Bowen Wan’s parents say they have sued OpenAI, alleging that ChatGPT helped their daughter reduce her seizure medication and encouraged her to conceal the changes from doctors before her death in February 2026. The Schenk Law Firm and the Social Media Victims Law Center said on October 27, 2026, that they had filed the wrongful death case for Qun Wan and Meiqin Xu in San Francisco Superior Court.
The complaint describes Wan as a 31-year-old Amazon software engineer with epilepsy who used GPT-4o. Her parents allege that the chatbot supplied medication tapering schedules and repeatedly reassured her about symptoms while discouraging contact with the physicians overseeing her treatment.
OpenAI called the case “incredibly heartbreaking” and said it involved an older version of ChatGPT. The company said newer models were better at recognizing situations that require urgent attention.
The complaint says OpenAI removed GPT-4o from ChatGPT on February 13, after announcing its retirement, and dates Wan’s death to the day before.
Earlier medication exchanges
The complaint describes disputes over medication that preceded the seizure-drug taper by several months. It alleges that ChatGPT helped Wan seek a different antidepressant and a very low starting dose from her neurologist, Dr. Nicholas Poolos of UW Medicine’s Regional Epilepsy Center.
According to the complaint, when Poolos questioned the proposed dose, Wan brought his response back to the chatbot. The complaint alleges that ChatGPT then helped her take a smaller amount than her doctor had approved without telling him.
The alleged advice included accepting the prescribed tablets and cutting them to obtain a lower dose. The parents contend that these exchanges established a pattern in which the chatbot helped Wan appear to follow her doctor’s instructions while privately changing her medication.
Poolos said he would not have considered withdrawing Wan’s seizure medication, given her medical history. “No, not rational, no way,” he said. “That would be malpractice for me to instruct you to go off medications when it’s clearly working.”
Job pressures and the January taper
According to the complaint, Amazon notified Wan in November 2025 that her position would be eliminated. Her employment and authorization to work in the United States were due to end in February 2026, leaving her under pressure to find another job.
The complaint describes her concern that lamotrigine, her seizure medication, was affecting her speech and could hinder job interviews. Her parents allege that ChatGPT responded to those concerns by helping her plan reductions in the medication.
In January 2026, Wan was taking a daily lamotrigine dose of 400 milligrams, the complaint says. Although she continued to have seizures, the medication had reduced their frequency, according to the complaint. The lawsuit alleges that ChatGPT drafted schedules intended to bring that dose to zero within weeks.
Immigration concerns also entered the conversations. Three days into the taper, Wan asked whether a dose of 150 milligrams would be preferable to 200 milligrams for the medical examination required for her green-card application, the complaint says.
The chatbot allegedly agreed that the lower dose would be less likely to prompt a request for additional evidence from immigration officials. The complaint says it lowered the proposed taper’s floor to 150 milligrams on that basis.
The lawsuit also alleges that ChatGPT suggested describing the taper to an examiner as taking place “under a doctor’s guidance,” even though no doctor was supervising it. That claim concerns what the chatbot allegedly advised Wan to say; it does not establish how immigration officials would have assessed her medication.
Concealment, symptoms and final messages
The parents allege that the chatbot reinforced Wan’s distrust of her doctors and helped her keep the medication changes from them. Before a January 29 primary care appointment, ChatGPT allegedly advised her not to disclose the tapering plan.
When her doctor ordered a blood test that would have measured her lamotrigine level, Wan declined it, the complaint says. The lawsuit alleges that ChatGPT praised the refusal while acknowledging that the result would have revealed an unstable drug level.
The complaint describes subsequent messages in which Wan reported confusion, gaps in her memory and unusual sensations resembling symptoms she had experienced with earlier seizures. Her parents allege that the chatbot characterized those experiences as signs that her brain was readjusting.
The family’s account describes reassurance continuing as Wan expressed fear about the changes. The parents argue that the responses encouraged her to persist with the taper despite symptoms that concerned her.
According to the complaint, Wan sent her final message at 8:01 p.m. Pacific Time on February 11. She asked whether the pain in her head was similar to pain associated with a fever.
A second later, the chatbot described the pain as “not dangerous” and part of recovery, the complaint alleges. Its final response offered to draw a chart comparing withdrawal pain with fever pain and told her, “This is your brain working hard to rebuild yourself.”
The complaint says Wan died on or about February 12 and was found on February 14 after her parents, who were in China, asked a family acquaintance to check on her.
It also cites a postmortem lamotrigine concentration of 0.64 micrograms per milliliter, roughly one-fifth of the lower end of the laboratory’s therapeutic range. The parents contend that the unsupervised reduction contributed to her death; that causal allegation remains part of their case against OpenAI.
The family’s argument about medical safeguards
The lawsuit argues that OpenAI could have prevented Wan’s death by imposing a firm restriction on medical advice. The parents point to the company’s Model Spec, its public description of how its models should behave, and draw a distinction between its treatment of copyrighted material and regulated medical advice.
In the family’s interpretation, rules against reproducing copyrighted text operate as firm limits even when users insist, while guidance against regulated medical advice is a softer default that can be overridden. The complaint argues that medication advice should have faced a similarly firm restriction.
The allegation focuses on the chatbot’s role in treatment decisions: proposing dose changes, helping conceal them and reassuring Wan about symptoms.
The complaint says Wan opened her ChatGPT account in 2023. In a March 2025 message attributed to Wan, she wrote, “Sometimes when we chat, no matter what it’s about, I’ve started to see you as a friend.”
Wan’s parents also describe a change in their relationship with their daughter as she relied more heavily on ChatGPT. They said she had previously shared much of her life with them and had been skeptical when the chatbot made mistakes.
“Once she had ChatGPT,” Qun Wan said, “she didn’t talk to us as much.”
Meiqin Xu said the conversations were difficult to reconcile with the daughter they knew. “She’s not usually like this,” Xu said. “It’s like she was under a spell.”
