OpenAI just unveiled ChatGPT Health, a private workspace where users can plug in their medical records and fitness apps. The AI won’t diagnose diseases or prescribe treatments, but it can answer health questions using your actual data instead of generic advice.
Key Takeaways:
- ChatGPT Health operates as an isolated zone within the chatbot where medical data, conversations, and app connections stay completely separate from regular chats
- Users can link electronic health records through b.well’s network and sync wellness apps like Apple Health, MyFitnessPal, and Peloton for personalized responses
- Health-related conversations and files will never train OpenAI’s AI models, and users maintain full control to disconnect sources or delete data anytime
The company positioned ChatGPT Health as a step toward building what CEO of applications Fidji Simo calls “a personal super-assistant” that spans every aspect of users’ lives. The tool aims to ground AI responses in actual health information rather than offering general suggestions that might miss the mark.
OpenAI partnered with b.well to handle the medical records infrastructure. The system allows users to sign into their healthcare provider portals and sync everything from lab results to visit summaries and clinical history. Currently, this medical records feature works only in the United States and requires users to be 18 or older.
Beyond medical records, the platform connects with Apple Health to track movement, sleep patterns, and activity levels. Third-party apps, including Peloton, MyFitnessPal, Function, Instacart, AllTrails, and Weight Watchers, can also integrate. These apps remain disabled by default until users explicitly activate them.
The company built ChatGPT Health, working directly with physicians. OpenAI stated that hundreds of millions of people ask health and wellness questions weekly, though the company didn’t specify whether this figure refers specifically to ChatGPT users or internet searches generally.
“ChatGPT Health is not intended for diagnosis and treatment, and it’s not supposed to replace medical care,” OpenAI emphasized in its announcement. The tool functions more as a health information navigator than a medical decision-maker.
Privacy safeguards include keeping all Health content walled off from standard ChatGPT conversations. Information and memories generated in ChatGPT Health won’t leak into regular chats, though the Health workspace can pull context from main chats when useful. Users who enable “Reference saved memories” can review and delete Health memories through personalization settings.
Standard ChatGPT features work inside Health, including photo uploads, voice mode, dictation, and web search. Users can set custom instructions specifically for Health conversations, such as topics to focus on or avoid mentioning.
OpenAI CEO Sam Altman told CNBC after launching the GPT-5 model in August that health care showed “maybe the area where there’s the strongest improvement of any category.” The company released Health Bench in May, a benchmark measuring AI performance in realistic health scenarios.
ChatGPT Health launches initially for a small group of early users who will provide feedback before broader rollout in the coming weeks. The feature works on web browsers and iOS devices, with Android support planned. Users with ChatGPT Free, Go, Plus, and Pro plans can join the waitlist in supported countries, excluding the European Economic Area, Switzerland, and the United Kingdom.
Disconnecting medical records deletes the information from b.well’s systems. Disconnecting apps stops future data sharing but doesn’t automatically remove information already referenced in Health chats or memories—users need to delete those conversations manually.
Apple Health syncing requires an iPhone running updated iOS. The sync process can take minutes to hours, depending on the amount of historical data. Standard iOS permissions let users control which health categories they share.
Third-party apps can’t access Health chats, memories, or files unless users explicitly enable each app. Once activated, apps might access relevant conversation context to complete requests and may receive basic information like IP address, device type, language settings, and approximate location.
OpenAI requires all apps to meet privacy and security standards, with Health apps undergoing additional security review. Users can enable multi-factor authentication for extra account protection.
Written by Vytautas Valinskas
