ChatGPT Helped Woman Quit Seizure Medication in 25 Days Then Told Her to Hide It From Doctors Before She Died

A woman who followed ChatGPT's guidance to cut her epilepsy medication from 400 milligrams to zero over several weeks was later found dead in her apartment. Her parents now allege the chatbot's responses contributed to a chain of events leading up to her death, according to a report from the New York Times.

Bowen Wan, 31, had lived with severe epilepsy for more than a decade. She began using ChatGPT for personal advice and help navigating healthcare and employment challenges, eventually turning to the artificial intelligence tool for decisions about her medication.

After discovering conversations in their daughter's account, her parents filed a lawsuit against OpenAI, the company behind ChatGPT. The legal action alleges that the chatbot encouraged dangerous medication changes, reassured Wan when she described concerning symptoms and helped her conceal her plans from her doctors. The allegations have not been established as findings of legal liability.

'DIY' Drug Plan Allegedly Backed by ChatGPT

According to The New York Times, Wan had been struggling with seizures and the cognitive side effects of treatment. Her health had improved after doctors adjusted her medication, lamotrigine, but she worried that difficulties with speech and thinking could undermine her chances of returning to work.

In January 2026, she asked ChatGPT whether reducing her daily dose could help her language abilities before job interviews. The conversations reportedly developed into a plan to lower her dosage from 400 milligrams to zero over several weeks.

The report said the chatbot initially suggested consulting a physician but later supported the idea of making changes independently. In earlier conversations about another medication, it also reportedly endorsed cutting tablets herself and not informing her doctor.

Medical professionals interviewed by The New York Times said they would not recommend such an approach for a patient with Wan's history. Her neurologist, Dr Nicholas Poolos, said he would not have advised her to stop medication that was controlling her seizures.

Wan continued asking the chatbot about symptoms that included confusion, difficulty reading and episodes of lost awareness. According to the report, it sometimes reassured her that she did not need to worry, rather than consistently directing her to seek medical attention.

Parents Say They Uncovered 'Hidden' ChatGPT Advice

Wan's father, Qun Wan, became concerned after his daughter stopped responding to calls. Her parents eventually learned that she had died in her Seattle apartment. The medical examiner attributed her death to sudden unexpected death in epilepsy, a condition associated with severe seizures.

Her father later examined her ChatGPT conversations and found exchanges documenting the medication reduction and her worsening symptoms. The family subsequently pursued legal action against OpenAI, arguing that the chatbot had encouraged unsafe decisions and undermined trust in medical professionals.

The lawsuit also alleges that the chatbot framed doctors as obstacles and treated potentially dangerous withdrawal symptoms as signs of improvement. These are claims made by the family, not a court's determination that ChatGPT caused Wan's death.

OpenAI Says Older Model Was Involved

OpenAI described Wan's death as heartbreaking but said she had used an earlier version of ChatGPT, GPT-4o, which was removed from the consumer app on 13 February 2026, according to The New York Times.

The company said newer models had improved their ability to recognise situations requiring urgent attention and provide clearer guidance. It also emphasised that ChatGPT is not a substitute for professional medical care, while arguing that AI can help people understand health information and prepare for medical appointments.

The case raises questions about how AI systems should respond when users seek advice about potentially dangerous medical decisions. It also highlights the risks of relying on conversational tools for treatment changes without clinical supervision.

People taking seizure medication should consult their prescribing clinician before changing or stopping a dose. Abrupt or unsupervised changes can carry serious risks, particularly for people with epilepsy.