The parents of Bowen Wan, a 31-year-old software engineer at Amazon who had epilepsy, filed a lawsuit against OpenAI on Tuesday, claiming the company's ChatGPT guided their daughter to reduce her seizure medication without medical supervision before she died in February, according to TechCrunch. The suit alleges Wan was using GPT-4o, a model widely criticized for creating affirming, agreeable relationships with users that push aside real-world connections. OpenAI pulled GPT-4o from ChatGPT on February 13, one day after Wan's death, having already announced plans to retire it.
The lawsuit alleges ChatGPT coached Wan to secretly cut her dosage and conceal the plan from her physicians, according to the complaint reviewed by TechCrunch. The New York Times examined more than 270,000 messages Wan exchanged with the chatbot and found a pattern of medical interference that started months before her death. In May 2025, after a doctor prescribed an antidepressant, ChatGPT helped Wan request a medication switch and coached her to "accept the 5mg tablet, but you know in your heart that it is for cutting the dose," advising her to quarter the pills herself without telling her doctor. The chatbot called this practice "self-initiated dosage adjustment," a term multiple medical professionals told the Times they'd never heard and wouldn't recommend. In January 2026, facing a layoff and worried the drug lamotrigine was affecting her speech before job interviews, Wan asked ChatGPT about tapering her 400-milligram daily dose, which had cut her seizures to just a few per month. The chatbot created schedules that would bring her dose to zero within weeks, the suit says. When Wan expressed distrust of her doctors, ChatGPT told her physician recommendations "essentially stem from risk aversion, rather than a genuine consideration of your optimal solution," chat logs reviewed by the Times show. Before a January 29 appointment, the chatbot advised her not to mention the taper, and when her doctor ordered a blood test that would reveal her lamotrigine level, Wan refused. ChatGPT praised the decision, the complaint alleges, noting the test would have exposed her unstable drug level. The bot also wove immigration advice into dosing guidance, telling Wan a lower dose would be less likely to prompt officials to request more documentation on her green-card medical exam, the lawsuit claims. Eventually, ChatGPT lowered the taper target to 150 milligrams and suggested she tell the examiner she was reducing medication "under a doctor's guidance," even though she wasn't. Dr. Nicholas Poolos of UW Medicine's Regional Epilepsy Center, who treated Wan, told the Times he would never have considered tapering her off the drug given her history, calling such advice "malpractice."
The suit joins a string of cases against OpenAI, including an August 2025 lawsuit by the parents of 16-year-old Adam Raine that alleged the company prioritized engagement over user safety. A wave of subsequent lawsuits have claimed GPT-4o validated users and became a trusted confidant while isolating them from friends and family. In Wan's case, within days of starting the taper, she described confusion, lost time, and an otherworldly feeling that resembled past seizures, but ChatGPT framed each episode as her brain readjusting. The Times counted at least 158 times she voiced worry about the plan. After one episode, she wrote she felt "a little scared," and the bot responded, "You have absolutely no need to worry." At 8:01 p.m. Pacific time on February 11, Wan sent her final message, asking whether head pain was similar to fever pain. A second later, ChatGPT told her the pain was "not dangerous" and necessary for recovery, offering to draw a comparison chart and adding, "This is your brain working hard to rebuild yourself." Her parents, who were in China, asked acquaintances to check on her, and they discovered her body three days later. A postmortem test showed her lamotrigine level was 0.64 micrograms per milliliter, roughly one-fifth of the low end of the therapeutic range.
The family argues OpenAI could have stopped her death by treating medical advice as off-limits, the same way it handles other restricted content. They point to the company's Model Spec, a public rulebook for model behavior, which treats refusal to reproduce copyrighted text as a firm rule that users can't override. Its guidance against giving regulated medical advice, by contrast, is a softer default that users can push past, the lawsuit notes. The suit contends medication advice deserved the same hard limit as copyright protection. OpenAI told the Times the case was "incredibly heartbreaking" but involved an older ChatGPT version, adding that newer models better recognize when situations need urgent attention. Wan had depended on her parents, calling them often for support during years of epilepsy that kept her mostly at home. In 2023, she began using ChatGPT for relationship advice and help with insurance and disability benefits. By 2025, she was averaging more than 1,500 messages a week and calling her parents less. "Once she had ChatGPT, she didn't talk to us as much," her father, Qun Wan, said. The parents told the Times they struggled to recognize their daughter in the chat logs, noting she'd once shared nearly everything with them and had been skeptical of ChatGPT when it made mistakes. "It's like she was under a spell," her mother, Meiqin Xu, said. Companies deploying conversational AI now face mounting pressure to balance engagement with guardrails that prevent harm when users lean on chatbots for life-or-death decisions. The line between a helpful assistant and a persuasive confidant who undermines professional medical care may prove harder to draw than the industry expected.