If you already ask ChatGPT about labs, meds, sleep, or diet, and you want those answers to notice your connected data, Health in ChatGPT is the rollout to watch. On July 23, 2026, OpenAI began opening it to eligible U.S. users so medical records and Apple Health context can travel with you beyond a sealed health-only tab.
This is the practical map: who gets it, what you can connect, how permission works, and what OpenAI still says the product is not.
Watch OpenAI's Health rollout clip
OpenAI's own short puts the U.S. launch in one line: connect Apple Health and supported medical records so ChatGPT can help you understand information in context and keep track of what has changed. The rest of this article is the checklist the clip skips.
Who gets it and what you can connect
OpenAI says Health is beginning to roll out to logged-in ChatGPT users age 18+ in the U.S. on web and iOS, across Free, Go, Plus, and Pro. OpenAI also notes Health is not yet available in Codex.
From the Health sidebar you can securely connect:
Medical records from participating U.S. hospital and provider systems (OpenAI's health materials call out pathways tied to major EHR ecosystems such as Epic and Oracle Health, plus consumer clinical services in the partner mix)
Apple Health for movement, sleep, and activity patterns (Apple Health sync requires iOS)
Wellness and lifestyle apps that meet OpenAI's Health inclusion bar, including partners named in OpenAI's materials such as Function, MyFitnessPal, Weight Watchers, and others depending on account and region
Apps connect only with explicit permission, even if you already linked them for ordinary ChatGPT chats. Medical-record integrations and some apps remain U.S.-only in OpenAI's current framing.
Permission is the product rule
After sources are connected, ChatGPT can ask before using that context on a given answer. OpenAI's health materials describe the behavior like this:
With your permission, ChatGPT can consider relevant information you have connected, such as medications, lab results, recent visits, sleep, and activity, alongside the goals and context you share.
OpenAI also says connected health data and related conversations are excluded from foundation-model training and ad targeting, with additional encryption protections called out for Health-connected information. Still treat that as a policy claim you should verify in the in-product privacy controls for your account.
What changed from the earlier health hub test
During earlier testing, OpenAI said a large share of health-related questions (about 70% in its own testing commentary) happened outside a dedicated hub. The wider launch leans into that behavior: you can stay in the main thread, and ChatGPT can request permission to pull connected health context when it is relevant (a dietary restriction while picking a restaurant, a recent injury when planning lower-impact activity).
OpenAI still frames Health as support for understanding your information over time, not as a clinic. Company terms continue to say the service is not intended for diagnosis or treatment of a health condition. Verify anything clinical with a professional before you change meds, skip a visit, or act on an urgent symptom.
Confirmed vs not confirmed
Confirmed
U.S. 18+ rollout on web and iOS for Free through Pro (staged; not every account sees it at the same minute)
Apple Health and medical-record connections with permission prompts
Training/ad exclusion claims for connected health data and related chats
Main-chat use with explicit permission, not hub-only
Not available in Codex yet
Not confirmed
Same-day availability for every U.S. hospital system or every Apple Health category
Android feature parity on day one
That answers replace clinician judgment for meds, imaging, or urgent care
Worldwide medical-record connectors outside the U.S.
What to do next
Open ChatGPT on web or iOS and look for Health in the sidebar.
Connect only the sources you actually want referenced (start with one records system or Apple Health, not everything).
When ChatGPT asks permission mid-chat, read what it wants to pull before you allow it.
Keep correcting stale meds or labs in-thread; connected data is only as current as the source.
Follow the AI section feed for privacy or platform expansions.