Microsoft Launches Copilot Health as a Secure Personal Health Workspace
Original: Introducing Copilot Health View original →
Microsoft has announced Copilot Health, a new health-focused experience that sits in a separate, secure space within Copilot. The product is built around a common consumer problem: people often have more health information than they can actually use, spread across test results, wearable dashboards, clinic portals, and general web searches that do not connect the dots.
Copilot Health is Microsoft's attempt to turn that fragmented data into a coherent profile. The company said the service can bring together health records, wearable signals, and health history so users can understand patterns, prepare for clinician visits, and ask more informed questions. Microsoft is rolling it out in phases, starting with a waitlist and an initial launch in English for adults in the United States.
What Microsoft says the product includes
- Support for health data from more than 50 wearable devices, including Apple Health, Oura, and Fitbit.
- Health records from more than 50,000 U.S. hospitals and provider organizations through HealthEx.
- Provider search tied to real-time U.S. directories, with filters for specialty, location, languages spoken, and insurance coverage.
- Answers built from credible health organizations across 50 countries, plus expert-written answer cards from Harvard Health.
Microsoft also tied the launch to a broader medical AI roadmap. The company said research such as its Microsoft AI Diagnostic Orchestrator will inform future capabilities, but it also emphasized that any new AI features drawing on those systems will be released only after clinical evaluations and with clear labeling. That is an important qualifier because Copilot Health is being positioned as a preparation and understanding tool, not as an autonomous diagnostic system.
The governance language in the announcement is unusually detailed for a consumer AI launch. Microsoft said Copilot Health conversations and data are isolated from general Copilot, protected with encryption at rest and in transit, and not used for model training. Users can disconnect data connectors and delete information. The company also said the service was developed with an internal clinical team, informed by more than 230 physicians across 24 countries, and certified under ISO/IEC 42001.
Microsoft already says its consumer products answer more than 50 million health questions a day, so Copilot Health looks like an effort to move from general Q&A into a more personalized and persistent health workflow. The promise is significant, but Microsoft also draws a clear boundary: the product is not intended to diagnose, treat, or prevent disease and is not a substitute for professional medical advice. Source: Microsoft AI.
Related Articles
Microsoft Discovery became generally available on June 2 for organizations building governed R&D workflows. The platform connects specialized agents, scientific knowledge, simulation tools, validation data, and a new local preview app for researchers.
Microsoft Research, UC Berkeley, UCSF, and Columbia researchers introduced generative causal testing. The method converts black-box language-brain prediction models into short hypotheses, then tests them by having an LLM write stories meant to activate targeted cortical regions.
Meta’s Brain2Qwerty v2 raises non-invasive sentence decoding to 61% word accuracy, with the best participant reaching 78%. The project used about 22,000 MEG-recorded typing sentences from nine volunteers and ships with training code plus a public v1 dataset.