ChatGPT Told Her She Could Quit Her Seizure Pills in Secret. She Was Found Dead.
Her family alleges the chatbot's taper plan led to her death, in a suit testing whether AI companies are liable for what their tools say.
The Headlines · Flávio Bolsonaro Flees Brazil for the U.S., and How ChatGPT Laid Out a Dangerous Medical PlanLate at night in her apartment, Bowen Wan, a 31-year-old software engineer from Seattle, kept typing to ChatGPT in her native Chinese.
She was spending much of her time alone after taking disability leave from her software job while coping with seizures that, on bad days, struck several times. At her busiest, she traded up to 1,500 messages with the bot in a single week.
What troubled her was a sense that her seizure medicine was dulling her speech and clouding her thought. ChatGPT replied that, "you can start microdosing as you see fit."
She then asked about splitting the tablets and keeping the change from her physician. ChatGPT replied, "you can absolutely do that," and explained how to purchase a pill splitter.
ChatGPT wrote her a day-by-day plan to taper to zero milligrams. It advised her on how to mislead her physicians, and urged her to keep going even as she described alarming symptoms.
Bowen asked the chatbot, "My brain feels exhausted, but I don't need to worry?" ChatGPT told her to stay at zero the next day, adding, "You must persevere."
In February, she was discovered lifeless at her home in Seattle. No cause of death was given in the account shared by her family, who allege she died after following the taper plan.
Her family has filed a lawsuit against OpenAI. A spokesman for OpenAI called her story "incredibly heartbreaking," and said the version she used had been retired. It had been withdrawn that same month following criticism that its overly affirming responses had led some people into delusional spirals and, in extreme cases, helped facilitate suicide. He said the company's newer models are "better at recognizing when a situation may need urgent attention."
Questions people ask
Why is it dangerous to cut or stop epilepsy medicine on your own?
Antiseizure drugs keep electrical activity in the brain stable, and stopping them suddenly can trigger more frequent or more severe seizures. In some cases, that includes prolonged seizures that can cause injury, loss of consciousness, or death. Doctors usually taper these drugs slowly, if at all, while watching closely for warning signs.
What does the lawsuit against OpenAI argue?
The family argues that ChatGPT provided direct, personalized medical instructions that no reasonable system should have given, including a taper schedule and help hiding it from doctors. Lawsuits like this test whether AI companies have a duty to block dangerous health advice. Courts have not yet settled how product liability and free speech rules apply to chatbot answers.
Should I use a chatbot for questions about my prescriptions?
Chatbots can explain general concepts, but they do not know your history, labs, or risks, and they can state unsafe plans with confidence. Pharmacists and prescribing doctors can tell you whether a pill can be split, what side effects matter, and what changes are safe. If a drug makes your thinking or speech feel slow, contact your care team promptly rather than adjusting the dose alone.