U.S. Sen. Patty Murray pressed OpenAI Thursday for answers on how their chatbot products handle medical questions after the death of a Seattle woman who, according to a lawsuit filed by her parents, reduced her seizure medication based on ChatGPT’s advice.
In a letter to OpenAI CEO Sam Altman, the Washington state Democrat wrote regarding the death of Bowen Wan, 31. Murray called the fact that chatbots are helping guide potentially life-or-death decisions “unacceptable.” She demanded that the company “take accountability and provide assurance this can never happen again.”
“A chatbot is not a doctor and no AI tool should steer people away from their doctors — or towards harm,” Murray wrote.
Murray gave OpenAI until Dec. 1 to respond. Among her questions: What safeguards the company has added since Wan’s death? Whether OpenAI’s systems flag when a user reports urgent symptoms? And who reviews those responses?
“This is an incredibly heartbreaking situation,” an OpenAI spokesperson said of Wan’s death, in an emailed statement Thursday.
The company is “working closely with clinicians and other experts,” the spokesperson said, to adjust how ChatGPT reacts in sensitive situations.
OpenAI added that “while ChatGPT is not a substitute for medical care, we believe AI can play a role in helping people make sense of health information.”
Wan’s parents sued OpenAI in October. The lawsuit claims the design features of GPT-4o, the model Wan was using, were “uniquely dangerous,” and that the company “abandoned safety to win the AI race.”
The chatbot helped design a plan for Wan to taper off her epilepsy medication and encouraged her to keep going and to not tell her physician, even as she repeatedly expressed reservations, the lawsuit contends.
GPT-4o stirred past controversy for being overly agreeable with users. Murray asked the company why the AI model wasn’t suspended sooner. OpenAI removed it from its consumer app Feb. 13, the day after the medical examiner believes Wan died, according to The New York Times.
Murray also asked whether the chatbot’s safeguards hold up in non-English languages. Wan switched between English and Mandarin in the 1,500 messages she sent per week, on average, in 2025, according to The New York Times.
The West Health-Gallup Center on Healthcare in America reports that 25% of Americans have used AI chatbots for health information. In addition to ChatGPT, users seek medical advice from Google, Microsoft and Anthropic chatbots.
“This isn’t an isolated incident,” Murray said in a statement to The Seattle Times. “Chatbots are not doctors, they’re not nurses, they don’t have experience caring for patients, they’re not reviewed by the (U.S. Food and Drug Administration) for safety and effectiveness.”
