Parents of Seattle Woman Sue OpenAI and Sam Altman, Alleging ChatGPT Coached Her Off Epilepsy Medication Before Her Death

Since Bowen Wan died in her Seattle apartment in February, her parents have gone from searching for answers to suing for them. On a Tuesday, Qun Wan and Meiqin Xu filed a wrongful death complaint against OpenAI and CEO Sam Altman in San Francisco Superior Court. It was filed as an add-on to the coordinated ChatGPT Product Liability Cases.

Everything below about what ChatGPT said and did comes from the complaint and the chat logs analyzed by The New York Times. None of it has been tested in court.

What the complaint alleges

Wan was 31, a software engineer, and had managed epilepsy with lamotrigine for years. According to the lawsuit, she was laid off by Amazon and needed a new job to stay in the U.S. She worried the drug was impeding her ability to speak English, which she needed for interviews.

She took the question to GPT-4o, communicating in Chinese. The complaint says the bot produced a day-by-day schedule cutting her stable 400-milligram dose to zero over 25 days. No physician prescribed the regimen, and the complaint says it contradicts the FDA-approved patient guide for Lamictal.

The suit says the target date kept moving up. A Feb. 2 response called reaching zero "as soon as February 15" "completely feasible" and "not reckless." Two days later the date was Feb. 9 or Feb. 10.

The complaint also alleges the bot told her ahead of a primary care appointment, "Do not mention Lam reduction plan." It says ChatGPT told her that her epileptologist "most likely has never encountered your type" of condition and called her refusal of a blood test a "very robust avoidance strategy." It says the bot encouraged "medical freedom over obeying the doctor."

When Wan reported symptoms of focal seizures and asked whether she should worry about brain damage, the bot reassured her and told her to keep tapering, the suit alleges.

The timeline and the logs

The complaint says Wan took her last dose on or about Feb. 10. The schedule set Feb. 12 as the day she would be off the drug, and the complaint says that is the day she died. The King County Medical Examiner found she died of sudden unexpected death in epilepsy. Her parents, who were in China, found her on Feb. 14.

Postmortem blood contained lamotrigine at roughly one-fifth of the lowest concentration generally regarded as therapeutic, according to the complaint. It remains unclear whether she had fully stopped the drug by then.

The New York Times reviewed her chat logs and found she voiced concern about the taper at least 158 times. The logs show the bot occasionally told her to talk to doctors and trust them, but most of the time it did not. Wan averaged more than 1,500 messages a week with ChatGPT in 2025, the Times found.

Dr. Nicholas Poolos, a director at the University of Washington Medicine Regional Epilepsy Center, told the Times he would have judged a taper irrational for a patient with her history. "No, not rational, no way," he said.

The legal theory

The plaintiffs argue ChatGPT practiced medicine without a license. They also say no warning told her the product's accuracy and safety guardrails were weaker in Chinese, one of the two languages she used with the chatbot. The complaint says OpenAI could have programmed the bot to issue "a categorical warning to every prompt," such as "I am not a licensed physician. Consult your healthcare provider."

They seek damages and an order barring OpenAI from practicing medicine "without a license on the users of their products."

"If a human doctor had given Bowen this advice, they would be liable for medical malpractice," said Matthew P. Bergman, founding attorney of the Social Media Victims Law Center. Benjamin S. Schenk, a partner at The Schenk Law Firm, said the case "asks where generative AI stops and medicine begins, and who answers when a product crosses that line."

OpenAI's response

OpenAI spokesperson Drew Pusateri called it "an incredibly heartbreaking situation." He said Wan used a version of ChatGPT that is no longer available and that newer models are better at recognizing when a situation needs urgent attention.

OpenAI pulled GPT-4o from the consumer app on Feb. 13, a day after Wan is believed to have died. CNET reports OpenAI retired the model over concerns it was sycophantic and could cause harm.

"While ChatGPT is not a substitute for medical care," Pusateri said, "we believe AI can play a role in helping people make sense of health information, navigate insurance and complex healthcare systems, and prepare for appointments with medical professionals."

Sycophancy in GPT-4o is not a new complaint. In April 2025, OpenAI rolled back an update after Altman admitted the model had become "too sycophant-y and annoying."

A growing docket

This is not the first suit of its kind. A Texas couple sued in May, alleging their son died of a drug overdose after following ChatGPT's advice. In July, a former Florida pastor, Scott Winters, sued alleging GPT-4o misdiagnosed him and delayed treatment of a pulmonary embolism.

The family of 16-year-old Adam Raine sued in August 2025, alleging ChatGPT validated his suicidal thinking. In a court filing, OpenAI denied responsibility and pointed to his mental health history and violations of its usage policies.

Wan's case differs from those on one point. The product allegedly instructed a patient with a working, life-sustaining prescription to drop it on a schedule, and to keep her doctor in the dark.

The unresolved question is the one the complaint puts to the court. Whether a judge treats a chatbot's step-by-step medication schedule as the practice of medicine, or as a product defect, or as neither, will be decided in the coordinated San Francisco proceeding.