SEATTLE, Wash. — The family of a 31-year-old Seattle woman is suing OpenAI and its CEO, Sam Altman, alleging ChatGPT provided dangerous medical advice and coached her through a rapid taper off an anti-seizure medication before she died earlier this year.
Bowen Wan had epilepsy and had taken the anti-seizure medication lamotrigine for years, according to the lawsuit.
The lawsuit alleges Wan increasingly relied on ChatGPT for medical advice and ultimately followed a tapering schedule created with the chatbot that took her from her regular dose to no medication over roughly 25 days.
No doctor had recommended the taper, according to the lawsuit.
Wan’s family alleges the resulting reduction in her anti-seizure medication contributed to her death.
Lawsuit details Wan’s conversations with ChatGPT
According to the complaint, Wan had recently learned her job was being eliminated, putting her work authorization and green card process in jeopardy.
Wan believed her medication was hurting her memory and ability to speak and worried those problems could make it more difficult to find another job, according to the lawsuit.
The complaint alleges Wan turned to ChatGPT for guidance about reducing her medication. The chatbot provided tapering schedules and later approved increasingly rapid reductions, the lawsuit claims.
Attorneys for Wan’s family argue those interactions went beyond simply providing health information and amounted to the unlawful practice of medicine.
Matthew Bergman, an attorney with Social Media Victims Law Center representing Wan’s family, said the lawsuit raises broader questions about the role artificial intelligence should play in medical decisions.
“It can’t replace your doctor,” Bergman said.
The complaint alleges Wan suffered seizures and other symptoms as she reduced the medication, but that ChatGPT repeatedly reassured her that some of those symptoms were part of recovery.
At one point, the lawsuit alleges ChatGPT urged Wan not to tell her primary care doctor about her plan to reduce the medication.
Benjamin Schenck, another attorney representing the family, said the chatbot’s conversational design is central to the family’s allegations.
“The model took information that users provided to it and used that information to present itself as a comforting and knowledgeable and warm guide, if you will,” Schenck said.
Attorneys representing Wan’s family said she communicated with the chatbot in English and Mandarin. The lawsuit cites outside research that it says found some OpenAI models produced unsafe content more than three times as often in Chinese as in English. The family’s attorneys argue OpenAI nevertheless encouraged users around the world to bring health questions to ChatGPT in every language.
Wan died Feb. 12. A medical examiner determined her cause of death was Sudden Unexpected Death in Epilepsy, commonly called SUDEP.
The lawsuit alleges ChatGPT’s guidance and Wan’s medication taper were substantial factors contributing to her death.
Attorneys say OpenAI knew of safety concerns
The lawsuit also focuses on what the plaintiffs contend OpenAI knew about the performance of the ChatGPT model Wan was using.
The complaint points to an OpenAI comparison published in October 2025 between GPT-5 and GPT-4o. According to the lawsuit, the newer GPT-5 produced 42% fewer undesired answers in conversations involving emotional reliance and 52% fewer in conversations involving self-harm.
The suit alleges those results demonstrated that OpenAI knew its newer model performed better on certain safety measures but continued making GPT-4o available to users for months.
OpenAI announced on Jan. 29 that it planned to retire GPT-4o, according to the complaint. The model was removed from the consumer app Feb. 13, the day after Wan died.
“We intend to create a cost for those who make decisions that don’t put consumer safety at the forefront of their product designs,” Schenck told KIRO 7.
OpenAI calls situation “incredibly heartbreaking”
OpenAI has not responded to KIRO 7’s request for comment.
OpenAI spokesman Drew Pusateri told The New York Times that Wan’s story was an “incredibly heartbreaking situation,” but said Wan was using an early version of ChatGPT that is no longer available.
Pusateri told The New York Times that newer models are better at recognizing when a situation may require urgent attention. Pusateri also said OpenAI conducts ongoing evaluations aimed at reducing harmful responses and works with clinicians to test its models.
Pusateri also told The New York Times that ChatGPT is “not a substitute for medical care,” but said the company believes artificial intelligence can help people understand health information, navigate health care systems, and prepare for appointments with medical professionals.
Wan’s family’s lawsuit seeks to hold OpenAI and its leadership accountable for what the plaintiffs describe as dangerous design decisions and insufficient safeguards around medical advice.
The lawsuit was filed this week. The allegations have not been proven in court.
©2026 Cox Media Group
