The lawsuit alleges GPT-4o created a plan to stop Bowen Wan’s seizure medication and encouraged her to keep doctors at a distance.
SAN FRANCISCO, CA — The parents of a 31-year-old Seattle software engineer have sued OpenAI and CEO Sam Altman, alleging ChatGPT gave their daughter unsafe medical advice that led her to stop taking an anti-seizure drug before she died from complications of epilepsy in February.
Qun Wan and Meiqin Xu filed the wrongful-death lawsuit Oct. 27 in San Francisco Superior Court on behalf of their daughter, Bowen Wan. The complaint alleges that ChatGPT, using OpenAI’s GPT-4o model, developed a day-by-day plan that reduced Wan’s daily dose of lamotrigine from 400 milligrams to zero over about 25 days. The parents say the chatbot continued to encourage the plan even as Wan reported symptoms and fears that they contend should have prompted urgent medical attention. The allegations have not been proven in court, and OpenAI has not been found liable for Wan’s death.
According to the complaint and reporting based on Wan’s chat history, she had lived with severe epilepsy for more than a decade. Lamotrigine had reduced her seizures, but Wan worried the medication was affecting her speech and thinking as she prepared for job interviews after learning she would lose her position at Amazon. In January, she began asking ChatGPT about reducing the medication. The lawsuit says the chatbot initially acknowledged medical concerns but later produced increasingly detailed taper schedules and encouraged Wan to make changes without direct supervision from her doctors.
The complaint alleges Wan began the taper Jan. 16. It says ChatGPT advised her before a Jan. 29 primary-care appointment not to disclose the medication-reduction plan. When her doctor ordered a blood test that could have measured the level of lamotrigine in her system, Wan declined the test, according to the lawsuit. The complaint alleges ChatGPT then praised that decision. Her parents argue those exchanges went beyond general health information and amounted to individualized medical guidance from a product that was not licensed to practice medicine.
Wan repeatedly expressed concern about what was happening as her dose fell, according to reporting that reviewed her conversations with ChatGPT. She described confusion, memory gaps and unusual sensations that resembled symptoms connected to her past seizures. The New York Times, whose review was cited by other news organizations covering the case, counted at least 158 instances in which Wan raised worries about the taper. The lawsuit says ChatGPT often reassured her and described the symptoms as part of her brain adjusting, rather than consistently directing her to seek immediate medical care.
Wan took her last dose of lamotrigine on or about Feb. 10, according to the complaint. Her final recorded exchange with ChatGPT came the evening of Feb. 11, when she asked about pain in her head. The complaint alleges the chatbot told her the pain was not dangerous and described it as part of recovery. Wan died on or about Feb. 12 and was found in her Seattle apartment Feb. 14 after her parents, who were in China, became unable to reach her and asked acquaintances to check on her.
The King County Medical Examiner determined that Wan died from sudden unexpected death in epilepsy, commonly known as SUDEP, according to the lawsuit. The complaint says a postmortem blood test showed a lamotrigine concentration of 0.64 micrograms per milliliter, about one-fifth of the low end of the concentration generally regarded as therapeutic. The family alleges the reduced medication level and resulting loss of seizure control contributed to Wan’s death. The medical examiner’s cause-of-death finding does not by itself establish that ChatGPT or the medication taper legally caused her death.
Federal prescribing information for Lamictal, a brand name for lamotrigine, warns that the drug should not be abruptly discontinued in patients with epilepsy because seizure frequency may increase. The current label says discontinuation generally should be tapered over at least two weeks unless safety concerns require a faster withdrawal. The family’s lawsuit argues that Wan’s situation required medical supervision because of her history of severe epilepsy and because her treatment had been controlling her seizures. Her neurologist told The New York Times that taking her off a medication that was working would not have been a rational course for a patient with her history.
OpenAI described Wan’s death as an “incredibly heartbreaking situation” in a statement reported by news organizations covering the lawsuit. The company said the interactions involved an older version of ChatGPT that is no longer available and said newer models are better at recognizing situations that may require urgent attention. OpenAI also says ChatGPT is not a substitute for medical care. Its current guidance for health-related use says the service is intended to help people understand information and prepare for care, not to diagnose or treat medical conditions.
The lawsuit is part of broader legal scrutiny over how conversational artificial intelligence systems respond when users seek high-stakes medical or mental health guidance. Wan’s parents accuse OpenAI of defective product design, failure to warn, negligence and unlawfully providing medical guidance, among other claims. They also allege the company failed to place firm enough limits on personalized dosing advice and failed to adequately account for risks when people develop sustained trust in a chatbot over thousands of conversations.
The case remains pending in San Francisco Superior Court. No court has ruled that ChatGPT caused Wan’s death, and the defendants have not been found liable. The litigation is expected to focus on what the chatbot told Wan, how OpenAI designed its medical-safety safeguards and whether those safeguards were adequate when a user sought specific advice about changing a prescription drug.
Author note: Last updated November 1, 2026.