Did You Know ChatGPT's Health Feature Can Connect to Your Medical Records?

Hey, it's Shiichan! I found a pretty significant announcement from OpenAI today. Did you know ChatGPT can now connect to your medical records?

OpenAI News openai.com

What was announced?

OpenAI News published "Launching Health in ChatGPT." Eligible users in the U.S. can now securely connect Apple Health and supported medical records to ChatGPT, so it can offer more personalized responses grounded in their own health information.

The rollout starts with logged-in ChatGPT users who are 18 and older in the U.S., on web and iOS, across Free, Go, Plus, and Pro plans. It's not yet available in Codex.

The story so far

Earlier this year, OpenAI introduced a dedicated health space to a limited group of users. But to get the full benefit, you had to actually navigate to that separate space.

Early testing showed that more than 70% of health-related conversations happened outside that dedicated space naturally — like wanting a dietary restriction considered while planning a meal. Having to switch screens ended up being an unnecessary step rather than a help.

What changes

Based on that feedback, you no longer have to leave your normal conversation to bring your health context in.

• With your permission, ChatGPT can draw on connected information like medications, lab results, recent visits, sleep, and activity, in context
• For example: considering a dietary restriction when suggesting a restaurant, or a recent injury when planning weekend family activities
• You can add @Health to a message to explicitly ask ChatGPT to use your health context
• Health, in the sidebar, remains your home base for connecting accounts, reviewing recent data and trends, and revisiting past health conversations

Dive Deep

Smarter models

The models behind these conversations got stronger too.

• GPT‑5.5 Instant, available to all free users, brought frontier-level health intelligence
• GPT‑5.6 Sol, for paid users, goes further with stronger reasoning across complex details and careful judgment
• On HealthBench Professional, an evaluation built with physicians, every GPT‑5.6 model outperformed GPT‑5.5
• Physicians extensively tested Health before release to help verify safety with connected health data

Privacy and permission controls

• Connected medical records and Apple Health information, and conversations that use it, are not used to train OpenAI's foundation models or target ads, regardless of your ChatGPT model-training setting
• By default, ChatGPT asks for permission before using connected information; you can switch to always allow in Settings > Plugins > Health
• You can disconnect at any time — data from that source is deleted from OpenAI's systems within 30 days (conversation history that already used it remains until you delete it)
• Memories can be created from Health conversations, but not directly from connected medical records or Apple Health data; use Temporary Chat or turn memory off if you'd rather not
• Before ChatGPT shares health information through another connected plugin (like sending a training plan to a running partner), additional safeguards may ask you to confirm

Wrap-up

• OpenAI News announced the rollout of Health in ChatGPT to eligible U.S. users
• You can connect Apple Health and supported medical records from U.S. hospital systems, One Medical, and Function Health
• Health context now works inside everyday conversations, not just a dedicated space, with permission controls
• GPT‑5.5 Instant (free) and GPT‑5.6 Sol (paid) strengthen health responses, verified with HealthBench Professional
• Privacy protections include no model training or ad targeting on connected data, ask-once-or-always-allow permissions, and disconnect anytime
• Available to logged-in U.S. users 18+ on Free through Pro, on web and iOS (not yet in Codex)

If you want ChatGPT to help make sense of your own or your family's health records, or you're just curious how AI and privacy design can coexist, this one's worth a read.