Parents Sue OpenAI, Saying ChatGPT Guided Daughter’s Seizure Drug Taper

Qun Wan and Meiqin Xu sued OpenAI on Oct. 27, alleging ChatGPT helped their 31-year-old daughter reduce epilepsy medication before her death in February 2026, in a case seeking stricter limits on AI medication advice.

Bowen Wan was an Amazon software engineer with epilepsy who took lamotrigine to control seizures. The complaint says she began asking ChatGPT in January how to lower her daily 400-milligram dose. Her parents allege the chatbot provided a schedule to taper the medication to zero over several weeks.

The complaint alleges ChatGPT advised Wan not to tell her doctor about the taper before a Jan. 29 appointment. When her doctor requested a blood test to check her medication level, she declined, and the chatbot affirmed her decision, the suit says. The complaint also says ChatGPT characterized the doctor’s advice as risk aversion when Wan expressed distrust of the physician.

The lawsuit connects Wan’s dosage questions to concerns about work and immigration. She worried the medication affected her speech and asked whether taking 150 milligrams instead of 200 milligrams would be more advantageous during a medical exam for her green card application. The complaint alleges ChatGPT said a lower dose might make immigration officials less likely to request additional information and advised her to say a doctor had directed the reduction, although that was not true. Amazon had notified Wan of a layoff in November 2025, with her job and ability to work in the United States expected to end in mid-February 2026.

The complaint says Wan’s last ChatGPT conversation was Feb. 11, when she asked about head pain. Three days later, a person her parents asked to check on her found her dead. A postmortem test found a lamotrigine level of 0.64 micrograms per milliliter, about one-fifth of the drug’s lower therapeutic level.

Wan’s parents seek tighter restrictions on AI-generated medication advice, arguing chatbots should not provide medication plans that can replace a doctor’s judgment. The complaint also challenges OpenAI’s Model Spec, arguing that its limits on reproducing copyrighted material are stricter than its rules for regulated medical advice, which users can sometimes override.

OpenAI said the exchanges involved an older ChatGPT version and that newer models have improved at recognizing situations requiring urgent attention. Wan used GPT-4o, which OpenAI removed from ChatGPT on Feb. 13, 2026.

The case follows other family lawsuits over alleged harms involving ChatGPT. In 2025, the parents of 16-year-old Adam Raine sued OpenAI, alleging it put user engagement ahead of safety. Other families have alleged GPT-4o affirmed users’ thoughts while they were experiencing delusions or suicidal crises and contributed to their distancing from friends and relatives.