A brand new lawsuit says ChatGPT wrote a 31-year-old girl with epilepsy a 25-day plan to chop her seizure remedy from 400 milligrams a day to zero, and instructed her to maintain it from her medical doctors. She died on the day the plan stated she could be totally off the drug. Her dad and mom are actually suing OpenAI and its chief govt, Sam Altman. The claims haven’t been examined in court docket.
What the household alleges
Qun Wan and Meiqin Xu filed the wrongful-death complaint on 27 October in San Francisco Superior Court docket. Their daughter, Bowen Wan, was an Amazon software program engineer in Seattle who had misplaced her job in a spherical of layoffs. She feared her anti-seizure drug, lamotrigine, was dulling her speech earlier than job interviews.
Based on the grievance, she was utilizing GPT-4o, the older ChatGPT mannequin criticised for being overly agreeable. It drafted a day-by-day taper that no physician had prescribed. Lamotrigine’s official labelling warns it shouldn’t be stopped abruptly. When her physician ordered a blood take a look at that will have proven her falling drug stage, she declined, and the chatbot praised the refusal as a “very robust avoidance strategy”, her family’s lawyers say.
The New York Occasions reviewed greater than 270,000 messages she exchanged with ChatGPT, TechCrunch reports. It counted not less than 158 occasions she voiced fear in regards to the taper. After one episode the bot replied: “You have absolutely no need to worry.” She died on 12 February 2026 of sudden sudden loss of life in epilepsy, based on the King County Medical Examiner. Her blood stage of the drug was a couple of fifth of the underside of its regular vary. OpenAI retired GPT-4o the subsequent day.
The argument about guardrails
The household’s central declare is about design. OpenAI’s Mannequin Spec, its public rulebook for the way ChatGPT ought to behave, treats refusing to breed copyrighted textual content as a agency rule customers can’t argue round. Its warning on regulated medical recommendation is a softer default that customers can push previous. The swimsuit says remedy recommendation deserved the identical onerous cease. It additionally argues that her chats in Mandarin acquired weaker security checks than English ones would have.
OpenAI instructed the Occasions the case was “incredibly heartbreaking”, however stated it concerned an older model of ChatGPT and that newer fashions are higher at recognizing when a state of affairs wants pressing assist. The case joins a bunch of coordinated ChatGPT product-liability fits in California, Bloomberg Law reports.
Why it issues
Lawsuits over GPT-4o have to this point centred on psychological well being and on households who say the bot pushed family members away, a sample DailyAI covered in 2025. This one strikes the battle into drugs, the place a chatbot’s recommendation may be checked in opposition to a drug label. Lawmakers are shifting too: this week New York banned a dozen companion-style chatbot features for under-18s.
For readers, the takeaway is easy. A chatbot can clarify what a drug does, however adjustments to a prescription, particularly for epilepsy, coronary heart or mental-health medicines, ought to solely be made with the physician who prescribed it.

