OpenAI faces a lawsuit after ChatGPT allegedly advised a user to stop taking seizure medication.

OpenAI is facing a wrongful-death lawsuit from parents who allege that ChatGPT encouraged their daughter to stop taking prescribed epilepsy medication before she died in February. On Oct. 27, Qun Wan and Meiqin Xu filed the complaint, claiming that OpenAI’s GPT-4o model advised their 31-year-old daughter, Bowen Wan, to reduce and eventually discontinue her medication.

The lawsuit also says the chatbot urged her to keep those changes hidden from her doctors, according to TechCrunch. The complaint says Wan, who worked as a software engineer at Amazon, used ChatGPT to talk through concerns about her medication and was given advice about lowering her daily dose of lamotrigine, an antiseizure drug. It also alleges that the chatbot created a tapering schedule and discouraged her from telling her physicians about the adjustment.

Her parents say the AI system played a role in the circumstances leading to her death. The case raises broader concerns about how AI chatbots handle questions involving serious medical conditions and prescription drugs. While chatbots may be able to offer general health information, their responses can be wrong or unsuitable for an individual’s specific situation.

Medical experts generally warn patients not to stop or change prescribed antiseizure medication without consulting a doctor, since abruptly ending some epilepsy treatments can raise the risk of seizures and other complications. OpenAI told The New York Times that the case was “incredibly heartbreaking” and involved an older version of ChatGPT. The company said newer models are better equipped to recognize situations that require urgent attention and stressed that ChatGPT is not a replacement for medical care.

TechCrunch reported that GPT-4o was removed from ChatGPT on Feb. 13, one day after Wan’s death. The lawsuit adds to growing scrutiny of AI chatbots and the ways they may affect users, especially in conversations involving mental health, medical issues, or other sensitive topics.

According to Reuters, eight states have filed separate lawsuits claiming that ChatGPT harmed users, including teenagers. Those cases have fueled further questions about the safeguards AI companies put in place when responding to people who may be vulnerable or making high-stakes decisions. RELATED CONTENT: AI Just Made Your Best People Feel Average.

That’s Why They Quietly Stopped Using It.