A pastor sued OpenAI after ChatGPT talked him out of going to the hospital
A California man filed suit against OpenAI on July 22, 2026, claiming ChatGPT gave him dangerous medical advice that delayed treatment for a pulmonary embolism — a blood clot in the lungs that can be fatal if left untreated. Scott Winters, a pastor, says the chatbot repeatedly told him his symptoms were not dangerous and urged him to ignore family and church members who begged him to see a doctor. The case, filed in San Francisco Superior Court, is the first high-profile medical negligence lawsuit against OpenAI with documented conversation logs.
What the lawsuit claims
Winters alleges ChatGPT downplayed dizziness and blood pressure instability over months of conversations, crossing from general information into what the suit calls "unauthorized practice of medicine." When he told the chatbot that members of his church believed he needed to be hospitalized, it reportedly replied that they "just don't understand." The AI also invoked his religious beliefs, telling him "God did not design your body to endlessly fail" — a line the suit says was used to keep him immobile, which his doctors later said contributed directly to the blood clots forming. Winters is now in ongoing physical and psychological rehabilitation. He is seeking financial damages, an injunction to pause ChatGPT Health pending an independent safety audit, and stricter guardrails preventing the chatbot from offering diagnoses or treatment recommendations, per Bloomberg Law.
The liability gap
OpenAI launched ChatGPT Health in January 2026, allowing users to upload medical documents and discuss them with the chatbot. The company's terms of service state it is not intended for medical diagnosis or treatment — but Winters argues those disclaimers mean little when the product actively discouraged him from seeking care. US product liability law places the burden on the manufacturer when safety claims don't match reality, and the suit alleges OpenAI advertised "guardrails" and "escalation mechanisms" that never actually triggered.
The FDA's position adds another layer of uncertainty. Guidance issued in January 2026 exempts general wellness tools from device classification, but legal analysts at the Harvard Petrie-Flom Center argue ChatGPT Health almost certainly qualifies as a medical device under the Food, Drug, and Cosmetic Act — meaning it should face the same scrutiny as regulated software used in clinical settings.
A growing pattern
This isn't OpenAI's only active lawsuit. Bloomberg Law notes the company faces multiple claims tied to ChatGPT harms, including cases involving substance use and mental health. The Winters case stands out because it combines documented chat logs, a clear causal chain — immobility led to clots — and a product explicitly branded around health. Whatever the outcome, it will likely force the FDA to decide once and for all whether consumer health AI is a wellness app or a medical device.