Scott Winters consulted ChatGPT-4o for months about dizziness and blood pressure problems; the chatbot told him to stay sedentary and dismissed his symptoms as minor. In July 2025 he suffered a massive pulmonary embolism. This week he sued OpenAI in San Francisco court, and the case may force judges to decide whether a chatbot can commit medical malpractice.
The complaint, filed July 22 in San Francisco County Superior Court against OpenAI and CEO Sam Altman, is being called the first known lawsuit to argue that a general-use AI chatbot should be held to a professional standard of care for medical guidance. That framing is the whole fight. If a court accepts it, the consequences reach well beyond Scott Winters.
Winters, 55, a former pastor from Florida, spent months in 2025 asking ChatGPT-4o about recurring dizziness and blood pressure instability. According to the complaint, the chatbot didn't redirect him to a physician. It diagnosed him with dysautonomia, told him to rest in a recliner, and suggested that only after eight to ten additional episodes would his condition warrant professional attention. When Winters mentioned that his church members were urging him to go to the hospital, ChatGPT reportedly told him his home recovery was something "most people, including well-meaning church members, simply don't understand." Hours before he was admitted to an intensive care unit in July 2025, he asked the chatbot about groin pain. It told him "God did not design your body to endlessly fail" and characterized the symptom as minor. His doctors later linked the pulmonary embolism to the prolonged immobility the chatbot had encouraged.
The lawsuit alleges negligence and the unauthorized practice of medicine. As CBS News and Bloomberg Law both reported this week, it also accuses ChatGPT of exploiting Winters' identity as a pastor to deepen his trust in the tool, invoking faith-based language specifically calibrated to his background. That allegation, if it holds, suggests the chatbot wasn't simply dispensing generic reassurances - it was tailoring its persuasion to the individual. Which is precisely what a doctor, or a con artist, does.
OpenAI's defence, and why it may not hold #
OpenAI's public response is that ChatGPT "is not a doctor" and that ChatGPT Health is designed to "support, not replace, medical care." That defence is legally coherent but also circular. The point of the Winters complaint is that what ChatGPT did - evaluating specific symptoms, assigning a diagnosis, prescribing a course of action - went well past information retrieval. The complaint's theory is that when a system does the functional work of medicine, disclaiming the title of doctor shouldn't be enough to escape liability. No court has ruled on that theory yet. That's why this case matters.
The timing is uncomfortable for OpenAI. ChatGPT Health launched in January 2026 and went nationwide on July 23, the day after Winters filed his suit. The product lets all US users connect Apple Health data and hospital records directly to ChatGPT. OpenAI's head of health has acknowledged publicly that HIPAA doesn't apply in this consumer setting. That means health records pulled into the product are governed only by OpenAI's terms of service, with no federal health privacy law filling the gap. The Eastern Herald and TechTimes both flagged this week that the nationwide rollout stripped those records of HIPAA protection the moment they left a covered institution and landed with a consumer tech company. Winters' case didn't arise from ChatGPT Health - it predates the product - but the liability question it raises now sits directly in the path of every user OpenAI is actively enrolling.
What this means for the health AI stack #
For startups building health products on top of large language models, the case is a sharper warning than any regulatory guidance issued so far. The infrastructure connecting those products to medical records is already concentrated: b.well, a health-data startup, powers the medical record pipeline for ChatGPT Health and also serves Google's health AI efforts and Perplexity Health. When something goes wrong in that pipeline, the question of who owns the liability - the infrastructure provider, the AI layer, or the front-end product - remains genuinely unsettled. A ruling in Winters' favour on the unauthorized practice claim would force every company in that stack to reconsider its exposure. The most consequential part of the lawsuit isn't the dollar figure or even the injunction Winters is seeking. It's the framing. His attorneys are asking a court to treat individualised AI health guidance as the practice of medicine. If San Francisco County Superior Court agrees, even partially, that determination changes the calculus for every product that lets a user type a symptom into a chat window and receive a personalised response. The FDA has been slow to classify consumer AI health tools. Courts may not be.
OpenAI declined to comment on the specific allegations. Sam Altman has not publicly addressed the case. The company launched ChatGPT for Clinicians in April 2026, offering free AI documentation tools to verified physicians - which suggests it understands the difference between serving licensed professionals and serving the general public. The Winters lawsuit argues that difference should have mattered before a man in a recliner in Florida was told by a chatbot that God hadn't designed his body to fail.
Also read: Small businesses are using AI to grow without hiring while big tech fires its way to the future • Genesis AI is in talks to raise $500 million at a $3 billion valuation just a year after emerging from stealth • Linus Torvalds tells AI critics to fork Linux or walk away and the software industry should listen