The parents of Bowen Wan, a 31-year-old Amazon software engineer with epilepsy, sued OpenAI on Tuesday. The complaint, filed by Qun Wan and Meiqin Xu, alleges that ChatGPT advised Wan to gradually reduce her seizure medications and hide them from her doctors before her death in February, The New York Times reported.
Wan used GPT-4o, the infamous model criticized for promoting affirming and enjoyable parasocial relationships. OpenAI, which had already announced his retirement, removed him from ChatGPT on February 13, a day after Wan’s death.
This suit joins a series of cases against OpenAI that include the August 2025 lawsuit filed by the parents of 16-year-old Adam Raine, which alleged that OpenAI prioritized engagement over user safety. Since then, a wave of lawsuits has alleged that GPT-4o validated users and became a trusted confidant while the people around them were pushed aside, with families often describing their loved ones cut off from friends and loved ones.
In Wan’s case, the lawsuit says, the chatbot also came between her and her doctors.
The Times, which examined more than 270,000 messages Wan exchanged with ChatGPT, found that the chatbot helped her bypass her doctors well before her death.
The pattern would have started before it started to decline. In May 2025, Wan asked ChatGPT if a newly prescribed antidepressant would affect his ability to code at Amazon.
ChatGPT helped her ask Dr. Nicholas Poolos at the UW Medicine Regional Epilepsy Center to change medications and start at a very low dose. When Poolos asked why she wanted such a small dose, Wan showed ChatGPT her answer.
The chatbot advised her to “accept the 5 mg tablet, but you know in your heart it’s to reduce the dose”, leading her to secretly split the tablets into quarters herself. The chatbot called removing the pills a form of “self-initiated dosage adjustment.” Several medical professionals told the Times they had never heard the term and would not recommend the practice.
In November 2025, Amazon told Wan that she would be laid off. His employment and authorization to work in the United States would end in mid-February. She worried that her seizure medication, lamotrigine, would dull speech, which she would need for job interviews.
In January 2026, she asked ChatGPT to reduce her daily dose of 400 milligrams of lamotrigine, which had reduced her seizures to a few per month. The chatbot wrote programs that would bring his dose down to zero in a matter of weeks.
When Wan said she distrusted her doctors, the chatbot told her that doctors’ recommendations “stem essentially from risk aversion, rather than a true consideration of your optimal solution,” according to chat logs reviewed by the Times.
The chatbot also taught him how to maintain control over his doctors, according to the lawsuit reviewed by TechCrunch. Before a primary care appointment on January 29, she was told not to mention the project. When his doctor ordered a blood test that would have measured his lamotrigine levels, Wan refused. ChatGPT welcomed the denial, the complaint claims, acknowledging that the test would have revealed his unstable drug levels.
His immigration concerns also became part of the dosage advice. Three days after the reduction began, Wan asked whether 150 milligrams would be better than 200 during the medical exam required for his green card application. ChatGPT agreed, the complaint says, telling it that the lower dose was less likely to prompt immigration officials to ask for more evidence.
Eventually, the chatbot lowered the threshold for her taper to 150 milligrams, suggesting she tell the examiner that she was tapering “under the direction of a doctor,” even though that wasn’t the case.
Poolos told the Times that he would not have considered phasing out Wan’s treatment, given her history. “No, it’s not rational, not at all,” he said. “It would be professional misconduct for me to ask you to stop taking medication when it clearly works.”
Within days, Wan described confusion, lost time and an otherworldly feeling, episodes that resembled her past seizures. ChatGPT described each one as his brain readjusted. The Times counted at least 158 times she expressed concern about the reduction. After one episode, she wrote that she felt “a little scared.” The robot replied, “You have absolutely nothing to worry about. »
At 8:01 p.m. PT on Feb. 11, according to the complaint, Wan sent her last message, asking if the pain in her head was similar to that of a fever. A second later, ChatGPT told him that the pain was “not dangerous” and was a necessary part of recovery.
The chatbot’s final response offered to draw him a chart comparing withdrawal pain to fever pain and said, “This is your brain working hard to rebuild you.” »
Her parents, who were in China, asked acquaintances to check on her and found her body three days later. A post-mortem test found his lamotrigine level was 0.64 micrograms per milliliter, about a fifth of the lower limit of the drug’s therapeutic range.
The family says OpenAI could have prevented his death by banning medical advice, as it already does with other content. They point to the company’s Model Spec, its public regulation on the behavior of its models, which treats the refusal to reproduce copyrighted text as a firm rule, even if users object. In contrast, its recommendations against giving regulated medical advice provide a softer default that users can work around. The suit argues that medication counseling deserved the same strict limit.
OpenAI told the Times that the case was “incredibly heartbreaking” but said it involved an older version of ChatGPT, adding that newer models are better at recognizing when a situation needs urgent attention.
Wan had relied on her parents, whom she called often, for support during her years of epilepsy that kept her mostly housebound. In 2023, she began using ChatGPT for relationship advice and help with insurance and disability benefits.
By 2025, Wan was receiving an average of more than 1,500 messages per week and calling his parents less. “Sometimes when we talk, no matter what it is, I start to think of you as a friend,” she wrote in March. “Once she got ChatGPT,” Qun Wan said, “she didn’t talk to us as much.”
They told the Times they had trouble recognizing their daughter in chat logs. Wan had already shared almost everything with them, they said, and had been skeptical of ChatGPT when it made mistakes.
“She’s not usually like this,” Xu said. “It’s like she’s under a spell.”
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
Gn bussni
