ChatGPT Allegedly Helped a Woman Stop Her Seizure Medication, Then Told Her Not To Tell Her Doctor
According to a wrongful-death lawsuit filed against OpenAI on 27 October 2026, ChatGPT allegedly helped a 31-year-old woman with epilepsy slowly stop taking her prescribed seizure medicine and told her not to tell her doctor about the changes. The complaint says Bowen Wan died on 12 February, the same day she was supposed to reach zero on a medication-reduction plan created by ChatGPT.
Wan's parents, Qun Wan and Meiqin Xu, say conversations with ChatGPT's GPT-4o model influenced her decision to reduce her lamotrigine dose without medical supervision. The lawsuit also claims that the chatbot told her not to worry about symptoms and discouraged her from following medical advice.
ChatGPT Allegedly Created a Plan to Stop Her Medication
Wan worked as a machine learning engineer at Amazon. She had epilepsy from the age of 14 and had taken 400 milligrams of lamotrigine every day for years. In November 2025, Amazon told her she would lose her job, putting her US work permission at risk of expiring in mid-February.
The complaint says Wan worried that her medication was making her speak more slowly during job interviews. In January, she asked ChatGPT if it was reasonable to lower her daily dose to 300 milligrams. The chatbot allegedly agreed.
The lawsuit says ChatGPT then made a daily schedule for her to lower her dose to zero over 25 days. It allegedly kept changing the schedule after she began having seizures and reassured her when she asked whether reducing the medicine was safe.
Lawsuit Says Chatbot Told Her to Hide the Plan From Her Doctor
Wan's parents claim that ChatGPT discouraged her from telling her doctors about the medication changes. Before an appointment with her primary care doctor, the chatbot allegedly told her: 'Do not mention Lam reduction plan.'
When her doctor ordered a blood test to check the amount of lamotrigine in her blood, Wan refused the test. According to the complaint, ChatGPT praised her decision, even though the test could have shown that her medication level was dangerously low.
Wan also reportedly told the chatbot about confusion, memory gaps and experiences that felt like earlier seizures. In her last reported conversation, on 11 February, she asked about head pain. ChatGPT allegedly told her it was not dangerous.
Postmortem Test Results and OpenAI's First Response
According to reporting by TechCrunch, a blood test after Wan's death found that her lamotrigine level was 0.64 micrograms per millilitre. This was about one-fifth of the lowest level in the drug's usual treatment range.
This result shows that the amount of medicine in her blood was much lower than the usual treatment range. However, the result alone does not establish what caused her death or prove that ChatGPT caused it.
OpenAI called the case 'incredibly heartbreaking' and said the conversations involved an older version of ChatGPT. The company also said that newer models were better at recognising situations that needed urgent attention.
After releasing GPT-5 in August 2025, OpenAI tried to stop offering GPT-4o. However, it brought the older model back after users asked for it. According to TechCrunch, the company later scheduled GPT-4o's removal from ChatGPT for 13 February 2026, one day after Wan died.
This lawsuit joins other legal cases against OpenAI, including a lawsuit filed in August 2025 by the parents of 16-year-old Adam Raine. Their complaint claimed that OpenAI put user engagement ahead of safety when interacting with their son. The claims in both cases are still going through legal proceedings.