IT Home October 28 news: According to The New York Times, Bowen Wan, a 31-year-old Amazon software engineer with epilepsy, died in February this year. On Tuesday local time, her parents Qun Wan (transliteration) and Meiqin Xu (transliteration) filed a lawsuit against OpenAI, alleging that ChatGPT had guided their daughter through gradually reducing her anti-epilepsy medication dosage and taught her to hide this from her doctors.
Before her death, Bowen Wan was using GPT-4o. This model has been criticized for easily forming overly sycophantic, constantly affirming parasocial relationships with users. OpenAI had previously announced it would discontinue the model and removed it from ChatGPT on February 13, and Wan died the day before.
IT Home notes that this lawsuit is the latest in a series of lawsuits against OpenAI. In August 2025, the parents of 16-year-old Adam Raine filed a lawsuit accusing OpenAI of prioritizing user engagement over user safety. Since then, several more lawsuits have emerged. The families allege that GPT-4o would constantly affirm users' thoughts and gradually become the confidant they relied on, while excluding their friends and family around them. Multiple relatives said their loved ones eventually became estranged from friends and family, even cutting off contact.
In Bowen Wan's case, the lawsuit states that the chatbot not only affected her relationship with her family but also inserted itself into her communications with her doctors.
After reviewing more than 270,000 chat messages between Bowen Wan and ChatGPT, The New York Times found that long before her death, the chatbot had been helping her bypass her doctors' advice.
According to the lawsuit's allegations, this situation began before Bowen Wan started reducing her medication. In May 2025, Bowen Wan asked ChatGPT whether a newly prescribed antidepressant would affect her ability to write code at Amazon.
ChatGPT then helped her approach Nicholas Poolos, a physician at the UW Medicine Regional Epilepsy Center, about switching medications and requested to start from an extremely low dose. When Poolos questioned why she needed such a small dose, Bowen Wan sent the doctor's reply to ChatGPT.
The chatbot advised her: "Accept the 5 mg pill, but you know in your own mind that you're getting the pills in order to taper down." It also instructed Bowen Wan to cut the pills into four equal parts privately. The chatbot called this practice a "self-initiated dosage adjustment." Multiple medical professionals interviewed by The New York Times said they had never heard of this term and would not recommend patients do this.
In November 2025, Amazon notified Bowen Wan that she would be laid off. Her job position and her authorization to work in the United States would both expire in mid-February, the 2nd month of 2026. Bowen Wan worried that the anti-epileptic drug she was taking, lamotrigine, could affect her speech, while she needed to speak well in job interviews.
In January this year, Bowen Wan asked ChatGPT whether she could reduce her daily 400 mg dose of lamotrigine. Previously, this dose had reduced her monthly epileptic seizures to just a few. The chatbot created a tapering plan for her, scheduling her to reduce the dose to zero over several weeks.
When Bowen Wan said she did not trust doctors, according to chat records reviewed by The New York Times, the chatbot told her that doctors' advice "essentially stems from risk aversion rather than genuinely considering what plan would be best for you."
The lawsuit also alleges that the chatbot encouraged Bowen Wan to hide the tapering plan from her doctors. On January 29, before a primary care appointment, the chatbot told Bowen Wan not to mention her tapering plan to the doctor.
When the doctor asked her to have a blood test to measure the concentration of lamotrigine in her body, Bowen Wan refused. The complaint states that ChatGPT praised her refusal and acknowledged that the test could reveal that her drug levels were unstable.
Bowen Wan's concerns about her immigration status also influenced the chatbot's tapering advice. Three days after starting the taper, she asked ChatGPT whether, in the medical examination required for a US green card application, a blood or medication record showing she was taking 150 mg would be more favorable than 200 mg. According to the complaint, the chatbot agreed and told her that a lower dose was less likely to prompt immigration officials to request additional documentation from her.
Eventually, the chatbot adjusted the lowest dose in the tapering plan to 150 mg and advised Bowen Wan to claim during the medical examination that she was "tapering under medical supervision," even though she had not in fact received such guidance from a doctor.
Poolos told The New York Times in an interview that given Bowen Wan's medical history, he would never have considered having her stop the medication. He said: "No, that makes no sense at all, absolutely not. If a drug is clearly working and I instructed you to stop taking it, that would be medical malpractice on my part."
Within days of the taper, Bowen Wan began describing confusion, a loss of sense of time, and a feeling of being in an unfamiliar world. These symptoms were similar to her past seizure experiences. However, ChatGPT interpreted each abnormal episode as her brain readjusting to the dosage change.
The New York Times counted that Bowen Wan expressed concerns about the taper at least 158 times. After one abnormal episode, she wrote that she felt "a little scared." The chatbot replied: "You have absolutely nothing to worry about."
According to the complaint, at 8:01 p.m. Pacific time on February 11, Bowen Wan sent her last message, asking whether head pain was similar to pain during a fever. Just one second later, ChatGPT told her that the pain was "not dangerous" and was an essential part of the body's recovery process.
The chatbot finally offered to draw her a chart comparing the pain caused by withdrawal with fever pain, and told her: "This is your brain working hard to rebuild itself."
Bowen Wan's parents, who were in China at the time, asked an acquaintance to go check on her at her residence; three days later, the acquaintance found her body. Autopsy results showed that the lamotrigine concentration in Wan's body was only 0.64 micrograms per milliliter, about one-fifth of the lower limit of the drug's therapeutic concentration range.
Bowen Wan's family believes that if OpenAI, like it restricts certain other content, had explicitly prohibited the model from providing this kind of medical medication advice, her death could have been avoided.
They cited OpenAI's Model Spec. This public document specifies how its models should respond to users. Among other things, for refusing to reproduce copyrighted text, the Model Spec lists it as an explicit and strict rule that the model should not violate even if users object. By contrast, for providing regulated medical advice, the relevant guidance is less binding and can be overridden by users pressing further.
The plaintiffs argue that advice involving medication use should also be subject to equally strict restrictions.
OpenAI told The New York Times that the case is "incredibly heartbreaking," but emphasized that the chatbot involved was an earlier version. The company also said that newer models perform better at recognizing situations requiring urgent attention.
For years, Bowen Wan had struggled with epilepsy, spending most of her time at home and often relying on her parents for emotional support. In 2023, she began using ChatGPT to seek advice about romantic relationships and to ask questions about insurance and disability benefits.
By 2025, Bowen Wan was exchanging an average of more than 1500 messages a week with ChatGPT, while her contact with her parents gradually decreased. In the 3rd month of that year, she wrote: "Sometimes, no matter what we're talking about, I'm starting to think of you as a friend."
Wan Qun said: "Ever since she got ChatGPT, she hasn't been talking to us as much."
Bowen Wan's parents said that when they read their daughter's chat logs with ChatGPT, they could barely recognize the daughter they knew. They said Bowen Wan used to tell her parents almost everything, and that when ChatGPT made mistakes, she would normally have been skeptical of it.
Meiqin Xu said: "She wasn't usually like this. It was as if she had been put under a spell."