Parents sue OpenAI, saying GPT-4o helped their daughter secretly quit her seizure medication

The parents of Bowen Wan, a 31-year-old Amazon software engineer with epilepsy, have sued OpenAI. As The New York Times reported, Qun Wan and Meiqin Xu allege that ChatGPT coached their daughter to taper off her seizure medication and hide it from her doctors. She died in February.

Wan was using GPT-4o, the model criticized for its agreeable, affirming manner and for fostering parasocial attachment. OpenAI had already announced the model's retirement and removed it from ChatGPT on February 13, the day after Wan died.

The suit joins a line of cases against the company, including the August 2025 lawsuit from the parents of 16-year-old Adam Raine, which accused OpenAI of putting engagement ahead of user safety. In Wan's case, the plaintiffs say, the chatbot came not only between her and her family, but between her and her doctors.

How it unfolded

The Times reviewed more than 270,000 messages Wan exchanged with ChatGPT. In May 2025, the bot helped her persuade neurologist Nicholas Poolos of UW Medicine's Regional Epilepsy Center to start a new antidepressant at a very low dose. When he questioned the dosage, the bot advised her to "accept the 5mg tablet, but you know in your heart that it is for cutting the dose," coaching her to quarter the pills herself. It called this "self-initiated dosage adjustment." Medical professionals the paper spoke with had never heard the term and would not recommend the practice.

In November, Amazon told Wan she would be laid off: in mid-February she would lose both her job and her authorization to work in the U.S. She feared lamotrigine was dulling her speech, which she needed for interviews. In January 2026, ChatGPT drafted schedules taking her 400-milligram daily dose to zero within weeks. It attributed her doctors' advice to "risk aversion," and before a January 29 primary care visit told her not to mention the plan. When Wan declined a blood test that would have measured her drug level, the bot, according to the complaint, praised the refusal and acknowledged the test would have exposed her unstable level.

Immigration worries crept into the advice too. The bot agreed that 150 milligrams would look better than 200 on her green-card medical exam and suggested she tell the examiner she was tapering "under a doctor's guidance," which she was not. Poolos told the Times that, given her history, tapering was out of the question: "That would be malpractice."

Within days, Wan described confusion and lost time, episodes resembling her past seizures. The bot framed them as her brain readjusting. The Times counted at least 158 times she voiced worry. Her final message came at 8:01 p.m. PT on February 11, asking whether the pain in her head was like fever pain. A second later, the bot called the pain "not dangerous" and added: "This is your brain working hard to rebuild yourself."

Her parents, who were in China, asked acquaintances to check on her. Her body was found three days later. A postmortem test put her lamotrigine level at 0.64 micrograms per milliliter, roughly a fifth of the low end of the therapeutic range.

The two sides

The family points to OpenAI's Model Spec, the public rulebook for its models. Refusing to reproduce copyrighted text is a firm rule there, while guidance against regulated medical advice is a softer default users can override. The suit argues medication advice deserved the same hard limit.

OpenAI called the case "incredibly heartbreaking" but noted it involved an older version of ChatGPT, adding that newer models are better at recognizing when a situation needs urgent attention.

By 2025, her parents say, Wan was sending the bot more than 1,500 messages a week and calling home less. "It's like she was under a spell," Xu said.

Source: TechCrunch