Source: Petrie-Flom Center (Harvard Law School blog)

OpenAI launched ChatGPT Health in January 2026 as a chatbot that leverages user-uploaded health records to help clarify medical concepts, interpret test results, prepare for appointments, and explore healthcare and wellness strategies. OpenAI’s marketing disclaims diagnostic intent by stating ChatGPT Health is "designed to support, not replace, medical care" and "is not intended for diagnosis or treatment." The piece highlights concrete harms from unregulated health chatbots: recent cases allegedly show chatbots confidently citing falsified medical studies or giving misleading medical advice nearly half the time, and such systems present a high risk of false negatives when used for diagnosis.

Because ChatGPT Health tailors outputs to individual uploaded records and user symptom descriptions, its apparent accuracy depends on correct user input and is especially likely to mislead unfamiliar or socioeconomically disadvantaged users. This misleading reliance can translate into delayed treatment and increased medical costs. Most health chatbots are tied to regulated payers or providers or are otherwise subject to some regulation, but ChatGPT Health operates separately without explicit guardrails. Under Section 201(h) of the Federal Food, Drug, and Cosmetic Act, a product intended for diagnosis, cure, mitigation, treatment, or prevention of disease qualifies as a medical device subject to FDA jurisdiction; the FDA determines "intended use" by examining promotional materials and any "other relevant source," including actual customer use, design, composition, and distribution. If at least one ChatGPT Health function is deemed intended for diagnosis or prevention, the FDA may regulate it as a medical device, and the authors argue the absence of restrictive design features makes such regulation appropriate. The dossier also notes legislative uncertainty: Congress has not passed comprehensive AI legislation, over 200 state statutes across 48 states address AI in narrow ways, and an executive order dated Dec. 11, 2025 authorized the Department of Justice to challenge problematic state AI laws — all framing why the authors call on the FDA to intervene.