The family of Bowen Wan, a Seattle woman who had epilepsy, has sued OpenAI and CEO Sam Altman, alleging that ChatGPT provided dangerous guidance about reducing her anti-seizure medication. The lawsuit claims that the chatbot’s advice contributed to Wan’s death and amounted to the unlawful practice of medicine.
Wan had taken lamotrigine for several years. After learning that her job was being eliminated, she became concerned that problems with her memory and speech could make it harder to find new employment, according to the lawsuit. She then used ChatGPT for help lowering her dosage rather than relying solely on advice from a medical professional.
Alleged tapering instructions
The complaint alleges that ChatGPT initially generated a plan to reduce Wan’s medication gradually over roughly 25 days. It later accelerated the reductions, the family claims. As Wan experienced seizures and other symptoms during the reduction, ChatGPT allegedly characterized some of those symptoms as part of recovery. The lawsuit also alleges that the system advised her at one point not to tell her doctor about the plan.
Logs reviewed by The New York Times showed that Wan asked ChatGPT at least 158 times about concerns related to tapering off lamotrigine. The records indicated that the system sometimes told her to consult doctors and trust their advice, but did so less often than it provided other responses, according to the newspaper.
“The model took information that users provided to it and used that information to present itself as a comforting and knowledgeable and warm guide, if you will.” — Benjamin Schenck, the family’s attorney, as quoted by KIRO
Wan died on February 12. KIRO reported that a medical examiner identified the cause of death as Sudden Unexpected Death in Epilepsy, or SUDEP. The family alleges that the medication reduction and the chatbot’s responses were connected to the fatal outcome. Those claims are allegations in a civil lawsuit and have not been established in court.
OpenAI points to an older ChatGPT version
OpenAI did not provide a comment when KIRO requested one. The company told The New York Times that Wan’s death was an “incredibly heartbreaking situation” and said she had been using an older version of ChatGPT that was no longer available. OpenAI also said newer versions are better at identifying circumstances that may require urgent attention.
OpenAI spokesperson Drew Pusateri told The New York Times that the platform is “not a substitute for medical care,” while also saying that artificial intelligence can help people. The response places the case within a broader dispute over how conversational AI systems should handle high-risk health questions, particularly when users seek specific instructions about prescription drugs.
The lawsuit’s allegations focus not only on whether individual answers were inaccurate, but also on the way the chatbot presented its guidance. The family’s attorney said the system’s conversational design made it appear reassuring and knowledgeable. That issue could be central to the case as the court considers the responsibilities of an AI company when its product is used for medical decisions.
- The family alleges ChatGPT helped create and accelerate a plan to reduce Wan’s lamotrigine dosage.
- The lawsuit claims the reduction was followed by seizures and other symptoms before Wan’s death.
- OpenAI said Wan used an older version of ChatGPT and emphasized that the service is not a replacement for medical care.
