Microsoft is rolling out Copilot Health in preview, positioning its Copilot chatbot as a place where users can ask health questions, connect wearable data and medical records, and get personalized insights—while stressing privacy, encryption, and that it won’t
By the time Microsoft’s Copilot Health finishes loading, the promise is already spelled out in the product’s pitch: one assistant for your questions, one place for your health history, one view that can tie together your wearables and the paperwork you’ve collected from providers.
Copilot Health is now in preview, offered through a dedicated space within the Copilot chatbot at copilot.microsoft.com/health. The concept is straightforward: ask health-related questions. and get answers built around a health profile Microsoft says you can customize with your background and goals—so the responses aim to be relevant rather than generic.
The feature also lives at the intersection of daily fitness tracking and clinical documentation. Microsoft says Copilot Health can connect Apple Health. with more wearable integrations coming soon. and it can link health records from over 50. 000 US provider organizations. With that setup. the assistant is designed to give personalized insights based on your overall profile. the connected data. and your conversations.
If you’ve ever stared at a blood test report and tried to translate numbers into meaning. that’s the use case Copilot Health leans into. Microsoft says Copilot Health can help you understand reports like blood test results. and it can guide users toward finding the right healthcare provider based on speciality. preferred language. location. and insurance coverage.
Microsoft is also trying to add weight to the medical side of an AI product. It says Copilot Health was developed with an external panel of over 250 physicians from more than 24 countries, and it reports achieving ISO/IEC 42001 certification.
The bigger question for users. of course. is trust—because asking an AI to handle health information isn’t like asking it to draft an email. Microsoft says Copilot Health conversations are not shared with the rest of Copilot. and that they are not used to train AI. It also says data is encrypted at rest and in transit. and that users can delete everything at any time to control what they share.
There’s a further layer to how Microsoft says it builds its answers. Copilot Health, Microsoft says, sources health information using principles published by the National Academy of Medicine and through a partnership with Harvard Health.
Still, Microsoft draws clear boundaries on what Copilot Health is and isn’t for. Like other health-focused AI tools such as “ChatGPT Health,” Copilot Health is not intended to diagnose, treat, or prevent any disease. The company also says it is not a substitute for professional medical advice.
This isn’t entirely starting from zero. Microsoft adds—pointing out that its consumer products already handle over 50 million health questions every day. Copilot Health. then. is framed less as an entirely new behavior and more as a formalized space for something Microsoft says people are already doing.
For now, the preview is limited to US users aged 18 and over who have a Microsoft 365 Personal, Family, or Premium subscription.
Microsoft Copilot Copilot Health AI health assistant medical records wearable data Apple Health privacy encryption ISO/IEC 42001 Harvard Health National Academy of Medicine Microsoft 365
So it’s like WebMD but with my records??
Idk how I feel about “connecting wearables and medical records.” Sounds convenient but also sounds like one more way companies know everything. They say privacy/encryption but it’s Microsoft, not a doctor.
Wait, so this thing can “find the right provider” based on insurance and location? That’s good I guess but like… won’t it just push whatever places pay the most? Also 50,000 orgs sounds huge, like are they really pulling all that from my history already or just when I connect it.
Blood test explanations sound cool, but AI is still gonna be like “your numbers are fine” and ignore the context. And preview means bugs, right? I saw something about ISO certification too which honestly sounds made up half the time. Next thing you know they’ll be using it to decide who gets care faster, and then you can’t opt out because it “learns from your conversations.”