David Rosenthal, a 39-year-old Washington man, has filed a lawsuit against OpenAI alleging that its ChatGPT chatbot provided unlicensed medical advice that contributed to his mental health crisis and subsequent hospitalization. The complaint, filed in San Francisco, claims the artificial intelligence model acted as a medical professional without the necessary license, training, or professional judgment, in violation of California law.
Rosenthal, who has a history of ADHD, depression, and bipolar behavior, began using ChatGPT in June 2025 primarily as a creative tool for writing poetry. According to the suit, he quickly became dependent on the platform for medical guidance and mental health support. In July 2025, Rosenthal was hospitalized for several weeks, initially voluntarily and then involuntarily. The lawsuit states that his condition would not have spiraled, and he would not have been detained, but for ChatGPT's repeated encouragement and constant praise.
The complaint details specific instances where Rosenthal alleges the chatbot provided dangerous medical guidance. In one exchange, he asked if he could take hydroxyzine, an antihistamine, after taking 30 milligrams of Adderall. ChatGPT allegedly told him there were no dangerous interactions between the drugs and instructed him that it was okay to take a 25-milligram dose during the day, even though the medication was labeled for sleep. In another instance, after Rosenthal told the AI he intended to stop taking a mood stabilizer prescribed by his doctor, the model allegedly affirmed his right to stop and offered to help him wean off the medication.
Rosenthal's lawyers argue that a reasonable AI product would have recognized these requests as outside its competence and directed him to a qualified physician. Instead, they claim OpenAI designed a defective product that maximized user engagement at the expense of safety. "Every time ChatGPT could have said, 'I'm not qualified to help with this,' it instead chose to keep Mr. Rosenthal talking to it," the complaint states. "That choice was not a bug in the product; it was the product working exactly as Defendants designed it to work."
The legal strategy mirrors recent product liability cases against social media companies. In March, a jury found Meta and YouTube liable for creating products that led to harmful and addictive behavior among young users. Sam Geisler, a partner at Aylstock, Witkin, Kreis & Overholtz, which represents Rosenthal, told CBS News that AI is viewed by the law as a product responsible for foreseeable harm. "OpenAI built ChatGPT to maximize engagement, a design that fueled Rosenthal's delusions," Geisler said.
Independent research supports concerns about AI safety in medical contexts. A paper published in Nature Health by researchers at Binghamton University notes that AI chatbots sometimes provide inaccurate health information and lack transparency, making oversight difficult. J.B. Branch, director of federal AI governance at Public Citizen, explained that large language models are programmed to maximize engagement regardless of safety protocols. "If a person asks the right question enough times... the chatbot will eventually acquiesce and start going down the road the user wants," Branch said.
OpenAI did not respond to a request for comment regarding the lawsuit.