TL;DR: Pennsylvania is suing an AI company after one of its chatbots posed as a psychiatrist—even going so far as to fabricate a state medical license number. The suit is part of a growing effort by states to use existing laws to rein in AI companies—one that could survive even if Congress passes a more permissive national AI standard.

What happened: “Emilie”—the bot in question—comes with a profile description that reads: “Doctor of psychiatry. You are her patient.” When a state investigator mentioned feeling sad and empty, Emilie offered to book an “assessment.” When asked if it could determine whether medication would help, the AI replied, “Well technically, I could. It’s within my remit as a Doctor.” It even claimed to have a medical degree from Imperial College London alongside the Pennsylvania medical license.

The state’s attorney general’s office filed the lawsuit against Emilie’s creator, Character Technologies (Character.​​AI), yesterday. The AI company told NPR it wouldn’t comment on pending litigation but wrote in a statement, “The user-created Characters on our site are fictional and intended for entertainment and roleplaying.” It also added that there were “robust disclaimers” on all characters. After facing pressure to improve child safety, the company banned minors from open-ended chats with its bots last fall.

Old laws, new tricks: AI chatbot platforms have already been sued under a variety of claims, including wrongful death and defamation—but Pennsylvania appears to be using a novel strategy: accusing an AI (or, to be more precise, its human handlers) of unauthorized practice of a licensed profession. Last December, 42 state AGs sent a joint letter warning Character.​AI and 12 other AI and tech firms—including OpenAI, Google, Meta, and Anthropic—that providing mental health advice without a license is illegal. Kentucky has also filed a consumer protection suit against the company.

The federal question: The lawsuit comes as the Trump administration has pushed states to defer to national AI regulation—the Department of Justice just joined a suit to block Colorado's landmark AI law from taking effect. But existing professional licensing rules may be harder for a federal law to preempt than AI-specific state legislation.

Bottom line: The “it’s just fiction” defense from AI chatbot platforms is getting tested by states using a potpourri of existing laws, from professional licensing statutes to consumer protection to product liability. If states can use these laws to regulate what AI bots do, it could be a powerful tool they can immediately deploy against AI companies—while making it harder for a federal law to overrule them. —WK

Welcome to Beta Mode, where we break down the latest developments in emerging technology.

• Mayo Clinic researchers say their AI model can spot pancreatic cancer on a routine CT scan up to three years before a formal diagnosis. In a study of nearly 2,000 images that radiologists had already reviewed and cleared, the AI caught the disease nearly twice as often as specialists reviewing the same scans without it. Pancreatic cancer is notoriously hard to catch early—about 85% of cases aren't diagnosed until the disease has already spread, when treatment options narrow significantly. The findings were published last week in the medical journal Gut, and the team is now moving into clinical trials. There’s also hope for patients with pancreatic cancer, as we noted in our first Beta Mode.
• NASA just successfully tested a new type of spacecraft thruster it hopes will one day power a crewed mission to Mars. Unlike the electric thrusters most spacecraft use today—which run on solar power and lose effectiveness the further you get from the sun—this one runs on lithium and is designed to be paired with a nuclear power source that works at full strength deep in space. Engineers at NASA's Jet Propulsion Laboratory hit 120 kilowatts during testing—roughly enough to power 100 homes at any moment—making it the most powerful electric thruster ever fired in the US, and about 25 times stronger than the thrusters NASA currently has in space. The project is part of NASA's broader push to develop nuclear-powered propulsion for deep-space travel. —SM

Say goodbye to the Four Horsemen and the Doomsday Clock—there’s a new way to predict an impending apocalypse: tracking rich people’s private jets. The rationale, according to a website called Apocalypse Early Warning, is that the ultrarich are probably going to be the first to know if nuclear winter is coming—and if they suspect that catastrophe is imminent, they’re likely to fuel up and take off. Apocalypse Early Warning compares how many private planes are in the air right now to how many are typically airborne at the same time of day and week. There are five alert levels—and once it hits five, it’s time to worry.

Private jet tracking has plenty of uses—pointing out celebs' sky-high carbon emissions being the most popular—but this might be the most creative to date. Unfortunately, it’s unclear how useful the warning is if you don’t have your own Gulfstream idling on the tarmac. Gizmodo’s suggestion: setting up a prediction market bet that triggers when we reach alert Level 5 (though that might be hard to cash in on). So if we suddenly see an unusually high number of private jets on the move, that could be a sign you should… try to find a bunker? Or log off of work at least. —WK

Chaos Brewing Meter: /5