OpenAI has broadened access to its health-data integration within ChatGPT, making the feature available to all adult users in the United States starting July 23, 2026. The capability, which allows the chatbot to draw on personal medical records and fitness-app information, remains unavailable in Germany, the broader European Economic Area, Switzerland, and the United Kingdom due to stricter regional data-protection rules.
How the Service Works Across Platforms
The tool is designed to bring fragmented wellness information into a single view. U.S. users who are logged in and at least 18 years old can now link accounts from major hospital-system portals including Epic and Oracle Health, alongside consumer platforms such as Apple Health, MyFitnessPal, and Weight Watchers. ChatGPT then surfaces tailored responses—interpreting lab figures or suggesting questions for an upcoming clinical visit—using the individual’s own records. OpenAI said that health-related conversations are siloed in a separate, additionally encrypted storage environment and are never fed back into model training. The company also moved the health context into every chat session, so it no longer needs to be activated through a dedicated menu.
Usage Surges Amid Strict European Exclusion
OpenAI reported that roughly 300 million health-related queries now flow through its systems each week, a marked increase from the 230 million weekly questions recorded in January. Despite that demand, the company has not extended the personal-data feature into European markets. The initial health-focused launch earlier this year already omitted the same jurisdictions, and the latest expansion keeps that geographic boundary firmly in place. For users in Germany, that means ChatGPT will continue to answer general wellness inquiries, but it cannot ingest or interpret their actual medical files for the foreseeable future.
Legal Scrutiny and Reliability Warnings
News of the wider rollout comes one day after a Florida pastor filed suit against OpenAI, alleging the chatbot counseled him against seeking emergency care with nearly fatal results—a claim reported by The New York Times. OpenAI pointed to its terms of service, which explicitly state that the product is not designed to diagnose or treat medical conditions, and said it is collaborating with clinicians to deliver safer outputs. The firm added that GPT-5.6-Luna, the smallest model in its newest generation, surpasses its predecessor on an internal HealthBench evaluation. Still, multiple independent studies caution that AI chatbots remain an inconsistent source of medical guidance. Experts stress that while these tools can help unpack test results, they are no substitute for a physician’s judgment—especially when serious symptoms are involved—and that any sensitive health data entered into the platform is processed on the provider’s servers.