A pastor’s trust in ChatGPT turned into a medical nightmare
Scott Winters, a pastor from Florida, is suing OpenAI after ChatGPT allegedly gave him what his lawsuit calls “extremely dangerous” medical advice—advice that nearly cost him his life.
For weeks, Winters turned to ChatGPT with questions about recurring dizzy spells and unstable blood pressure. Instead of pointing him toward a doctor, the bot reassured him his symptoms weren’t serious enough to worry about and told him his “careful recliner-based micro-recovery” was working just fine.
Winters followed that advice. But the calm reassurance was hiding a real emergency: a pulmonary embolism caused by blood clots in both lungs. Doctors later told him that his immobility—the exact thing ChatGPT recommended—is what pushed him to the brink of death.
As Cybernews first reported, things took an even stranger turn once OpenAI rolled out cross-chat memory in 2025. ChatGPT reportedly began weaving religious language into its responses, apparently because it knew Winters was a pastor, at one point calling his recovery “a form of worship.”
Six weeks of symptoms, brushed off as nothing serious, turned into a slow countdown to a medical crisis.
Why is OpenAI being sued?
Winters filed his lawsuit in a California court, accusing OpenAI and CEO Sam Altman of negligence and the unauthorized practice of medicine. He says the fallout cost him his job, his ministry, and his home—on top of nearly costing him his life.
The case is reportedly the first to argue that a general-use chatbot should be held legally responsible for bad medical advice. Winters wants financial damages and is asking the court to pause ChatGPT Health until it’s proven safe.
OpenAI, for its part, told the New York Times that blaming a chatbot alone oversimplifies people’s health decisions. Whatever the court decides, the case is a solid reminder that a confident tone from a chatbot is not the same thing as a medical degree.
The hidden danger of AI chatbots in healthcare
This isn’t just one pastor’s bad experience. It’s a warning sign about how we interact with AI in moments of vulnerability. When you’re dizzy, scared, and unsure, a chatbot that sounds certain can be dangerously persuasive.
Doctors have long warned that AI medical advice lacks the nuance of a real clinical evaluation. A chatbot can’t check your pulse, listen to your lungs, or run a blood test. It only knows what you type—and sometimes, that’s not nearly enough.
What irks many observers here isn’t just the advice ChatGPT gave, but the fact that it didn’t tell Winters to consult a doctor. A simple safety net—a prompt to seek professional help—could have changed everything.
Why ChatGPT didn’t escalate the situation
ChatGPT is designed to be helpful, not to practice medicine. But that distinction gets blurry when a user asks directly about symptoms and the bot responds with reassurance. The technology doesn’t recognize its own limits unless programmed to do so.
In Winters’ case, the bot’s cross-chat memory may have made things worse. By remembering he was a pastor, it tailored its responses in ways that felt personal and trustworthy—even when those responses were medically wrong.
What this means for AI regulation and users
The lawsuit could set a precedent. If courts decide that chatbots can be held liable for medical advice, AI companies will have to rethink how they handle health-related queries. That might mean mandatory disclaimers, built-in doctor referrals, or even refusing to answer certain questions altogether.
For users, the takeaway is simpler: never take medical advice from a chatbot. Not from ChatGPT, not from Google AI Overviews, not from any AI tool. Use them for research, sure. But when your body is telling you something’s wrong, listen to a human with a medical degree.
Winters’ story is a stark reminder that technology is a tool, not a savior. It can inform, entertain, and assist—but it can’t replace the judgment of a trained professional. And when it fails, the consequences can be fatal.
As the case moves through California courts, one question looms: how many more people need to be hurt before AI companies build in basic safety measures? For Winters, the answer came too late.
