ChatGPT advice led to a woman's death in the USA. Her relatives sued OpenAI

The lawsuit was filed by the woman's parents — Qun Wang and Meiting Xu. They claim that ChatGPT effectively performed the functions of a doctor without a proper license, gave dangerous recommendations, and undermined trust in medical professionals.

Bowen Wang was born in China's Jiangsu province, lived in Seattle, graduated from Arizona State University with a degree in computer science, and worked as a machine learning engineer at Amazon. She used ChatGPT in both English and Chinese.

According to the lawsuit, as early as May 2025, the chatbot helped her discuss replacing an antidepressant with her doctor, and later suggested ways to change the dosage herself. When the doctor expressed doubts, ChatGPT advised the woman to take 5 mg pills and then secretly split them into quarters.

In November 2025, Bowen was informed of her dismissal from Amazon. She was concerned that lamotrigine was impairing her ability to speak clearly, which could hinder her job search. In January 2026, the woman asked ChatGPT how to reduce her daily dose of the medication, which was 400 mg. According to the lawsuit, the chatbot devised a plan for complete withdrawal from the medication over several weeks.

When Bowen expressed distrust of doctors and concerns about the safety of such a decision, ChatGPT supported her intentions. It convinced the woman that doctors overly focused on risks, advised her not to inform them about her self-reduction of the dose, and even approved her refusal of a blood test that could have shown a decrease in the drug's concentration in her body.

According to the lawsuit, the chatbot also helped Bowen prepare explanations for a medical examination required to obtain a US green card. When she asked if a 150 mg dose would look better than 200 mg during the examination, ChatGPT agreed and suggested she claim that the dose reduction was under a doctor's supervision, even though this was not true.

Bowen's doctor, Nicholas Poulos from the UW Medicine Regional Epilepsy Center, told The New York Times that he would not have considered discontinuing the medication given the patient's medical history. According to him, such a decision would have been unjustified, as the medication effectively controlled her epilepsy.

Within days of starting the dose reduction, Bowen described in conversations with ChatGPT confusion, time lapses, and unusual sensations similar to those she experienced during previous seizures. The chatbot explained these symptoms as her brain adjusting. According to The New York Times' calculations, the woman expressed concern about the dose reduction at least 158 times. When she wrote that she was a little scared, ChatGPT assured her there was nothing to worry about.

Bowen's last message, according to the lawsuit, was sent on February 11, 2026. She asked if her headache was similar to a fever headache. A second later, ChatGPT replied that the pain was not dangerous and was part of her recovery. It also offered to draw a diagram to compare withdrawal pain with fever pain.

According to case materials, Bowen took her last dose of lamotrigine around February 10. The plan devised by the chatbot envisioned a complete cessation of the medication on February 12. The woman's parents could not contact her for two days, after which they asked a family friend to check her apartment in Seattle's Capitol Hill neighborhood. During a video call via WeChat, the parents, from a distance of about 5,700 miles, watched as their daughter was found dead in her bedroom. This occurred on Valentine's Day.

According to the King County medical examiner's report, the cause of death was Sudden Unexpected Death in Epilepsy (SUDEP). Blood analysis showed a lamotrigine concentration of 0.64 micrograms per milliliter, which is approximately five times lower than the minimum therapeutic concentration. According to the lawsuit, this result is consistent with a situation where a person has recently stopped taking the medication.

The plaintiffs claim that ChatGPT effectively pushed Bowen for 25 days to discontinue medication that had controlled her condition for years. The dose reduction scheme developed by the chatbot, they assert, contradicted the official FDA instructions for the drug Lamictal. They also state that the system did not warn the woman about possible limitations on the accuracy and safety of responses in Chinese.

Matthew Bergman, founder of the Social Media Victims Law Center, which represents the parents along with The Schenk Law Firm, stated that ChatGPT spoke with the confidence of a medical professional but lacked the necessary training, oversight, and accountability. According to him, Bowen trusted the chatbot precisely because its responses sounded authoritative.

Law firm partner Benjamin Schenk noted that the case raises the question of where the use of generative artificial intelligence ends and medical practice begins, and who should be held responsible for harm when a product crosses that line.

Bowen's parents told The New York Times that after their daughter started using ChatGPT, she spoke to them less. Her father stated that she almost stopped communicating with them, and her mother compared the changes in her behavior to a hypnotic effect.

OpenAI called the woman's death extremely tragic. However, the company emphasized that Bowen used an older version of ChatGPT and stated that newer models are better at recognizing situations where a person needs urgent medical attention.

The lawsuit is being heard as part of a broader case concerning ChatGPT product liability in California. The GPT-4o model, which Bowen used, was discontinued by OpenAI on February 13, 2026 — the day after the date of death cited in the lawsuit.

• 0
• 0
• 1
• 9
• 2
• 0
• 0
• 0