HugoScore hugoscore.org

Full review

Microsoft Copilot (consumer health) CAIHL draft report

Evidence-linked HugoScore draft report for a health AI tool that affects patients.

Agency checklist · October 30, 2026

Leans expands 8 of 12 yes

Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.

Assessed function: Dedicated Copilot Health space (preview) inside Microsoft Copilot, with optional wearable and health-record connections. General Copilot health answers are excluded.

A no can mean missing documentation or institution-dependent control; it does not by itself establish harmful behavior. These categories are not clinical safety or product-quality ratings.

  • Public documents do not establish actual product behavior.
  • The independent review screened rules and selected evidence; it did not replicate all twelve answers for this tool.
See all twelve answers and their evidence
QuestionAnswerEvidence / reviewSource and reasoning
1. Told it is AIYesdocumented
Not independently replicated
Copilot Health makes use of increasingly sophisticated AI to make sense of patterns in your health data Source

The support page itself does not label Copilot Health as AI. The product sits inside Microsoft Copilot. I did not verify an in-product consent screen.

2. Operator and model maker namedNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending. Source

Microsoft is named as operator. The page does not say which AI model powers Copilot Health. The announcement mentions an internal clinical team and MAI-DxO research but no model maker.

3. Patient chooses to use itYesdocumented
Not independently replicated
Sign up to be one of the first to try Copilot Health Source

Preview for Microsoft 365 Personal, Family and Premium subscribers. The user opens Health from the More menu. The help page says connecting wearables or records is 'If you choose'.

4. Patient sets the taskYesdocumented
Not independently replicated
a dedicated space in Microsoft Copilot where you can talk through private health questions Source

The user brings the questions.

5. Patient can decline or stopYesdocumented
Not independently replicated
Doing so will delete all your health data including your conversations, Health Profile, and connected data sources Source

Self-started tool per the Q3 source. The help page says deleting the account also withdraws consent for collection and processing.

6. Patient sees the AI outputYesdocumented
Not independently replicated
a high-level summary of your health Source

The page describes the Health Profile as an AI-built summary the user can view and edit. Conversations are stored and can be deleted.

7. Patient can correct or challengeYesdocumented
Not independently replicated
You can edit or delete anything in your health profile anytime. Source

The page also describes a thumbs-down on a response as feedback. Editing the profile is the stronger route.

8. Patient can get a copyNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending. Source

Neither the help page nor the announcement mentions export. The consumer health data policy covers access and clearing through the privacy dashboard, not a copy.

9. Patient can deleteYesdocumented
Not independently replicated
You can delete all your health conversations and previously saved health information anytime you want from your settings. Source

In settings, choose Health and then Delete my health data.

10. Training or sale does not override patient choiceYesdocumented
Not independently replicated
No. The information and data you share with Copilot Health is not used to train AI models. Source

The announcement says the same. This applies to the Health space only, not general Copilot conversations.

11. Patient can run it or take it elsewhereNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending. Source

Vendor-hosted only. No export and no self-host route.

12. No advertising or data businessNodocumented
Not independently replicated
for example, for advertising or marketing purposes Source

Conflict: the help page says Microsoft 'will not use the information you provide to Copilot Health to create an advertising profile' and that health data is not sold. The consumer health data policy lists advertising or marketing as a consent-based use and lists sharing categories, and it names Copilot Health as covered. I answered by the binding policy.

Initial coding: Claude Sonnet helpers, compiled by Claude Opus 5.5 (Anthropic). Selective review and corrections: GPT-6 Astra (OpenAI). Editor: Hugo Campos. Acceptance of corrections is not product verification.

Checklist rules · Independent review and limitations · Later evidence follow-up

Earlier report

Microsoft Copilot (consumer health): CAIHL reassessment

September 8, 2026. Published AI-assisted draft.

This assessment covers the dedicated Copilot Health preview. Its documented Health-space protections, citations, and connection controls must be distinguished from general Copilot health conversations. Preview access and subscription conditions remain relevant.

Scope and agency posture

Dedicated Copilot Health preview, with general Copilot health answers identified separately. Health-space protections must not be attributed to every health conversation in general Copilot.

Posture: Potentially agency-expanding, subscription-gated preview.

Axis: 71/100, retained provisionally. No numerical recalibration was performed.

Sources and documented findings

Patient authority

Inference from the documented workflow: Patients bring goals and questions, select connections and can disconnect them. The dedicated space provides meaningful direction, although recommendation and source-selection design remain vendor controlled.

Critical capacity

Editorial assessment: Citations and navigation support independent checking and action. Public evidence does not establish a patient workflow to correct generated insights or propagate corrections to source records.

Informed control

Editorial assessment: Distinguishing the protected Health space from general Copilot is essential. Privacy and certification claims need their proper scope. Patient-organization input is evidence of consultation, not patient governance.

Assessment and limits

The baseline combines two deployments with different conditions. Its positive but qualified placement remains defensible for the dedicated space. Price is an access limitation, not evidence that patient goals are overridden.

Confidence: Moderate for documented preview design.

Remaining uncertainty: Current in-account availability, connector retention and correction behavior were not tested

Published documentation is evidence of stated conditions, not proof of actual implementation. Unknowns did not receive automatic negative points. Funding, sponsorship, and public code do not determine agency by themselves.

Review provenance

  • Reviewer/model: OpenAI Codex / GPT-6.
  • Method: Focused public-source reassessment using CAIHL: patient authority, critical capacity, and informed control. Existing evidence plus one focused primary-source pass and at most one targeted follow-up. No live product testing. Numerical scores remain provisional editorial placements, not a new calculation.
  • Human review: Hugo Campos authorized publication of these AI-assisted draft reassessments on September 8, 2026. This does not claim comprehensive human verification of every finding.
  • Earlier review: 2026-06-10. Historical assessment. Earlier claims are not automatically reverified by this publication.