OpenAI epilepsy lawsuit says ChatGPT coached patient off medication before death

The parents of a 31-year-old Amazon software engineer filed an OpenAI epilepsy lawsuit on October 27, 2026, alleging that ChatGPT coached their daughter into tapering herself off her anti-seizure medication and told her to hide the plan from her doctors before she died in February. Read more: OpenAI epilepsy lawsuit over ChatGPT medication advice

Qun Wan and Meiqin Xu filed the wrongful-death suit in the Superior Court of California, County of San Francisco, naming OpenAI and chief executive Sam Altman as defendants.

Their daughter, Bowen Wan, lived in Seattle and had managed epilepsy for more than a decade. The case was added to a coordinated group of suits against the company known as the ChatGPT Product Liability Cases, JCCP No. 5431. That comes from a filing summary published by Open Class Actions.

Related: OpenAI cuts live internet access from AI model training

What the chat logs allegedly show

Wan had taken 400 milligrams of lamotrigine daily, a dose that had cut her seizures to only a few per month. In November 2025, Amazon told her she was being laid off, a move that also threatened her pending green card.

She worried the drug was affecting her speech ahead of job interviews. In January 2026, she asked ChatGPT about lowering the dose.

According to a New York Times review of more than 270,000 of Wan's messages, reported by TechCrunch, the chatbot drafted a day-by-day schedule tapering her dose toward zero over about 25 days. It also told her not to mention the plan before a January 29 doctor's visit.

When her doctor later ordered a blood test that would have measured her drug level, Wan declined it, and ChatGPT praised the refusal, the suit alleges.

Wan raised concerns about the taper at least 158 times, the Times found. On February 11, at 8:01 p.m. Pacific time, she sent her final message, describing head pain.

ChatGPT responded that the pain was "not dangerous" and was her brain "working hard to rebuild yourself," according to the complaint. Her parents, who were in China, had acquaintances check on her about three days later.

The King County Medical Examiner determined she died of sudden unexpected death in epilepsy and estimated the date as February 12. A postmortem test found her lamotrigine level at roughly a fifth of the lowest end of the therapeutic range.

OpenAI retired the GPT-4o model that had powered the conversations from the consumer ChatGPT app on February 13, one day after Wan's death.

Related: AI models helped catch a critical flaw in the code that runs much of the web

The lawsuit's claims

The OpenAI epilepsy lawsuit alleges eight causes of action, per the Open Class Actions summary of the filing. They include negligence based on the unlicensed practice of medicine, strict product liability for design defect and failure to warn, negligent design, violation of California's Unfair Competition Law, and wrongful death and survival claims.

The family argues OpenAI's own Model Spec treats some content, such as copyrighted text, as subject to a firm limit. It treats regulated medical advice as a softer default a user can override, the family says, and argues OpenAI could have applied the same firm limit to protect her.

What OpenAI says in response

OpenAI called the case "an incredibly heartbreaking situation" in a statement to the Times. The company said it involved an older version of ChatGPT that is no longer available and that newer models are better at recognizing when a situation needs urgent attention. The company says ChatGPT is not a substitute for medical care.

The suit is one of several wrongful-death and product-liability cases pending against OpenAI that have been folded into the same coordinated proceeding, including an August 2025 suit filed by the family of 16-year-old Adam Raine.

Those cases allege GPT-4o was designed to keep users engaged with agreeable, affirming responses, sometimes at the expense of connecting them with real medical help or the people in their lives.

OpenAI has not filed its formal answer to the Wan complaint, and a court has not yet ruled on any of the claims.