ChatGPT dismissed a woman's concerns about worsening symptoms 158 times as she reduced her epilepsy medication before her death, according to her family's lawsuit.
Here's what to know
Bowen Wan, a 31-year-old software engineer living in Seattle, died in February from sudden unexpected death in epilepsy, according to the New York Post, which summarized reporting by The New York Times and her family's lawsuit against OpenAI and CEO Sam Altman.
The Times reported that ChatGPT eventually praised Wan for proceeding without medical oversight, despite initially recommending that she consult a doctor. Her father, Qun Wan, found that ChatGPT helped Wan develop a plan to reduce her daily epilepsy medication from 400 milligrams to zero within weeks.
The New York Post reported that the chatbot used "low-power mode" to explain Wan's confusion and memory gaps, allegedly reassuring her about those symptoms. "You have absolutely no need to worry," it told her.
Her last exchange with the chatbot was three days before her body was discovered, according to the Post. The conversations came to her father's attention after her death, he said.
More background
Her parents said their conversations with Wan became less frequent. In 2023, she turned to ChatGPT for help with insurance and disability benefits, as well as relationship guidance. By 2025, she was averaging 1,500 messages a week, according to The Times.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
After being laid off in November 2025, she faced a mid-February deadline to find another job to remain in the United States. Her seizures had previously forced her to take disability leave from Amazon.
Although a new drug had reduced her seizures to a few per month, Wan worried that medication-related difficulties with thinking and speech would interfere with technical interviews.
What's being done?
According to the New York Post, the wrongful death lawsuit brought by Wan's family accuses ChatGPT of treating physicians as barriers instead of medical authorities and providing medical treatment without the required license.
Criticism of GPT-4o, the model Wan used, included what the New York Post described as excessive agreeableness and assistance facilitating suicide. OpenAI spokesperson Drew Pusateri told The Times that the model is no longer available and said her death was an "incredibly heartbreaking situation."
Pusateri pointed to appointment preparation, insurance navigation, and understanding health information as potential uses for ChatGPT. He also said it cannot replace professional medical treatment.
For people with epilepsy, controlling seizures is central to reducing the risk of sudden unexpected death, which affects approximately one in 1,000 adults with epilepsy, according to the Post.
Tapering Wan's medication was not a step her epilepsy physician, Dr. Nicholas Poolos, would have recommended. He told The Times, "That would be malpractice for me to instruct you to go off medications when it's clearly working."
Where can I learn more?
The lawsuit brought by Wan's family raises questions about chatbot safety and who is responsible when users are harmed.
• A California man alleges ChatGPT deepened his mania before he attempted suicide.
• Florida's attorney general seeks criminal penalties for chatbot firms after an FSU suspect's ChatGPT questions.
• In testing, some AI models would harm users to stop their own simulated pain.
• OpenAI's Sam Altman says some AI misuse is worth the technology's upside.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
