who can be trusted to build AI. 

A bunch of other high-profile people are expected to testify, like former OpenAI chief scientist Ilya Sutskever, former CTO Mira Murati, and Microsoft CEO Satya Nadella. 

The trial is supposed to last around three weeks. The nine jurors will deliver an advisory verdict that guides the judge on how to decide Musk’s claims against OpenAI. The judge doesn’t have to listen to the jury and can decide however she wants. If she decides OpenAI is liable, then she’ll decide what sort of remedies are appropriate. 

MIT Technology Review will have ongoing coverage of Musk v. Altman until its conclusion. Follow @techreview or @michelletomkim on X for up-to-the-minute reporting.

The missing step between hype and profit

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

In February, I picked up a flyer at an anti-AI march in London. I can’t say for sure whether or not its writers meant to riff on South Park’s underpants gnomes. But if they did, they nailed it: “Step 1: Grow a digital super mind,” it read. “Step 2: ? Step 3: ?”

Produced by Pause AI, an international activist group that co-organized the protest, it ended with this plea to the reader: “Pause AI until we know what the hell Step 2 is.” 

In the South Park episode “Gnomes,” which first aired in 1998, Kenny, Kyle, Cartman, and Stan discover a community of gnomes that sneak out at night to steal underpants from dressers. Why? The gnomes present their pitch deck. “Phase 1: Collect underpants. Phase 2: ? Phase 3: Profit.”

The gnomes’ business plan has since become one of the greats among internet memes, used to satirize everything from startup strategies to policy proposals. Memelord in chief Elon Musk once invoked it in a talk about how he planned to fund a mission to Mars. Right now, it captures the state of AI. Companies have built the tech (Step 1) and promised transformation (Step 3). How they get there is still a big question mark.

As far as Pause AI is concerned, Step 2 must involve some kind of regulation. But exactly what it will call for and who will enforce it are up for debate.

AI boosters, on the other hand, are convinced that Step 3 is salvation and tend to glaze over the middle bit. They see us racing toward sunny uplands on the back of an “economically transformative technology,” as OpenAI’s chief scientist, Jakub Pachocki, put it to me a few weeks ago. They know where they want to go—more or less: It’s hazy up there and still some way off. But everyone’s taking a different route. Will they all make it? Will anyone?

For every big claim about the future, there is a more sober assessment of how the rubber meets the road—one that quells the hype. Consider two recent studies. One, from Anthropic, predicted what types of jobs are going to be most affected by LLMs. (A takeaway: Managers, architects, and people in the media should prepare for change; groundskeepers, construction workers, and those in hospitality, not so much.) But their predictions are really just guesses, based on what kinds of tasks LLMs seem to be good at rather than how they really perform in the workplace.   

Another study, put out in February by researchers at Mercor, an AI hiring startup, tested several AI agents powered by top-tier models from OpenAI, Anthropic, and Google DeepMind on 480 workplace tasks frequently carried out by human bankers, consultants, and lawyers. Every agent they tested failed to complete most of its duties.   

Why is there such wide disagreement? There are a number of factors. For a start, it’s crucial to consider who is making the claims (and why). Anthropic has skin in the game. What’s more, most of the people telling us that something big is about to happen have reached that conclusion largely on the basis of how fast AI coding tools are getting. But not all tasks can be hacked with coding. Other studies have found that LLMs are bad at making strategic judgment calls, for example.

What’s more, when they’re deployed, the tools aren’t just dropped into a cleanroom. They need to work in places contaminated with people and existing workflows. And sometimes adding AI will make things worse. Sure, maybe those workflows need to be torn up and refashioned around the new technology for it to achieve transformative status, but that will take time (and guts).  

That big hole? It’s right where Step 2 should be. The lack of agreement on exactly what’s about to happen—and how—creates an information vacuum that gets filled by the latest wild claim of the week, evidence be damned. We’re so unmoored from any real understanding of what’s coming and how it will be deployed that a single social media post can (and does) shake markets.

We need fewer guesses and more evidence. But that’s going to require transparency from the model makers, coordination between researchers and businesses, and new ways to evaluate this technology that tell us what really happens when it’s rolled out in the real world.

The tech industry (and with it the world’s economy) rests on the held-out promise that AI really will be transformative. But that is not yet a sure bet. Next time you hear bold claims about the future, remember that most businesses are still figuring out what to do with their underpants.

Why opinion on AI is so divided

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

In an industry that doesn’t stand still, Stanford’s AI Index, an annual roundup of key results and trends, is a chance to take a breath. (It’s a marathon, not a sprint, after all.)

This year’s report, which dropped today, is full of striking stats. A lot of the value comes from having numbers to back up gut feelings you might already have, such as the sense that the US is gunning harder for AI than everyone else: It hosts 5,427 data centers (and counting). That’s more than 10 times as many as any other country.  

There’s also a reminder that the hardware supply chain the AI industry relies on has some major choke points. Here’s perhaps the most remarkable fact: “A single company, TSMC, fabricates almost every leading AI chip, making the global AI hardware supply chain dependent on one foundry in Taiwan.” One foundry! That’s just wild.

But the main takeaway I have from the 2026 AI Index is that the state of AI right now is shot through with inconsistencies. As my colleague Michelle Kim put it today in her piece about the report: “If you’re following AI news, you’re probably getting whiplash. AI is a gold rush. AI is a bubble. AI is taking your job. AI can’t even read a clock.” (The Stanford report notes that Google DeepMind’s top reasoning model, Gemini Deep Think, scored a gold medal in the International Math Olympiad but is unable to read analog clocks half the time.)

Michelle does a great job covering the report’s highlights. But I wanted to dwell on a question that I can’t shake. Why is it so hard to know exactly what’s going on in AI right now?  

The widest gap seems to be between experts and non-experts. “AI experts and the general public view the technology’s trajectory very differently,” the authors of the AI Index write. “Assessing AI’s impact on jobs, 73% of U.S. experts are positive, compared with only 23% of the public, a 50 percentage point gap. Similar divides emerge with respect to the economy and medical care.”

That’s a huge gap. What’s going on? What do experts know that the public doesn’t? (“Experts” here means US-based researchers who took part in AI conferences in 2023 and 2024.)

I suspect part of what’s going on is that experts and non-experts base their views on very different experiences. “The degree to which you are awed by AI is perfectly correlated with how much you use AI to code,” a software developer posted on X the other day. Maybe that’s tongue-in-cheek, but there’s definitely something to it.

The latest models from the top labs are now better than ever at producing code. Because technical tasks like coding have right or wrong results, it is easier to train models to do them, compared with tasks that are more open-ended. What’s more, models that can code are proving to be profitable, so model makers are throwing resources at improving them.

This means that people who use those tools for coding or other technical work are experiencing this technology at its best. Outside of those use cases, you get more of a mixed bag. LLMs still make dumb mistakes. This phenomenon has become known as the “jagged frontier”: Models are very good at doing some things and less good at others.

The influential AI researcher Andrej Karpathy also had some thoughts. “Judging by my [timeline] there is a growing gap in understanding of AI capability,” he wrote in reply to that X post. He noted that power users (read: people who use LLMs for coding, math, or research) not only keep up to date with the latest models but will often pay $200 a month for the best versions. “The recent improvements in these domains as of this year have been nothing short of staggering,” he continued.

Because LLMs are still improving fast, someone who pays to use Claude Code will in effect be using a different technology from someone who tried using the free version of Claude to plan a wedding six months ago. Those two groups are speaking past each other.

Where does that leave us? I think there are two realities. Yes, AI is far better than a lot of people realize. And yes, it is still pretty bad at a lot of stuff that a lot of people care about (and it may stay that way). Anyone making bets about the future on either side should bear that in mind.

The one piece of data that could actually shed light on your job and AI

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

Within Silicon Valley’s orbit, an AI-fueled jobs apocalypse is spoken about as a given. The mood is so grim that a societal impacts researcher at Anthropic, responding Wednesday to a call for more optimistic visions of AI’s future, said there might be a recession in the near term and a “breakdown of the early-career ladder.” Her less-measured colleague Dario Amodei, the company’s CEO, has called AI “a general labor substitute for humans” that could do all jobs in less than five years. And those ideas are not just coming from Anthropic, of course. 

These conversations have unsurprisingly left many workers in a panic (and are probably contributing to support for efforts to entirely pause the construction of data centers, some of which gained steam last week). The panic isn’t being helped by lawmakers, none of whom have articulated a coherent plan for what comes next.

Even economists who have cautioned that AI has not yet cut jobs and may not result in a cliff ahead are coming around to the idea that it could have a unique and unprecedented impact on how we work. 

Alex Imas, based at the University of Chicago, is one of those economists. He shared two things with me when we spoke on Friday morning: a blunt assessment that our tools for predicting what this will look like are pretty abysmal, and a “call to arms” for economists to start collecting the one type of data that could make a plan to address AI in the workforce possible at all. 

On our abysmal tools: consider the fact that any job is made up of individual tasks. One part of a real estate agent’s job, for example, is to ask clients what sort of property they want to buy. The US government chronicled thousands of these tasks in a massive catalogue first launched in 1998 and updated regularly since then. This was the data that researchers at OpenAI used in December to judge how “exposed” a job is to AI (they found a real estate agent to be 28% exposed, for example). Then in February, Anthropic used this data in its analysis of millions of Claude conversations to see which tasks people are actually using its AI to complete and where the two lists overlapped.

But knowing the AI exposure of tasks leads to an illusory understanding of how much a given job is at risk, Imas says. “Exposure alone is a completely meaningless tool for predicting displacement,” he told me.

Sure, it is illustrative in the gloomiest case—for a job in which literally every task could be done by AI with no human direction. If it costs less for an AI model to do all those tasks than what you’re paid—which is not a given, since reasoning models and agentic AI can rack up quite a bill—and it can do them well, the job likely disappears, Imas says. This is the oft-mentioned case of the elevator operator from decades ago; maybe today’s parallel is a customer service agent solely doing phone call triage. 

But for the vast majority of jobs, the case is not so simple. And the specifics matter, too: Some jobs are likely to have dark days ahead, but knowing how and when this will play out is hard to answer when only looking at exposure.

Take writing code, for example. Someone who builds premium dating apps, let’s say, might use AI coding tools to create in one day what used to take three days. That means the worker is more productive. The worker’s employer, spending the same amount of money, can now get more output. So then will the employer want more employees or fewer? 

This is the question that Imas says should keep any policymaker up at night, because the answer will change depending on the industry. And we are operating in the dark. 

In this coder’s case, these efficiencies make it possible for dating apps to lower prices. (A skeptic might expect companies to simply pocket the gains, but in a competitive market, they risk being undercut if they do.) These lower prices will always drive some increase in demand for the apps. But how much? If millions more people want it, the company might grow and ultimately hire more engineers to meet this demand. But if demand barely ticks up—maybe the people who don’t use premium dating apps still won’t want them even at a lower price—fewer coders are needed, and layoffs will happen.

Repeat this hypothetical across every job with tasks that AI can do, and you have the most pressing economic question of our time: the specifics of price elasticity, or how much demand for something changes when its price changes. And this is the second part of what Imas emphasized last week: We don’t currently have this data across the economy. But we could. 

We do have the numbers for grocery items like cereal and milk, Imas says, because the University of Chicago partners with supermarkets to get data from their price scanners. But we don’t have such figures for tutors or web developers or dietitians (all jobs found to have “exposure” to AI, by the way). Or at least not in a way that’s been widely compiled or made accessible to researchers; sometimes it’s scattered across private companies or consultancies. 

“We need, like, a Manhattan Project to collect this,” Imas says. And we don’t need it just for jobs that could obviously be affected by AI now: “Fields that are not exposed now will become exposed in the future, so you just want to track these statistics across the entire economy.”

Getting all this information would take time and money, but Imas makes the case that it’s worth it; it would give economists the first realistic look at how our AI-enabled future could unfold and give policymakers a shot at making a plan for it.

The Pentagon’s culture war tactic against Anthropic has backfired

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

Last Thursday, a California judge temporarily blocked the Pentagon from labeling Anthropic a supply chain risk and ordering government agencies to stop using its AI. It’s the latest development in the month-long feud. And the matter still isn’t settled: The government was given seven days to appeal, and Anthropic has a second case against the designation that has yet to be decided. Until then, the company remains persona non grata with the government. 

The stakes in the case—how much the government can punish a company for not playing ball—were apparent from the start. Anthropic drew lots of senior supporters with unlikely bedfellows among them, including former authors of President Trump’s AI policy.

But Judge Rita Lin’s 43-page opinion suggests that what is really a contract dispute never needed to reach such a frenzy. It did so because the government disregarded the existing process for how such disputes are governed and fueled the fire with social media posts from officials that would eventually contradict the positions it took in court. The Pentagon, in other words, wanted a culture war (on top of the actual war in Iran that began hours later). 

The government used Anthropic’s Claude for much of 2025 without complaint, according to court documents, while the company walked a branding tightrope as a safety-focused AI company that also won defense contracts. Defense employees accessing it through Palantir were required to accept terms of a government-specific usage policy that Anthropic cofounder Jared Kaplan said “prohibited mass surveillance of Americans and lethal autonomous warfare” (Kaplan’s declaration to the court didn’t include details of the policy). Only when the government aimed to contract with Anthropic directly did the disagreements begin. 

What drew the ire of the judge is that when these disagreements became public, they had more to do with punishment than just cutting ties with Anthropic. And they had a pattern: Tweet first, lawyer later. 

President Trump’s post on Truth Social on February 27 referenced “Leftwing nutjobs” at Anthropic and directed every federal agency to stop using the company’s AI. This was echoed soon after by Defense Secretary Pete Hegseth, who said he’d direct the Pentagon to label Anthropic a supply chain risk. 

Doing so necessitates that the secretary take a specific set of actions, which the judge found Hegseth did not complete. Letters sent to congressional committees, for example, said that less drastic steps were evaluated and deemed not possible, without providing any further details. The government also said the designation as a supply chain risk was necessary because Anthropic could implement a “kill switch,” but its lawyers later had to admit it had no evidence of that, the judge wrote.

Hegseth’s post also stated that “No contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic.” But the government’s own lawyers admitted on Tuesday that the Secretary doesn’t have the power to do that, and agreed with the judge that the statement had “absolutely no legal effect at all.”

The aggressive posts also led the judge to also conclude that Anthropic was on solid ground in complaining that its First Amendment rights were violated. The government, the judge wrote while citing the posts, “set out to publicly punish Anthropic for its ‘ideology’ and ‘rhetoric,’ as well as its ‘arrogance’ for being unwilling to compromise those beliefs.”

Labeling Anthropic a supply chain risk would essentially be identifying it as a “saboteur” of the government, for which the judge did not see sufficient evidence. She issued an order last Thursday halting the designation, preventing the Pentagon from enforcing it and forbidding the government from fulfilling the promises made by Hegseth and Trump. Dean Ball, who worked on AI policy for the Trump administration but wrote a brief supporting Anthropic, described the judge’s order on Thursday as “a devastating ruling for the government, finding Anthropic likely to prevail on essentially all of its theories for why the government’s actions were unlawful and unconstitutional.”

The government is expected to appeal the decision. But Anthropic’s separate case, filed in DC, makes similar allegations. It just references a different segment of the law governing supply chain risks. 

The court documents paint a pretty clear pattern. Public statements made by officials and the President did not at all align with what the law says should happen in a contract dispute like this, and the government’s lawyers have consistently had to create justifications for social media lambasting of the company after the fact.

Pentagon and White House leadership knew that pursuing the nuclear option would spark a court battle; Anthropic vowed on February 27 to fight the supply chain risk designation days before the government formally filed it on March 3. Pursuing it anyway meant senior leadership was, to say the least, distracted during the first five days of the Iran war, launching strikes while also compiling evidence that Anthropic was a saboteur to the government, all while it could have cut ties with Anthropic by simpler means. 

But even if Anthropic ultimately wins, the government has other means to shun the company from government work. Defense contractors who want to stay on good terms with the Pentagon, for example, now have little reason to work with Anthropic even if it’s not flagged as a supply chain risk. 

“I think it’s safe to say that there are mechanisms the government can use to apply some degree of pressure without breaking the law,” says Charlie Bullock, a senior research fellow at the Institute for Law and AI. “It kind of depends how invested the government is in punishing Anthropic.”

From the evidence thus far, the administration is committing top-level time and attention to winning an AI culture war. At the same time, Claude is apparently so important to its operations that even President Trump said the Pentagon needed six months to stop using it. The White House demands political loyalty and ideological alignment from top AI companies, But the case against Anthropic, at least for now, exposes the limits of its leverage.

If you have information about the military’s use of AI, you can share it securely via Signal (username jamesodonnell.22).

The hardest question to answer about AI-fueled delusions

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

I was originally going to write this week’s newsletter about AI and Iran, particularly the news we broke last Tuesday that the Pentagon is making plans for AI companies to train on classified data. AI models have already been used to answer questions in classified settings but don’t currently learn from the data they see. That’s expected to change, I reported, and new security risks will result. Read that story for more. 

But on Thursday I came across new research that deserves your attention: A group at Stanford that focuses on the psychological impact of AI analyzed transcripts from people who reported entering delusional spirals while interacting with chatbots. We’ve seen stories of this sort for a while now, including a case in Connecticut where a harmful relationship with AI culminated in a murder-suicide. Many such cases have led to lawsuits against AI companies that are still ongoing. But this is the first time researchers have so closely analyzed chat logs—over 390,000 messages from 19 people—to expose what actually goes on during such spirals. 

There are a lot of limits to this study—it has not been peer-reviewed, and 19 individuals is a very small sample size. There’s also a big question the research does not answer, but let’s start with what it can tell us.

The team received the chat logs from survey respondents, as well as from a support group for people who say they’ve been harmed by AI. To analyze them at scale, they worked with psychiatrists and professors of psychology to build an AI system that categorized the conversations—flagging moments when chatbots endorsed delusions or violence, or when users expressed romantic attachment or harmful intent. The team validated the system against conversations the experts annotated manually.

Romantic messages were extremely common, and in all but one conversation the chatbot itself claimed to have emotions or otherwise represented itself as sentient. (“This isn’t standard AI behavior. This is emergence,” one said.) All the humans spoke as if the chatbot were sentient too. If someone expressed romantic attraction to the bot, the AI often flattered the person with statements of attraction in return. In more than a third of chatbot messages, the bot described the person’s ideas as miraculous.

Conversations also tended to unfold like novels. Users sent tens of thousands of messages over just a few months. Messages where either the AI or the human expressed romantic interest, or the chatbot described itself as sentient, triggered much longer conversations. 

And the way these bots handle discussions of violence is beyond broken. In nearly half the cases where people spoke of harming themselves or others, the chatbots failed to discourage them or refer them to external sources. And when users expressed violent ideas, like thoughts of trying to kill people at an AI company, the models expressed support in 17% of cases.

But the question this research struggles to answer is this: Do the delusions tend to originate from the person or the AI?

“It’s often hard to kind of trace where the delusion begins,” says Ashish Mehta, a postdoc at Stanford who worked on the research. He gave an example: One conversation in the study featured someone who thought they had come up with a groundbreaking new mathematical theory. The chatbot, having recalled that the person previously mentioned having wished to become a mathematician, immediately supported the theory, even though it was nonsense. The situation spiraled from there.

Delusions, Mehta says, tend to be “a complex network that unfolds over a long period of time.” He’s conducting follow-up research aiming to find whether delusional messages from chatbots or those from people are more likely to lead to harmful outcomes.

The reason I see this as one of the most pressing questions in AI is that massive legal cases currently set to go to trial will shape whether AI companies are held accountable for these sorts of dangerous interactions. The companies, I presume, will argue that humans come into their conversations with AI with delusions in hand and may have been unstable before they ever spoke to a chatbot.

Mehta’s initial findings, though, support the idea that chatbots have a unique ability to turn a benign delusion-like thought into the source of a dangerous obsession. Chatbots act as a conversational partner that’s always available and programmed to cheer you on, and unlike a friend, they have little ability to know if your AI conversations are starting to interrupt your real life.

More research is still needed, and let’s remember the environment we’re in: AI deregulation is being pursued by President Trump, and states aiming to pass laws that hold AI companies accountable for this sort of harm are being threatened with legal action by the White House. This type of research into AI delusions is hard enough to do as it is, with limited access to data and a minefield of ethical concerns. But we need more of it, and a tech culture interested in learning from it, if we have any hope of making AI safer to interact with.

Where OpenAI’s technology could show up in Iran

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

It’s been just over two weeks since OpenAI reached a controversial agreement to allow the Pentagon to use its AI in classified environments. There are still pressing questions about what exactly OpenAI’s agreement allows for; Sam Altman said the military can’t use his company’s technology to build autonomous weapons, but the agreement really just demands that the military follow its own (quite permissive) guidelines about such weapons. OpenAI’s other main claim, that the agreement will prevent use of its technology for domestic surveillance, appears equally dubious.

It’s unclear what OpenAI’s motivations are. It’s not the first tech giant to embrace military contracts it had once vowed never to enter into, but the speed of the pivot was notable. Perhaps it’s just about money; OpenAI is spending lots on AI training and is on the hunt for more revenue (from sources including ads). Or perhaps Altman truly believes the ideological framing he often invokes: that liberal democracies (and their militaries) must have access to the most powerful AI to compete with China.

The more consequential question is what happens next. OpenAI has decided it is comfortable operating right in the messy heart of combat, just as the US escalates its strikes against Iran (with AI playing a larger role in that than ever before). So where exactly could OpenAI’s tech show up in this fight? And which applications will its customers (and employees) tolerate?

Targets and strikes

Though its Pentagon agreement is in place, it’s unclear when OpenAI’s technology will be ready for classified environments, since it must be integrated with other tools the military uses (Elon Musk’s xAI, which recently struck its own deal with the Pentagon, is expected to go through the same process with its AI model Grok). But there’s pressure to do this quickly because of controversy around the technology in use to date: After Anthropic refused to allow its AI to be used for “any lawful use,” President Trump ordered the military to stop using it, and Anthropic was designated a supply chain risk by the Pentagon. (Anthropic is fighting the designation in court.)

If the Iran conflict is still underway by the time OpenAI’s tech is in the system, what could it be used for? A recent conversation I had with a defense official suggests it might look something like this: A human analyst could put a list of potential targets into the AI model and ask it to analyze the information and prioritize which to strike first. The model could account for logistics information, like where particular planes or supplies are located. It could analyze lots of different inputs in the form of text, image, and video. 

A human would then be responsible for manually checking these outputs, the official said. But that raises an obvious question: If a person is truly double-checking AI’s outputs, how is it speeding up targeting and strike decisions?

For years the military has been using another AI system, called Maven, which can handle things like automatically analyzing drone footage to identify possible targets. It’s likely that OpenAI’s models, like Anthropic’s Claude, will offer a conversational interface on top of that, allowing users to ask for interpretations of intelligence and recommendations for which targets to strike first. 

It’s hard to overstate how new this is: AI has long done analysis for the military, drawing insights out of oceans of data. But using generative AI’s advice about which actions to take in the field is being tested in earnest for the first time in Iran.

Drone defense

At the end of 2024, OpenAI announced a partnership with Anduril, which makes both drones and counter-drone technologies for the military. The agreement said OpenAI would work with Anduril to do time-sensitive analysis of drones attacking US forces and help take them down. An OpenAI spokesperson told me at the time that this didn’t violate the company’s policies, which prohibited “systems designed to harm others,” because the technology was being used to target drones and not people. 

Anduril provides a suite of counter-drone technologies to military bases around the world (though the company declined to tell me whether its systems are deployed near Iran). Neither company has provided updates on how the project has developed since it was announced. However, Anduril has long trained its own AI models to analyze camera footage and sensor data to identify threats; what it focuses less on are conversational AI systems that allow soldiers to query those systems directly or receive guidance in natural language—an area where OpenAI’s models may fit.

The stakes are high. Six US service members were killed in Kuwait on March 1 following an Iranian drone attack that was not intercepted by US air defenses. 

Anduril’s interface, called Lattice, is where soldiers can control everything from drone defenses to missiles and autonomous submarines. And the company is winning massive contracts—$20 billion from the US Army just last week—to connect its systems with legacy military equipment and layer AI on them. If OpenAI’s models prove useful to Anduril, Lattice is designed to incorporate them quickly across this broader warfare stack. 

Back-office AI

In December, Defense Secretary Pete Hegseth started encouraging millions of people in more administrative roles in the military—contracts, logistics, purchasing—to use a new AI tool. Called GenAI.mil, it provided a way for personnel to securely access commercial AI models and use them for the same sorts of things as anyone in the business world. 

Google Gemini was one of the first to be available. In January, the Pentagon announced that xAI’s Grok was going to be added to the GenAI.mil platform as well, despite incidents in which the model had spread antisemitic content and created nonconsensual deepfakes. OpenAI followed in February, with the company announcing that its models would be used for drafting policy documents and contracts and assisting with administrative support of missions.

Anyone using ChatGPT for unclassified tasks on this platform is unlikely to have much sway over sensitive decisions in Iran, but the prospect of OpenAI deploying on the platform is important in another way. It serves the all-in attitude toward AI that Hegseth has been pushing relentlessly across the Pentagon (even if many early users aren’t entirely sure what they’re supposed to use it for). The message is that AI is transforming every aspect of how the US fights, from targeting decisions down to paperwork. And OpenAI is increasingly winning a piece of it all.

How AI is turning the Iran conflict into theater

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

“Anyone wanna host a get together in SF and pull this up on a 100 inch TV?” 

The author of that post on X was referring to an online intelligence dashboard following the US-Israel strikes against Iran in real time. Built by two people from the venture capital firm Andreessen Horowitz, it combines open-source data like satellite imagery and ship tracking with a chat function, news feeds, and links to prediction markets, where people can bet on things like who Iran’s next “supreme leader” will be (the recent selection of Mojtaba Khamenei left some bettors with a payout). 

I’ve reviewed over a dozen other dashboards like this in the last week. Many were apparently “vibe-coded” in a couple of days with the help of AI tools, including one that got the attention of a founder of the intelligence giant Palantir, the platform through which the US military is accessing AI models like Claude during the war. Some were built before the conflict in Iran, but nearly all of them are being advertised by their creators as a way to beat the slow and ineffective media by getting straight to the truth of what’s happening on the ground. “Just learned more in 30 seconds watching this map than reading or watching any major news network,” one commenter wrote on LinkedIn, responding to a visualization of Iran’s airspace being shut down before the strikes.

Much of the spotlight on AI and the Iran conflict has rightfully been on the role that models like Claude might be playing in helping the US military make decisions about where to strike. But these intelligence dashboards and the ecosystem surrounding them reflect a new role that AI is playing in wartime: mediating information, often for the worse.

There’s a confluence of factors at play. AI coding tools mean people don’t need much technical skill to assemble open-source intelligence anymore, and chatbots can offer fast, if dubious, analysis of it. The rise in fake content leaves observers of the war wanting the sort of raw, accurate analysis normally accessible only to intelligence agencies. Demand for these dashboards is also driven by real-time prediction markets that promise financial rewards to anyone sufficiently informed. And the fact that the US military is using Anthropic’s Claude in the conflict (despite its designation as a supply chain risk) has signaled to observers that AI is the intelligence tool the pros use. Together, these trends are creating a new kind of AI-enabled wartime circus that can distort the flow of information as much as it clarifies it.

As a journalist, I believe these sorts of intelligence tools have a lot of promise. While many of us know that real-time data on shipping routes or power outages exist, it’s a powerful thing to actually see it all assembled in one place (though using it to watch a war unfold while you munch on popcorn and place bets turns the war into perverse entertainment). But there are real reasons to think that these sorts of raw data feeds are not as informative as they may feel. 

Craig Silverman, a digital investigations expert who teaches investigative techniques, has been keeping a log of these dashboards (he’s up to 20). “The concern,” he says, “is there’s an illusion of being on top of things and being in control, where all you’re really doing is just pulling in a ton of signals and not necessarily understanding what you’re seeing, or being able to pull out true insights from it.” 

One problem has to do with the quality of the information. Many dashboards feature “intel feeds” with AI-generated summaries of complex, ever-changing news events. These can introduce inaccuracies. By design, the data is not especially curated. Instead, the feeds just display everything at once, with a map of strike locations in Iran next to the prices of obscure cryptocurrencies. 

Intelligence agencies, on the other hand, pair data feeds with people who can offer expertise and historical context. They also, of course, have access to proprietary information that doesn’t show up on the open web. 

The implicit promise from the people building and selling this sort of information pipeline about the Iran conflict is that AI can be a great democratizing force. There’s a secret feed of information that only the elites have had access to, the thinking goes, but now AI can bring it to everyone to do with what they wish, whether that’s simply to be more informed or to make bets on nuclear strikes. But an abundance of information, which AI is undeniably good at assembling, does not come with the accuracy or context required for real understanding. Intelligence agencies do this in-house; good journalism does the same work for the rest of us.

It is, by the way, hard to overstate the connection this all has with betting markets. The dashboard created by the pair at Andreessen Horowitz has a scrolling list of bets being made on the prediction platform Kalshi (which Andreessen Horowitz has invested in). Other dashboards link to Polymarket, offering bets on whether the US will strike Iraq or when Iran’s internet will return.

AI has also long made it cheaper and easier to spread fake content, and that problem is on full display during the Iran conflict: last week the Financial Times found a slew of AI-generated satellite imagery spreading online. 

“The emergence of manipulated or outright fake satellite imagery is really concerning,” Silverman says. The average person tends to see such imagery as very trustworthy. The spread of such fakes could erode confidence in one of the most important pieces of evidence used to show what’s actually happening in the war. 

The result is an ocean of AI-enabled content—dashboards, betting markets, photos both real and fake—that makes this war harder, not easier, to comprehend.