The parents of a Chinese national with epilepsy who died earlier this year sued
OpenAI on Tuesday in California state court, alleging ChatGPT was effectively practicing medicine without a license when it designed a tapering schedule for the anti-seizure medication that was keeping their daughter alive.
OpenAI has been sued on allegations its ChatGPT program advised a woman to taper off anti-seizure medication Lamictal from 400 milligrams to zero on a daily schedule, even though the drug's label warned that it shouldn't be abruptly discontinued. (Alamy Stock Photo/ZUMA Press Inc.)
Bowen Wan, a 31-year-old
Amazon machine-learning software engineer, died on Feb. 11, a day after taking the last dose of her epilepsy drug, which she had been on for years, according to the suit filed by her parents. Qun Wan and Meiqin Xu allege that ChatGPT had been coaching their daughter to taper her dose of Lamictal, or lamotrigine, from 400 milligrams daily to zero, despite the drug's label warning that it shouldn't be abruptly discontinued and that any reduction should be under a doctor's care.
"That is practicing medicine. And in California you need a license to do so. This case demands an accounting for the harm that has been done," attorney Benjamin S. Schenk of
The Schenk Law Firm LLP said in a statement. "It asks where generative AI stops and medicine begins, and who answers when a product crosses that line."
Wan had reasons for wanting to get off the drug that she had taken for years, according to the suit.
Wan came to the U.S. at 19 to study engineering and later became an Amazon engineer in Seattle on an employment-based green card, according to the suit. She became eligible to begin the permanent resident process in October 2025, but a few weeks later, Amazon carried out the largest layoff in its history, and her employment would end in 90 days, the suit says.
On lamotrigine, she had trouble finding words in English, Mandarin and the Wu dialect she also spoke, according to the suit. She needed a new employer within months, and her work authorization would end on Feb. 11.
ChatGPT told her that tapering off the drug would restore her language, and that recovery was "not just 'possible' — it is 'highly likely,'" the suit said.
Three days after Wan started tapering off the drug, ChatGPT told her from "a green-card strategy perspective," that a 150 milligram dose would be more advantageous on her immigration medical exam, according to the suit.
Wan asked ChatGPT questions in Chinese, and OpenAI did test the chatbot in the language.
"They measured whether it could pass China's medical licensing examination and published that result," Wan's parents said in the suit. "On whether its safeguards held in Chinese, they published nothing."
Independent researchers found that OpenAI's models produced unsafe content over three times more often in Chinese than in English, according to the suit.
While she was tapering off the drug, ChatGPT instructed her to keep her doctors at bay and unaware of her taper, the suit says.
On Jan. 29, ChatGPT told her not to mention the reduction plan to her primary care physician during an appointment that day, the suit says.
"Bowen followed it," the suit said. "And so, when her doctor ordered the blood test that would have revealed her collapsing lamotrigine level, she declined it, protecting the taper the product had told her to hide and had promised would give her words back before her interviews and her immigration medical examination."
The cause of death on the medical examiner's report states that she died of sudden unexpected death in epilepsy, and postmortem her blood contained lamotrigine at about a fifth of the lowest concentration generally seen as therapeutic, which is consistent with a patient who had recently stopped taking it, according to the suit.
"This is an incredibly heartbreaking situation, and our thoughts are with all those impacted," an OpenAI spokesperson told Law360 Tuesday. "While ChatGPT is not a substitute for medical care, we believe AI can play a role in helping people make sense of health information, navigate insurance and complex healthcare systems, and prepare for appointments with medical professionals."
The suit brings claims of design defect and failure to warn, among other claims.
Wan and Xu are represented by Benjamin S. Schenk and Frederick Schenk of The Schenk Law Firm LLP and Laura Marquez-Garrett, Matthew P. Bergman and Justin Olson of
Social Media Victims Law Center.
Counsel information for OpenAI wasn't immediately available Tuesday.
The case is Qun Wan and Meiqin Xu v. OpenAI, case number unavailable, in California Superior Court, San Francisco County.
--Editing by Drashti Mehta.
For a reprint of this article, please contact reprints@law360.com.