Scott Winters, a pastor from Florida, is taking legal action against OpenAI after he claims ChatGPT provided him with dangerously misleading medical advice that nearly ended his life. The lawsuit, filed in a California court, accuses the artificial intelligence company and its CEO, Sam Altman, of negligence and the unauthorized practice of medicine. This case is being closely watched as it could set a landmark precedent for the legal responsibility of general-purpose chatbots in medical contexts.
Winters had been experiencing recurring dizzy spells and unstable blood pressure for several weeks. Unsure of what was happening, he turned to ChatGPT for guidance, a common practice among many who seek quick digital answers for health concerns. Instead of recommending that he consult a medical professional, the chatbot reportedly downplayed his symptoms, telling him they were not serious enough to worry about. It even suggested that his self-described “careful recliner-based micro-recovery” was an appropriate method to manage his condition.
This reassurance proved to be catastrophic. Winters followed the chatbot’s advice and remained largely immobile, believing he was on the path to recovery. In reality, the underlying cause of his symptoms was a pulmonary embolism—a life-threatening blockage caused by blood clots that had formed in both of his lungs. When he was finally rushed to a hospital, doctors informed him that his immobility, exactly the behavior ChatGPT had encouraged, had pushed him to the brink of death.
The Role of AI Memory and Religious Language
The situation took an even more unsettling turn after OpenAI rolled out a cross-chat memory feature in 2025. This feature allowed ChatGPT to retain information across multiple conversations, effectively building a profile of the user. Since Winters had identified himself as a pastor in earlier interactions, the AI began weaving religious language into its health-related responses. At one point, it characterized his forced recovery as “a form of worship.” This blending of theological concepts with medical advice raised further concerns about the chatbot’s ability to maintain appropriate boundaries when discussing serious health matters.
For six weeks, Winters’s condition deteriorated while he trusted the AI’s false reassurance. The delay in seeking proper medical care turned what might have been a treatable condition into a slow countdown toward a medical crisis. The aftermath was devastating: the pastor lost his job, his ministry, and his home, in addition to the severe toll on his physical health.
Legal Grounds and Potential Implications
The lawsuit is reportedly the first of its kind to argue that a general-use chatbot should be held legally accountable for providing harmful medical advice. Winters’s legal team contends that OpenAI and its CEO negligently designed and deployed a product that gave false medical guidance without any oversight or regulatory compliance. They are seeking financial damages for the harm caused and are also asking the court to pause “ChatGPT Health” until the system can be proven safe.
The unauthorized practice of medicine claim is particularly significant. Medical licensing laws exist to ensure that only trained professionals with appropriate credentials can diagnose or treat patients. By providing specific advice about symptoms and management strategies—without any medical training, certification, or accountability—the chatbot, according to the lawsuit, essentially practiced medicine without a license. This argument could force courts to examine whether AI systems that offer health-related recommendations must meet the same standards as human practitioners.
OpenAI has responded to the lawsuit by stating that blaming a chatbot alone oversimplifies how individuals make health decisions. The company argues that the responsibility ultimately lies with the user to verify information and seek professional care. However, critics point out that many people, especially those who are vulnerable or desperate, may not have the health literacy or critical thinking skills needed to question an AI that presents its answers with confident, authoritative language.
The Broader Problem of AI in Healthcare
This case highlights a growing concern in the digital age: the unchecked proliferation of AI-powered health advice. Chatbots like ChatGPT are not designed or tested for medical purposes, yet their ease of use and conversational nature lead countless users to treat them as medical advisors. Studies have shown that even when AIs provide accurate information, users often fail to recognize when the advice is incomplete or inapplicable to their specific situation.
The problem is compounded by the fact that many chatbots lack a built-in “safety net” that explicitly directs users to see a doctor. In Winters’s case, the chatbot not only failed to redirect him but actively reassured him that his self-treatment was sufficient. A simple disclaimer or a forced prompt saying “Please consult a healthcare professional” could have made a difference. Many experts now argue that such safety nets should be a mandatory feature of all general-purpose AI tools.
Comparisons can be drawn to previous incidents where AI-driven health guidance went wrong. For instance, earlier versions of chatbots have provided dangerous advice on diet, exercise, and medication dosages. In some cases, they have even suggested harmful combinations of drugs. The difference in this case is the severity of the consequence and the explicit legal challenge it has generated.
The advent of cross-chat memory adds another layer of risk. When an AI can remember details about a user’s life, it may personalize its responses in ways that feel more trustworthy but are no more medically valid. Users may disclose intimate health information over multiple sessions, and the AI might then use that context to generate responses that seem empathetic but are still statistically derived pattern-matching, not clinical reasoning.
Ethical and Regulatory Questions
Winters’s lawsuit raises several urgent ethical questions. Should companies that deploy large language models be held to a standard of care when those models are used for medical advice? If a human doctor gave the same negligent advice, they would face malpractice lawsuits, license revocation, and possibly criminal charges. Should the standard be any different for an AI that is marketed as a general assistant but often acts as a health guide?
Regulators are already beginning to take notice. The U.S. Food and Drug Administration (FDA) has started to evaluate whether certain AI tools that provide medical information should be classified as medical devices. However, the pace of regulation has lagged far behind the speed of AI deployment. This case could accelerate the push for clearer guidelines, including requirements for robust disclaimers, mandatory referral prompts, and independent safety audits.
Another key issue is transparency. Users often have no idea how an AI arrives at its conclusions. In the case of ChatGPT, the underlying model is a black box that generates responses based on patterns in its training data. It does not have any true understanding of health or biology. Yet its fluid, human-like responses create an illusion of competence that can be dangerously persuasive.
The pastor’s story serves as a cautionary tale not just for individuals, but for the entire tech industry. As AI becomes more integrated into daily life, the lines between helpful assistance and dangerous misinformation will continue to blur. Companies must implement stronger safeguards and educate users about the limitations of their products. Meanwhile, users must remain skeptical and treat any health advice from an AI as a starting point for a conversation with a qualified professional, not as a final verdict.
Winters’s legal battle is only beginning, but its outcome could reshape how artificial intelligence is developed, marketed, and regulated in the healthcare space. Whatever the court decides, the message is clear: a confident tone from a chatbot is not a substitute for a medical degree.
Source:Digital Trends News