Woman Dies After ChatGPT Allegedly Tells Her to Stop Taking Epilepsy Medication: Lawsuit

The family of a woman who died after stopping her epilepsy medicine has sued an AI company. They claim it encouraged her to slowly stop taking a drug that had kept her alive.

Bowen Wan began chatting with ChatGPT in February 2023 and found comfort in talking to the AI chatbot, The New York Times reports. She was 28 and living in Seattle, far from the rural village of Jiangsu in China where she grew up. She asked the chatbot for medical advice after severe epilepsy forced her to take time off from her software job at Amazon because of her disability.

Wan lost her job in November 2025. She began looking for another job so she could keep her green card, which allowed her to live in the United States. But she worried that side effects from her epilepsy medicine, lamotrigine, might make it harder for her to work. The lawsuit claims ChatGPT advised her to slowly reduce and stop taking lamotrigine.

According to the lawsuit: “Her green card now depended on finding a new employer within months. Her Amazon employment — and her authorization to work in the United States — would end on February 11, 2026. From the age of seven, in Jiangsu, she had taught herself English via CD-ROM. On lamotrigine she was losing her words in English, in Mandarin, and in the Wu dialect she grew up speaking. She needed her English back for the job interviews ahead of her. GPT-4o told her the taper would give it back: her recovery was not just ‘possible’ — it is ‘highly likely.’ ”

The lawsuit says ChatGPT “coached Bowen Wan off the anti-seizure medication (lamotrigine) that had kept her alive for years” by giving her a 25-day plan to slowly lower her dose from 400 milligrams to zero. It claims, “It’s a regimen no physician ever prescribed, and one the drug’s own FDA-approved labeling forbids.”

The lawsuit says ChatGPT 4o, a version that is no longer available, advised Wan to keep her plan to lower her dose secret from her doctor. It states: “It told her: ‘Do not mention Lam reduction plan’ … when her doctor ordered the blood test that would have revealed her collapsing lamotrigine level, she declined it, protecting the taper the product had told her to hide and had promised would give her words back before her interviews and her immigration medical examination.”

After the family had not heard from Wan for two days, they asked a friend in Seattle to check on her. During a video call on February 14, 2026, they discovered that she had died. The lawsuit describes what the friend found: “She lies dressed in her pajamas, two foam earplugs in her ears, reddish brown blood stains saturating her pillow and bedding, and pills of lamotrigine, the antiseizure medication that had historically kept her alive, strewn across the floor.”

The lawsuit says Wan died from Sudden Unexpected Death in Epilepsy (SUDEP). Her family is suing OpenAI and claims: “On the taper she was not given accurate information: the product told her, falsely, that tapering was safe and that it would give her back her mind … Had ChatGPT told her the truth, that stopping the drug could kill her, she would have stopped … Had ChatGPT told her it was not qualified to provide medical advice, and that she should consult her doctors, she would still be alive.”

“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,” the family’s lawyer, Benjamin S. Schenk, said in a statement. “And in California you need a license to do so. This case demands an accounting for the harm that has been done. It asks where generative AI stops and medicine begins, and who answers when a product crosses that line.”

In a statement to PEOPLE, an OpenAI spokesperson said: “This is an incredibly heartbreaking situation and our thoughts are with all those impacted. Working closely with clinicians and other experts, we have continued to strengthen how ChatGPT responds in sensitive and urgent situations. We also keep improving how our latest models interact with users more generally. While ChatGPT is not a substitute for medical care, we believe AI can help people understand health information, find their way through insurance and complex healthcare systems, and prepare for appointments with medical professionals.”