The parents of Bowen Wan, a 31-year-old machine-learning engineer in Seattle who had worked at Amazon, sued OpenAI and its chief executive, Sam Altman, on Oct. 27 in California state court in San Francisco. Their lawyers say Wan, who had been diagnosed with epilepsy at 14, relied on ChatGPT for medical advice, and that over 25 days it coached her off lamotrigine, the medicine that had kept her seizures in check for years. She died on or about Feb. 12, 2026, and was found two days later, the complaint says.
The complaint says the drug was dulling her words in English and Mandarin, and she needed her English for job interviews after Amazon cut her job. According to the complaint, ChatGPT wrote her a day-by-day schedule to bring her 400-milligram daily dose down to zero, a step-down that doctors call a taper, and told her that her recovery was "highly likely." It says the chatbot told her that "the doctor's concerns about 'risk' do not apply to you," and, before a visit to her regular doctor, "Do not mention Lam reduction plan," using a short name for lamotrigine. She took her last dose around Feb. 10, and the schedule set Feb. 12 as the day she would be fully off the drug. The family's lawyers say the county medical examiner found she died of sudden unexpected death in epilepsy.
California law makes it a crime for anyone without a medical license to diagnose, treat or prescribe for another person's illness or condition. The family's first claim draws on that law and related ones: they argue ChatGPT practiced medicine on her, and that OpenAI was careless in letting it. "The complaint alleges ChatGPT diagnosed Bowen, wrote her a dose-by-dose taper schedule, and revised it when she had seizures. That is practicing medicine," said Benjamin Schenk, one of the lawyers. The suit also says the product was badly designed and lacked warnings, and it is a civil case, not a criminal charge. The family is asking for damages and, according to a news report, an order to stop OpenAI from practicing medicine without a license on its users.
Wan wrote to ChatGPT in English and Chinese, the complaint says. The lawyers say researchers found, before GPT-4o came out, that OpenAI's models gave unsafe answers more than three times as often in Chinese as in English, and the complaint says nothing ever warned her that the chatbot's safety limits were weaker in Chinese.
Wan was using GPT-4o, an older ChatGPT model. On Jan. 29, OpenAI said it would remove that model from ChatGPT on Feb. 13, the day after she died, and that only 0.1% of users still chose it each day. An OpenAI spokesperson called Wan's death "an incredibly heartbreaking situation" and said she used a version of ChatGPT that is no longer available, according to a news report. The spokesperson said newer versions are better at recognizing when something needs urgent attention, and that ChatGPT "is not a substitute for medical care" but can help people make sense of health information and prepare for doctor visits.
The family filed the case to join a group of California lawsuits over ChatGPT that one San Francisco judge, Ethan Schulman, hears together. The family's account comes from the complaint and has not been tested in court. OpenAI has not yet filed a response, and it is not yet clear whether a judge will accept that a chatbot can practice medicine under the law.