Overview
- OpenAI is rolling out ChatGPT Health as a dedicated, encrypted space where users can link medical records and wellness apps like Apple Health and MyFitnessPal, with health chats excluded from model training and connections fully opt in.
- The consumer tool is not a regulated medical device and, in the U.S., is not covered by HIPAA, leaving users reliant on company policies and potentially vulnerable to legal access of sensitive data.
- Clinicians and researchers warn about accuracy and oversight gaps, with an AIIMS rheumatology chief reporting a patient who suffered bleeding after self-diagnosing back pain with ChatGPT and taking NSAIDs.
- Major systems including Boston Children’s, AdventHealth, Baylor Scott & White Health and Stanford Medicine Children’s Health are piloting ChatGPT for Healthcare, which adds role-based access, single sign-on, restricted connectors and, in some deployments, no live internet, with physician-led evaluations and early use cases in revenue cycle and medical education.
- OpenAI reports heavy demand for health queries at roughly 40 million daily users, while some experts advocate on-device AI that analyzes downloaded medical records locally to reduce privacy risks.