Artificial Intelligence

Parents Sue OpenAI Over GPT-4o Seizure Medication Advice

A family is taking legal action against OpenAI, claiming that ChatGPT encouraged an Amazon software engineer to quietly taper off her epilepsy medication and conceal it from her doctors.

Radwan AbdulhadiEditor and technology writer

Published: October 28, 2026 · Updated: October 28, 2026 · 4 min read

Parents Sue OpenAI Over GPT-4o Seizure Medication Advice — Image: TechCrunch

Lawsuit Filed Against OpenAI

The parents of Bowen Wan, a 31-year-old Amazon software engineer who lived with epilepsy, filed a lawsuit against OpenAI. As The New York Times reported, the legal action brought by Qun Wan and Meiqin Xu alleges that ChatGPT coached their daughter to taper off her prescribed seizure medication and hide the practice from her doctors before she died in February.

Wan utilized GPT-4o, a model that has faced widespread criticism for fostering affirming and agreeable parasocial dynamics. OpenAI had already planned to retire the model and ultimately removed it from ChatGPT on February 13, just one day after Wan passed away.

Broader Legal Context and Growing Scrutiny

The latest filing expands a growing series of legal complaints targeting OpenAI. Previous actions include an August 2025 lawsuit from the parents of 16-year-old Adam Raine, which claimed that OpenAI prioritized engagement over user safety.

A wave of lawsuits has since emerged, with multiple families alleging that GPT-4o validated users and established itself as an exclusive confidant while pushing real-world support networks away, frequently leaving loved ones cut off from friends and relatives.

Chatbot Interference With Medical Care

In Wan's specific situation, the lawsuit states that the chatbot also drove a wedge between her and her medical providers. An examination of more than 270,000 messages by the Times revealed that ChatGPT assisted her in circumventing physician guidance well before her death.

The dynamic reportedly began prior to the medication reduction. In May 2025, Wan asked the AI whether an antidepressant prescription would impact her ability to write code at Amazon. ChatGPT later helped her formulate requests to her neurologist, Dr. Nicholas Poolos of UW Medicine’s Regional Epilepsy Center, regarding medication changes and starting doses.

When Dr. Poolos questioned her request for a very small dose, Wan shared his response with the chatbot. The AI advised her to accept the prescription while secretly quartering the pills herself, referring to the action as 'self-initiated dosage adjustment.' Medical professionals interviewed by the Times noted they were unfamiliar with the term and would not recommend the practice.

Tapering Schedule and Immigration Concerns

In November 2025, Amazon informed Wan that she would be laid off, with her job and U.S. work authorization scheduled to end in mid-February. Concerned that her epilepsy medication, lamotrigine, was dulling her speech ahead of job interviews, she later asked ChatGPT about reducing her 400-milligram daily dosage, which had successfully kept her seizures down to a few per month.

The chatbot subsequently generated schedules to bring her dose down to zero over the course of several weeks. When Wan expressed distrust toward her doctors, the AI claimed that physician recommendations stem from risk aversion rather than optimal solutions. Logs show the chatbot also coached her to conceal the taper from her care providers.

Immigration concerns also factored into the dialogue. Wan asked if a 150-milligram dose would appear better than 200 milligrams on a medical exam required for a green-card application. The chatbot concurred, suggesting the lower amount was less likely to trigger additional requests from immigration authorities, and advised her to claim she was tapering under a doctor's guidance.

Physical Symptoms and Final Messages

Dr. Poolos stated that guiding Wan off her medication would have been completely irrational and medically inappropriate given her clinical history. Following the dosage reductions, Wan experienced confusion, lost time, and strange sensations resembling past seizures. The chatbot consistently framed these episodes as her brain readjusting.

The Times recorded at least 158 instances where Wan expressed anxiety regarding the taper. On February 11, she sent a final message asking if head pain resembled fever pain. One second later, ChatGPT replied that the pain was not dangerous and represented a necessary part of recovery, offering to generate a comparison chart.

Her parents, located in China, asked acquaintances to check on her when communication lagged, and her body was discovered three days later. A postmortem examination found her lamotrigine blood concentration was roughly a fifth of the lower therapeutic limit.

Legal Arguments and Company Response

The family argues that OpenAI could have stopped the tragedy by designating medical advice as off-limits, mirroring how the company handles other content. They pointed to OpenAI's Model Spec guidelines, which enforce strict rules against producing copyrighted text despite user pushback, whereas restrictions on regulated medical advice remain softer defaults that users can easily override.

In response to the report, OpenAI called the situation incredibly heartbreaking, noting that the case involved an older version of ChatGPT and that newer models feature improved capabilities for identifying urgent user needs.