ChatGPT Health already lets eligible users connect records. The September 28 change makes one interaction quicker: select a chart, metric, or record and ask for an explanation without writing a prompt.
OpenAI has added personalized summaries in the Health tab. Its release notes say users can select a chart, metric, or record to receive an explanation based on connected health information. This updates Health in ChatGPT, not the launch of the broader Health product. OpenAI says the feature is intended to help people understand their information and is not for diagnosis or treatment.
What changed in the Health tab
Previously, a user could ask a question about connected information. Now the Health tab can start from a selected item. OpenAI gives the example of choosing a lab result to see an explanation and how the result changed over time. The response may include an interactive chart. That makes it easier to explore a record, but the selection does not verify that the underlying information is complete, current, or clinically interpreted correctly.
Health can use information a person chooses to connect, including Apple Health and supported U.S. medical records. It can also work with supported sources such as One Medical and Function Health. Availability differs by provider and app. A wearable metric may not transfer through Apple Health, and an app-specific score may be defined differently from the raw measurement shown in ChatGPT.
Who can use it
According to OpenAI’s Health page, the experience is rolling out to logged-in U.S. adults aged 18 or older on Free, Go, Plus, and Pro plans. It supports web and iOS; an Apple Health connection requires an iPhone. Some eligible users will not see the feature yet because the rollout is gradual.
Check the source before acting on a summary.
- Open the original record. Check the date, unit, reference range, and provider note when available.
- Look for missing context. A trend can reflect a changed device, lab, medication, or measurement method rather than a change in health.
- Use the explanation to prepare questions. If a result may affect care, discuss it with a qualified clinician rather than changing treatment from a generated summary.
- Seek urgent care for urgent symptoms. A chatbot summary is not an emergency assessment.
This is a product-use checklist, not a way to interpret an individual’s value. The new interaction is navigational: it can point users back to a specific record and help them formulate a question. The clinical decision remains outside the tool.
Privacy and connection controls matter
OpenAI says it doesn’t use connected medical records, Apple Health information, or conversations that use them to train foundation models or target ads. Users choose which accounts to connect and can remove them. Its help page says it deletes synced data from a disconnected source from OpenAI’s systems within 30 days, while information already included in chat history remains until those conversations are deleted.
The same page says new Health plugin connections follow the user’s global user permission setting; without a change, the default is “Allow low-risk actions.” People who want a tighter boundary can review Settings → Plugins → Health. OpenAI also says the consumer Health experience is not intended for clinical or covered-entity use and does not offer a Business Associate Agreement.
The new summaries reduce the effort needed to ask about a record. They do not reduce the need to verify the source, control what data is connected, or involve a professional when a result could change care. Our earlier report on ChatGPT and Epic records covers a related but separate integration and why a safety figure should not be mistaken for clinical clearance.
Sources
- OpenAI: ChatGPT release notes (September 28, 2026).
- OpenAI: Health in ChatGPT (accessed September 28, 2026).
