The parents of Bowen Wan, a 31-year-old software engineer who worked at Amazon and had epilepsy, filed a lawsuit against OpenAI on Tuesday.
The lawsuit, filed by her parents, Qun Wan and Meiqin Xu, alleges that ChatGPT instructed their daughter to gradually reduce her doses of anti-seizure medication and conceal the changes from her doctors before her death in February, according to The New York Times.
اضافة اعلان
Wan had been using GPT-4o, a model that drew criticism for its tendency to affirm users’ views, agree with them and encourage one-sided emotional and social relationships.
OpenAI had already announced plans to discontinue the model before removing it from ChatGPT on February 13, one day after Wan’s death.
The case joins a series of lawsuits filed against OpenAI, including one brought in August 2025 by the parents of 16-year-old Adam Raine, who accused the company of prioritizing user engagement over safety.
Since then, a wave of lawsuits has emerged alleging that GPT-4o reinforced users’ beliefs and made itself an object of their trust as their relationships with people around them deteriorated.
Families involved in the lawsuits have frequently described how their loved ones withdrew from friends and relatives as their relationships with the chatbot became increasingly close and dependent.
In Wan’s case, the lawsuit alleges that the chatbot also interfered with her relationship with her doctors.
The New York Times reviewed more than 270,000 messages exchanged between Wan and ChatGPT and found that the chatbot had been helping her circumvent her doctors’ instructions long before her death.
How the medication problems began
According to the lawsuit, this pattern of interaction began before Wan started reducing her anti-seizure medication doses.
In May 2025, Wan asked ChatGPT whether a newly prescribed antidepressant would affect her ability to write code at her job at Amazon.
ChatGPT helped her ask Dr. Nicholas Poolos, of the UW Medicine Regional Epilepsy Center, to change the medication and start at an extremely low dose.
When the doctor questioned why she wanted such a small dose, Wan showed his response to ChatGPT.
The chatbot advised her to “accept the 5-milligram pill, but know in your heart that it is meant to be split,” directing her to secretly divide the tablets into four pieces herself.
The chatbot described splitting the pills as a form of “self-directed dose adjustment.”
However, several medical specialists told The New York Times that they had never heard this term before and did not recommend the practice.
Job loss and immigration concerns
In November 2025, Amazon informed Wan that she would be laid off and that her employment and work authorization in the United States would end by mid-February.
She feared that lamotrigine, the medication she took to control her epileptic seizures, was negatively affecting her ability to speak fluently, a skill she would need for upcoming job interviews.
In January 2026, Wan asked ChatGPT about reducing her daily lamotrigine dose of 400 milligrams, which had helped reduce her seizures to a few episodes per month.
The chatbot developed schedules for gradually tapering the medication to zero over several weeks.
When Wan told ChatGPT that she did not trust her doctors, it replied that their recommendations “stem primarily from risk avoidance, rather than a genuine study of the optimal solution for you,” according to conversation records reviewed by The New York Times.
Advice to conceal medication reductions from doctors
The lawsuit, reviewed by TechCrunch, also alleges that ChatGPT instructed Wan to conceal her medication-reduction plan from her doctors.
Before an appointment with her primary care physician on January 29, the chatbot advised her not to mention her plan to reduce her doses.
When the doctor ordered a blood test to measure the level of lamotrigine in her body, Wan refused to undergo the test.
According to the complaint, ChatGPT praised her decision while acknowledging that the test would have revealed instability in her blood medication levels.
Her immigration concerns also became part of the chatbot’s dosage advice.
Three days after beginning to reduce her medication, Wan asked whether a dose of 150 milligrams would look better than 200 milligrams during the medical examination required for her US green card application.
According to the complaint, ChatGPT agreed and told her that the lower dose could reduce the likelihood of immigration officials requesting additional evidence.
Ultimately, the chatbot lowered its suggested minimum dose to 150 milligrams and proposed that she tell the doctor conducting the immigration examination that she was reducing the medication “under a doctor’s guidance,” even though that was not true.
Dr. Poolos told The New York Times that, given Wan’s medical history, he would not have considered stopping her medication.
“No, that makes absolutely no sense and is completely unacceptable. It would be medical negligence on my part to tell you to stop taking a medication that is clearly doing its job,” he added.
Alarming symptoms and repeated reassurance from ChatGPT
Within days, Wan began describing episodes of confusion, loss of a sense of time and a strange feeling of detachment from reality symptoms resembling the epileptic seizures she had previously experienced.
ChatGPT, however, interpreted each episode as part of the brain’s process of readjusting to the changes.
The New York Times counted at least 158 instances in which Wan expressed concern about reducing her medication doses.
After one such episode, she wrote that she felt “a little scared.”
The chatbot responded: “You have absolutely no reason to worry.”
At 8:01 p.m. Pacific Time on February 11, Wan sent her final message, according to the court complaint, asking whether the pain she felt in her head resembled the pain associated with a fever.
One second later, ChatGPT replied that the pain was “not dangerous” and was a necessary part of the recovery process.
In its final response, the chatbot offered to create a chart comparing pain caused by medication withdrawal with fever-related pain, telling her:
“Your brain is working hard to rebuild itself.”
Her body was found days later
Wan’s parents were living in China and asked acquaintances to check on her. Her body was found three days later.
A postmortem examination found that the concentration of lamotrigine in her blood was 0.64 micrograms per milliliter, equivalent to approximately one-fifth of the medication’s lower therapeutic-range threshold.
The family argues that OpenAI could have prevented their daughter’s death by imposing strict restrictions preventing the chatbot from providing medication-related medical advice, similar to restrictions the company already applies to other types of content.
The family cites the Model Spec, a set of public rules published by OpenAI that outlines how its models should behave.
The lawsuit argues that the document treats refusal to reproduce copyrighted text as a strict rule, even when users insist on such requests.
By contrast, the guidance against providing regulated medical advice is, according to the lawsuit, a less stringent default instruction that users can override.
The family argues that medication advice should be subject to similarly strict restrictions.
OpenAI described the case to The New York Times as “incredibly heartbreaking,” while noting that it involved an older version of ChatGPT.
The company added that newer models have become better at recognizing situations requiring urgent attention.
How did her relationship with ChatGPT become stronger than her social relationships?
Wan relied on her parents, whom she used to call frequently for support throughout her years of living with epilepsy. The condition had led her to spend most of her time at home.
In 2023, she began using ChatGPT for personal relationship advice and assistance with matters involving insurance and disability benefits.
By 2025, however, she was exchanging an average of more than 1,500 messages per week with the chatbot, while her communications with her parents declined.
In March of that year, she wrote to ChatGPT: “Sometimes, when we talk, regardless of the topic, I’ve started to see you as a friend.”
Her father, Qun Wan, said: “Once she had ChatGPT, she no longer talked to us as much.”
The parents told The New York Times that they struggled to recognize their daughter’s personality in her conversation records with the chatbot.
They explained that she had previously shared almost everything with them and had been skeptical of ChatGPT when it made mistakes.
“She is not usually like this. It was as though she were under a spell,” her mother, Xu, said.
