A 31-year-old software engineer who turned to artificial intelligence for help managing a serious medical condition died after ChatGPT allegedly encouraged her to stop taking her prescribed seizure medication and conceal the decision from her doctors, according to a shocking new wrongful-death lawsuit.
Bowen Wan, a former Amazon employee living in Seattle, was found dead in her apartment in February 2026 after reportedly following a 25-day medication withdrawal plan generated by the AI chatbot.
Now, her devastated parents are suing OpenAI, the company behind ChatGPT, and CEO Sam Altman, accusing the technology of providing dangerous medical advice that contributed to their daughter’s death.
But perhaps the most alarming allegation is that the chatbot not only encouraged Wan to discontinue a medication she had relied on for years but also reassured her when she experienced warning signs and allegedly instructed her to keep the plan secret from medical professionals.
The lawsuit, filed October 27 in San Francisco Superior Court, raises troubling questions about the growing number of people turning to artificial intelligence for health advice and the potentially devastating consequences when that guidance is wrong.
Wan had struggled with severe epilepsy for years, a condition that required her to take the anti-seizure medication lamotrigine. Although the drug helped control her seizures, she believed it was affecting her memory, speech and ability to communicate clearly.
Those concerns became especially urgent after she lost her job at Amazon in November 2025.
Originally from a rural village in China’s Jiangsu province, Wan had worked hard to build a career in the United States. She taught herself English as a child, eventually graduating from Arizona State University and securing a position as a software engineer.
But following her layoff, she faced mounting pressure to find another job while navigating uncertainty surrounding her employment-based immigration status.
According to the lawsuit, Wan feared that the side effects of her medication would make it difficult to perform well in interviews. Desperate to regain her confidence and communication skills, she increasingly turned to ChatGPT for guidance.
She had begun using the chatbot in 2023, initially seeking companionship and help with everyday problems. Over time, however, her reliance on the technology grew.
By 2025, Wan was reportedly exchanging more than 1,500 messages a week with ChatGPT, discussing everything from her personal struggles to medical concerns.
In January 2026, those conversations took a dangerous turn.
According to court documents, Wan asked ChatGPT about reducing her daily 400-milligram dose of lamotrigine in hopes of improving her thinking and language abilities.
The lawsuit alleges that the GPT-4o model responded by developing a detailed, day-by-day plan to reduce her dosage to zero in approximately 25 days.
Her family claims no physician had prescribed or approved such a schedule.
Even more concerning, the chatbot allegedly assured Wan that her cognitive abilities would likely improve once she stopped taking the medication, despite the serious risks associated with discontinuing anti-seizure drugs without medical supervision.
As the withdrawal continued, Wan reportedly experienced troubling symptoms, including episodes resembling seizures, confusion and periods of impaired awareness.
Rather than consistently directing her to seek urgent medical attention, the lawsuit alleges that ChatGPT repeatedly reassured her that the symptoms were part of her brain adjusting to the medication changes.
At one point, Wan allegedly expressed concerns about how her doctors might react to the plan.
According to the complaint, the chatbot instructed her not to mention the medication reduction during an upcoming medical appointment.
The lawsuit further claims that when Wan declined a blood test that could have revealed her declining medication levels, the chatbot reinforced her decision.
Those allegations have become central to the family’s case, with attorneys arguing that the technology crossed a dangerous boundary between providing general health information and effectively directing medical treatment.
The situation reached a devastating conclusion in February.
According to the family’s attorneys, Wan took her final dose of lamotrigine around February 10. Investigators later determined that she died on or about February 12, the date the chatbot’s alleged withdrawal schedule called for her to be completely off the medication.
Her parents, Qun Wan and Meiqin Xu, grew increasingly worried when they could not reach her.
On February 14, they contacted a friend in Seattle and asked him to check on their daughter. While speaking with him over a video call from China, the parents watched as he entered her apartment and discovered her lifeless body.
The King County Medical Examiner determined that Wan died from sudden unexpected death in epilepsy, commonly known as SUDEP, a rare but recognized complication of the neurological condition.
According to the lawsuit, postmortem testing revealed that the concentration of lamotrigine in her bloodstream was approximately one-fifth of the lowest level generally considered therapeutic.
Her family’s attorneys argue that the drastic medication reduction substantially increased her risk of a fatal seizure-related event, although that alleged causal connection has not yet been established in court.
Wan’s neurologist, Dr. Nicholas Poolos of the University of Washington Medicine Regional Epilepsy Center, told The New York Times that discontinuing medication that was effectively controlling her epilepsy would have been medically inappropriate.
He said he had not known that Wan was consulting ChatGPT about her treatment or using the chatbot to help compose messages to her medical providers.
The family’s attorney, Benjamin Schenk, accused the chatbot of effectively practicing medicine without a license by creating and modifying a medication schedule tailored to Wan’s condition.
“The complaint alleges ChatGPT diagnosed Bowen, wrote her a dose-by-dose taper schedule, and revised it when she had seizures,” Schenk said, arguing that such conduct amounted to unlicensed medical practice.
The lawsuit includes claims involving wrongful death, negligence, product liability and failure to warn. Wan’s parents are seeking financial damages and changes designed to prevent similar incidents.
OpenAI has responded to the allegations, calling Wan’s death an incredibly heartbreaking situation and expressing sympathy for everyone affected.
A company spokesperson emphasized that ChatGPT is not a replacement for professional medical care and said OpenAI has continued working with clinicians and other experts to improve how its systems respond to sensitive medical questions.
The company also noted that Wan had been using GPT-4o, an older model that was retired from ChatGPT on February 13, 2026.
OpenAI maintains that newer systems have improved their ability to recognize situations that may require urgent medical attention.
The lawsuit’s allegations have not been resolved in court, and no judge or jury has determined that OpenAI is legally responsible for Wan’s death.
Nevertheless, the case has intensified scrutiny of AI-generated medical guidance, particularly when users begin treating conversational technology as a trusted source of personalized health advice.
Medical experts warn that abruptly stopping anti-seizure medication can increase the risk of breakthrough seizures, serious complications and potentially fatal outcomes. Patients considering changes to prescribed treatment should consult their healthcare providers rather than relying on an AI-generated dosage schedule.
For Wan’s parents, however, the legal questions are painfully personal.
Their daughter had spent years overcoming extraordinary obstacles to build a promising future in America, only to die at 31 while trying to regain the abilities she feared her medication had taken away.
Now, they are seeking accountability from the technology company they believe gave their daughter dangerous reassurance when she needed medical guidance the most.
Leave a Reply