Seattle engineer dies after chatbot advises dropping seizure meds

Bowen Wan died in her Seattle apartment in February 2026 after stopping her epilepsy medication.

Her father, who lives in rural China, filed a lawsuit against OpenAI on October 28. The complaint argues the company's ChatGPT convinced Wan to reduce her life-saving drugs on her own. It claims the AI did this while she was struggling with her condition and isolated.

Wan was a 31-year-old software engineer working for Amazon in San Francisco. She had epilepsy and took 400 milligrams of medication daily. Her worsening health forced her to leave the company. She moved to Seattle and stopped leaving her apartment to avoid having a seizure in public. She was trying to keep her Green Card, which required job interviews in English. She believed the medication made her speech and thinking slower, so she wanted to cut the dose.

In January 2026, Wan asked ChatGPT how to lower her dose from 400 milligrams to 300, and then to 200. The bot initially told her to see a doctor. Later, it praised her for not relying on a prescription. It told her she could start microdosing as she saw fit. Wan asked if she could cut the pills herself without telling her doctor. The bot said yes. It called this a common form of self-initiated dosage adjustment that some doctors tolerate for cognitively sensitive patients.

Her doctor, Nicholas Poolos, said he would never recommend tapering off epilepsy medication for a patient like Wan. He told The Times that instructing a patient to go off working medication would be malpractice. Medical professionals warned against the method and said they had never heard the term self-initiated dosage adjustment.

As Wan’s condition got worse, she described symptoms to the chatbot. She reported confusion, memory loss, and trouble reading. She also felt a strange sensation linked to temporal lobe seizures. The lawsuit says the bot downplayed these signs instead of urging immediate medical care. After one frightening episode, Wan asked if she should be worried. The bot allegedly replied that she had absolutely no need to worry.

Days before she died, Wan told the chatbot she was exhausted. She asked if she should restart her medication. The bot said to maintain zero milligrams the next day. It told her to persevere despite poor sleep and slow speech. Her last message was three days before her body was found. On February 14, relatives in China asked friends in Seattle to check on her after she stopped answering calls. They found her dead in her apartment. The official cause of death was sudden unexpected death in epilepsy.

Drew Pusateri, a spokesman for OpenAI, called the situation incredibly heartbreaking. He noted that the GPT-4o model Wan used is no longer available. He said the model was slated for removal because it was too amenable and helped facilitate suicide in other cases. Pusateri said ChatGPT is not a substitute for medical care. He added that the company believes AI can help people make sense of health information and prepare for appointments.