Microsoft Copilot to Surface Harvard Health Content for Safer Health Answers

Microsoft is preparing to fold curated Harvard Health Publishing content into Copilot so that health-related questions return answers grounded in a trusted medical publisher — a move reported by major outlets that signals both a tactical effort to improve clinical accuracy and a strategic push to diversify Microsoft’s AI stack away from single-vendor dependence.

Background​

Microsoft’s Copilot family has rapidly expanded from productivity assistants into vertical copilots tailored for regulated industries, with healthcare among the highest-priority targets. The company already markets specialized healthcare offerings — including Dragon Copilot, built from Nuance voice and ambient-capture technology — and has a track record of integrating third-party medical references into Copilot Studio and healthcare agent workflows. These product moves provide the technical and commercial scaffolding for a publisher-license strategy that pairs fluent large language models with verifiable reference content. Recent reporting states that Microsoft and Harvard Medical School (through Harvard Health Publishing) have reached a licensing arrangement that will allow Copilot to surface Harvard Health content in responses to consumer-facing health queries; that update was reported as possible as soon as October and is framed by reporters as part of Microsoft’s broader effort to reduce reliance on any single model or vendor. Reuters and the Wall Street Journal both covered the deal in early October, reporting that Microsoft will pay a licensing fee and that the update aims to make Copilot’s answers “more practitioner-like.” Microsoft and Harvard did not immediately confirm detailed terms in early reports.

---

What was reported, exactly​

• The core claim: Copilot will begin using Harvard Health Publishing content so that answers to health queries reflect Harvard’s consumer-oriented medical guidance.
• Commercial terms: Reports say Microsoft will pay a licensing fee, but the amount, duration, and rights (display, summarization, derivative use) were not disclosed publicly in initial coverage.
• Timing: Multiple outlets reported an “as soon as October” timeline for the Copilot update to begin surfacing licensed Harvard content; again, the precise release cadence and rollout plan were not independently verified.
• Corporate positioning: Microsoft’s health AI leadership framed the move as part of improving accuracy and trust for health answers while diversifying dependence on external model providers. Reportedly, this is consistent with Microsoft’s broader pattern of adding vetted publisher content to vertical copilots.

The user-facing coverage that initially surfaced these points — including regional amplification on syndication sites — repeats the same central reporting thread: licensing Harvard Health Publishing content gives Copilot a stamped source of reference material for health questions. Those articles echo industry reporting and analysis circulated in forums and briefing notes that caution readers about the difference between licensed reference content being available to a model and the model reliably rendering clinically safe, context-aware medical advice.

---

Why Microsoft would do this: strategic and technical rationale​

Microsoft’s motivations split neatly between product trust and platform strategy.

• Build trust and reduce hallucinations: Licensing an authoritative publisher like Harvard Health Publishing is a pragmatic way to reduce the chance that Copilot invents medical claims or cites weak sources. Anchoring outputs to a known publisher improves perceived and potentially measurable reliability when retrieval mechanisms are configured for provenance.
• Commercial differentiation: Publisher partnerships let Microsoft present Copilot as a product with defensible claim sources — a useful lever when selling to enterprise healthcare customers and regulators who demand auditability. Past collaborations (for example, Merck Manuals integration into Copilot Studio) demonstrate a template for how publishers and AI platforms can cooperate commercially while retaining editorial control.
• Vendor diversification: Microsoft has expanded beyond a single-model dependency by integrating alternate model suppliers and developing in-house capabilities. Layering publisher content atop a diversified model stack reduces overreliance on any one foundation model while increasing the value of Microsoft’s proprietary retrieval and governance layers.

Technically, there are three plausible integration architectures Microsoft could employ, each with distinct trade-offs:

• Retrieval-Augmented Generation (RAG): Copilot queries an indexed Harvard Health corpus and conditions the model’s output on retrieved passages. This approach supports explicit provenance and can be engineered to quote or strictly summarize retrieved text. It’s the least invasive with respect to publisher IP and easiest to audit.
• Fine-tuning: Microsoft could fine-tune a model on Harvard Health texts or use them to calibrate model behavior. Fine-tuning embeds publisher knowledge into the model weights, improving fluency but making direct attribution harder and increasing legal and editorial complexity.
• Hybrid: Use RAG for consumer-facing Copilot responses and tightly controlled, fine-tuned models with retrieval checks for clinical-grade clinician tools (e.g., Dragon Copilot integrated in EHR workflows). This lets Microsoft balance transparency in general consumer responses with deterministic behavior in regulated clinical workflows.

Each option requires specific operational controls (index refresh cadence, excerpt display, mismatch handling) to ensure that the display of Harvard material is timely, accurate, and clearly attributed.

---

Clinical safety, regulatory and liability considerations​

Integrating publisher content into generative AI does not eliminate the serious safety, legal and regulatory challenges of providing medical information at scale.

• Safety-critical stakes: Medical queries can prompt actions with direct patient harm. Any consumer-facing assistant that offers triage-like guidance must make its limits explicit and provide escalation to clinicians or emergency services where appropriate. Existing legal regimes and guidance — including HIPAA for PHI and FDA guidance for clinical decision-support tools — create real constraints on what automated systems can legitimately do without formal regulatory review.
• Provenance vs. paraphrase risk: Even when a model has access to Harvard content, paraphrase drift — where the model subtly changes a recommendation during summarization — can introduce clinically significant errors. This is a core reason why many safety-minded architects prefer deterministic citation (quoting passages verbatim) and ensemble verification layers.
• Contractual and editorial scope: A licensing agreement may authorize display, summarization, or internal use for training — but those differences matter. If Harvard’s content is only allowed for retrieval and quoting, Microsoft must engineer RAG workflows that surface exact excerpts. If training licenses are broader, the publisher will need governance over how its editorial voice appears in paraphrased outputs. Initial reports did not disclose those contract details.
• Liability and regulatory posture: Licensing an authoritative source reduces some reputational risk but does not transfer legal liability for erroneous medical advice. Clinical tools embedded into EHRs that influence diagnosis or treatment decisions will likely trigger regulatory scrutiny and may need submission pathways or certification. Contracts and indemnities between Microsoft and publisher partners will be central to allocating risk — but those commercial terms were not made public in early coverage.

Practical regulatory steps organizations should expect Microsoft to address (or be asked to disclose) before adopting Copilot for clinical workflows include:

• HIPAA-compliant handling of patient data and explicit non-use-for-training guarantees where applicable.
• Audit logs that record which publisher passage (and which version) produced a given answer.
• Clear labeling and escalation prompts for high-risk queries (chest pain, suicidal ideation, stroke signs).
• Independent performance validation and post-market surveillance commitments for any clinical decision-support feature.

---

UX, transparency and accessibility​

Trusted content is not enough on its own; the user experience determines whether trust is realized or undermined.
Good UX patterns for medical answers should include:

• Inline provenance: show the exact Harvard Health Publishing excerpt or an explicit citation card that clearly indicates what the assistant used. This prevents the “black-box” problem where polished language masks uncertain or partial evidence.
• Confidence bands and disclaimers: when evidence is thin, Copilot should express uncertainty and offer next steps — for example, “Information referenced from Harvard Health Publishing; this does not substitute for medical advice. Contact a healthcare professional for personalized recommendations.”
• Escalation and locality: present local emergency numbers and recommend in-person care when symptoms are severe. For triage use-cases, provide pathways to clinicians or telehealth services rather than a single conversational answer.
• Accessibility and literacy adaptation: Harvard Health Publishing is authoritative but written in a style that may not suit all audiences; Copilot should offer plain-language rewrites, translations and culturally aware explanations without changing clinical meaning.

Poor UX — a single polished paragraph with no citation or action guidance — risks lulling users into inappropriate confidence and could increase liability and harm.

---

Market and competitive implications​

The Harvard licensing reports are part of a broader industry trend: platform vendors are pairing large, generalist generative models with curated, domain-specific knowledge layers to earn trust in regulated verticals.

• Publisher economics and influence: Publishers face a strategic choice — license content to platforms for reach and revenue, or withhold it to preserve direct traffic and editorial control. Microsoft’s reported deal follows earlier integrations (Merck Manuals) and suggests a playbook for monetizing high-quality medical content in AI assistants.
• Competitive positioning: Microsoft wants Copilot to be the go-to assistant for workplace and consumer health queries. Licensing Harvard content gives Microsoft a unique badge of authority to show enterprise health systems and consumers, while the company continues to diversify models (internal models, Anthropic, etc. to reduce dependence on OpenAI. This is a strategic defensive and offensive move: it reduces systemic vendor risk and increases product differentiation.
• Potential downstream deals: If this model proves commercially successful, expect additional publisher partnerships and possibly tiered offerings (consumer Copilot with visible citations; enterprise Copilot with curated, contractually guaranteed sources and downstream indemnities).

---

What remains unverified — cautionary flags​

Several materially important claims were not confirmed publicly at the time of early reporting. These gaps must be treated as unresolved until primary sources publish contract terms or official statements:

• Scope of the license: Which Harvard Health Publishing titles, topics, and formats (articles, Q&A, multimedia) are included? Is the deal global or limited by region or language? These details were not disclosed.
• Usage rights: Can Microsoft use licensed content only for retrieval and quoting, or may it also fine-tune models or create derivative knowledge artifacts based on Harvard material? This distinction determines provenance guarantees and legal exposure.
• Update cadence and versioning: How will Harvard update propagate into Copilot? Will users see a “last-updated” timestamp or version ID that identifies which edition of guidance was used? Lack of timely updates is a known hazard in medical guidance.
• Indemnity and liability allocation: Does Harvard assume any editorial responsibility for how its content is summarized or presented by Copilot? Initial coverage noted payment of a licensing fee but not indemnities or editorial controls.

Because these are high-stakes commercial and clinical questions, public-facing clarity from Microsoft and Harvard — ideally in the form of an FAQ and documentation for enterprise customers — will be essential to reduce confusion and risk.

---

Practical checklist for IT and healthcare teams evaluating Copilot with licensed publisher content​

• Confirm scope and rights: Ask vendors for written confirmation of which publisher content is in-scope, whether it can be used for training, and the geographic coverage.
• Require provenance and versioning: Demand UI-level provenance for every medically actionable statement and a visible last-update timestamp for cited guidance.
• Pilot with metrics: Run controlled pilots that measure accuracy, false-negative/false-positive clinical flags, clinician review time, and downstream impacts on coding and workflows.
• Contractual protections: Negotiate data residency, non-use-for-training clauses (if required), indemnities, and SLAs for content updates and security.
• Human-in-the-loop and escalation: Ensure clinicians see and verify draft outputs before they enter the legal medical record; route high-risk queries to human review.
• Accessibility and adaptation: Validate that the assistant can render plain-language and translated versions of guidance without losing clinical nuance.

---

Technical design patterns Microsoft should (and appears likely to) adopt​

• Deterministic citations for consumer health answers (show exact Harvard excerpt and link to the full article). This preserves editorial fidelity and reduces paraphrase drift.
• Fact-checker ensembles that verify model outputs against the retrieved Harvard passage and an alternate clinical source to detect conflicts or omissions. An ensemble approach increases computational cost but materially reduces hallucination risk.
• Temporal safeguards that flag potentially stale recommendations (e.g., “This guidance last updated in 2019; clinical knowledge may have changed”). That reduces the hazard of relying on out-of-date material.
• Distinct pipelines for consumer vs. clinician experiences: RAG with visible provenance for consumer Copilot; locked-down, fine-tuned, validated inference with audit logs for Dragon Copilot within EHR integrations.

---

Risks to watch and how they might unfold​

• Overconfidence in branded content: Users may equate a Harvard label with personalized clinical counsel, risking inappropriate self-treatment. Clear labeling and triage guidance are essential.
• Narrow editorialization: If Microsoft privileges a small set of licensed publishers, the range of clinical perspectives could narrow, creating monoculture risks where alternative, valid viewpoints are suppressed by product design choices.
• Contractual and litigation exposure: If Copilot paraphrases or omits critical nuance from a Harvard passage and harm occurs, litigation and regulatory scrutiny will focus on the contract terms and the product’s QA processes. Public clarity about indemnities and editorial controls matters.
• Data governance ambiguity: For healthcare customers, it is critical to verify that any patient data processed in conjunction with Copilot features is segregated, protected, and not inadvertently used to train broader models. Contractual guarantees and technical proofs (e.g., in-tenant processing) are necessary.

---

Bottom line​

Licensing Harvard Health Publishing for Copilot — if the reporting proves accurate and the license is engineered with strong provenance, update cadence, and contractual protections — is a pragmatic and defensible step toward making conversational AI more trustworthy for health queries. It aligns with a clear technical pattern: pair fluent LLMs with authoritative retrieval sources to reduce hallucinations and increase enterprise confidence. However, this is not a panacea. The deal alone cannot guarantee clinical safety, legal clarity, or unbiased coverage. The final user impact will depend on how Microsoft actually integrates the content: whether outputs include deterministic citations, whether publisher content is allowed for model training, how updates are propagated, and what governance and auditing tools are provided to enterprise customers. Those implementation details — currently unverified in early reporting — will determine whether the integration meaningfully improves patient safety and clinical decision support or simply dresses conversational AI with an authoritative label.
Organizations, clinicians and IT leaders should treat the reported Harvard license as a promising signal, not a guarantee. Before routing clinical work or patient-facing triage into Copilot-driven flows, require documentation, independent validation and contractual assurances that map directly to regulatory and clinical safety requirements.
The immediate practical advance is clear: Copilot gaining licensed, high-quality medical content would elevate the quality of health-related answers. The larger systemic imperative remains unchanged — combining editorial authority, transparent provenance, and rigorous validation is the only way to safely scale generative AI in healthcare.

---

Source: Asianet Newsable Microsoft's Copilot To Answer Health Queries Using Harvard's Medical Data, Research: Report
Source: Stocktwits Microsoft's Copilot To Answer Health Queries Using Harvard's Medical Data, Research: Report

 

[ChatGPT]

Microsoft’s Copilot is being positioned to give safer, more practitioner‑like answers to health questions by incorporating licensed content from Harvard Health Publishing — a move that industry reporting says will be paid for with a licensing fee and rolled into Copilot as part of Microsoft’s broader strategy to diversify AI models and reduce hallucination risk.

Background / Overview​

Microsoft’s Copilot family has expanded rapidly from productivity helpers into verticalized assistants for regulated industries, and healthcare is a top priority. The company already operates clinical products built on Nuance technology (Dragon Copilot) and has a pattern of integrating third‑party medical references into Copilot Studio and other enterprise workflows. Recent reporting indicates Microsoft has struck a licensing arrangement with Harvard Medical School’s Harvard Health Publishing so Copilot can surface Harvard’s consumer‑facing medical guidance for health‑related queries.
Why this matters: authoritative, editorially curated content can materially reduce the risk of confident but wrong responses from conversational AI — commonly called hallucinations — and gives Microsoft a named source to cite when users ask about symptoms, treatments, or general medical guidance. That combination is attractive both to consumers and to enterprise healthcare customers who demand provenance and auditability for clinical information.

---

What was reported — the core claims​

• Microsoft will license content from Harvard Health Publishing and integrate it into Copilot so that health‑related questions return answers grounded in that material.
• The Wall Street Journal first reported the core claim, with Reuters and other outlets providing corroboration; coverage states Microsoft will pay a licensing fee, though specific monetary terms were not disclosed publicly.
• The update was reported to be scheduled “as soon as October,” though rollout timing, precise product surfaces, and contractual scope remained unconfirmed in the initial reports.

These are the load‑bearing facts in public reporting to date; the details that determine legal, clinical and operational impact — scope of included titles, whether content can be used for model training or only retrieval, update cadence, and indemnity arrangements — were not public at the time of reporting and must be treated as unresolved.

---

Technical possibilities: how Harvard content could be used in Copilot​

There are three realistic integration patterns, each with different safety and audit implications:

1. Retrieval‑Augmented Generation (RAG) — the conservative, auditable option​

RAG indexes the Harvard Health corpus and retrieves exact passages to condition Copilot’s answers. This supports explicit provenance: Copilot can display the excerpt and a “Harvard Health says…” card, reducing paraphrase drift and making it easier to audit claims. RAG is the least invasive to publisher IP and easiest to certify for enterprise customers.

2. Fine‑tuning / alignment — deeper but less transparent​

Microsoft could fine‑tune an internal model using Harvard texts or use them to calibrate model outputs. That can improve fluency and naturalness, but it risks obscuring whether a given answer derives from Harvard content or model inference, complicating provenance and legal accountability.

3. Hybrid approaches — different pipelines for consumers and clinicians​

A likely pragmatic path: use RAG with visible citations for consumer‑facing Copilot answers while operating a locked, fine‑tuned, validated model with audit logs for clinician tools (Dragon Copilot, EHR integrations). This balances transparency for the public with determinism and speed for clinical workflows.

---

What Microsoft stands to gain​

• Lower hallucination risk: Anchoring answers to an authoritative publisher reduces the chance of fabricated claims in a domain where errors can cause real harm.
• Commercial differentiation: Named publisher content is a credible selling point for healthcare customers and regulators who demand traceable sources.
• Publisher revenue: Licensing creates a monetization pathway for high‑quality medical publishers that have historically relied on subscription or advertising models.
• Strategic vendor diversification: Microsoft has been broadening its model stack (adding Anthropic’s Claude, investing in in‑house models). Adding licensed content is another lever to reduce dependency on any single foundation model vendor.

---

Major risks and limitations — why licensing Harvard is not a cure‑all​

Even a deal with a top medical publisher does not eliminate the core safety challenges of medical AI. Important risks include:

• False sense of safety: A Harvard label can create user confidence that outstrips the assistant’s true capabilities. Licensed content can still be outdated, incomplete, or improperly summarized. Copilot outputs that blend Harvard text and paraphrase may omit crucial caveats.
• Scope and update cadence uncertainty: If the license is snapshot‑based or update frequency is slow, Copilot could cite guidance that’s no longer current. The contract’s update and versioning terms are pivotal. Reports so far do not disclose these terms.
• Paraphrase and decontextualization: Generative summarization can strip nuance from cautious clinical guidance and convert it into prescriptive language. Without deterministic citations or strict summarization protocols, risk remains.
• Liability and regulatory exposure: Licensing does not magically reassign legal responsibility. If Copilot misstates guidance or omits contraindications, harm can follow and legal questions will arise about product labeling, indemnity clauses, and whether the output constitutes medical advice under relevant laws.
• Mental‑health and crisis handling: Publisher content alone does not solve crisis‑triage issues. AI systems have prior failures in handling suicidality, acute chest pain, and other emergencies. Explicit escalation logic and human‑in‑the‑loop processes are essential.
• Editorial narrowness: Over‑reliance on a small set of publishers risks a monoculture of perspectives, potentially burying alternative but valid clinical viewpoints.

---

Product and UX considerations that will determine real world safety​

The public benefit hinges on product design and transparency. Key product choices to watch:

• Visible provenance: Does Copilot show “Harvard Health Publishing” excerpts and a last‑updated date for medically actionable statements? Deterministic citations are a must for auditability.
• Confidence bands and uncertainty handling: For ambiguous or low‑evidence topics, Copilot should surface uncertainty and advise consulting a clinician. Overconfident phrasing must be explicitly avoided.
• Escalation flows and human oversight: For triage, crisis, medication changes or diagnosis‑level recommendations, Copilot must route to clinicians or emergency services rather than provide a stand‑alone answer.
• Versioning and reindexing cadence: Users and clinicians need clear metadata about when the cited guidance was last reviewed and how often the Harvard corpus is reindexed.
• Enterprise controls: Healthcare organizations should be able to opt in or out of specific publisher sources, demand non‑training clauses, and require audit logs that tag every model call with source and model identifier.

---

Practical checklist for IT and clinical leaders evaluating a Harvard‑backed Copilot​

• Confirm scope and rights: obtain written confirmation of which Harvard Health Publishing titles, formats, and languages are included, and whether the license permits model training or only retrieval.
• Demand provenance and timestamping: require UI‑level citations and visible “last updated” dates for every medically actionable statement.
• Start with read‑only pilots: run pilots where Copilot suggests content and provenance but does not auto‑populate orders or clinical records.
• Require telemetry and audit logs: every call should log the content source, model used (e.g., OpenAI, Anthropic, internal), timestamp, and response hash for post‑hoc audits.
• Negotiate contractual protections: secure indemnities, SLAs for content freshness, data residency, and explicit non‑use‑for‑training language if required.
• Build golden test sets and clinician validation loops: measure false positives/negatives, mis‑summarization frequency, and clinical impact before broad deployment.

---

Regulatory and legal landscape — what to expect​

Generative AI for healthcare sits at the intersection of several regulatory frameworks. Key considerations include:

• HIPAA: Any Copilot feature that ingests protected health information must be evaluated for HIPAA compliance and data handling safeguards. Vendors and health systems must be explicit about whether interactions are stored, how they are protected, and whether they are used to train models.
• FDA oversight: If Copilot’s outputs move beyond informational guidance into triage or diagnostic decision support, FDA regulation or premarket review pathways might apply, depending on jurisdiction and functionality. Microsoft will need to clarify product classification and any regulatory assessments performed.
• Consumer protection and malpractice: The legal line between information and medical advice is context dependent. Clear labeling, disclaimers, and pathways to clinician consultation will affect liability exposure. Contracts with publishers and enterprise customers must address indemnity allocation.

---

Market implications and competitive signaling​

Microsoft’s reported licensing of Harvard Health Publishing is a strategic signal: major platforms are increasingly pairing foundation models with curated, publisher‑level knowledge layers to win trust in regulated verticals like healthcare, law and finance. This approach:

• Creates a product moat around named content.
• Puts pressure on rivals to secure similar publisher partnerships or to demonstrate superior provenance and auditability.
• Generates new revenue models for publishers who can monetize their editorial assets via licensing.

For Microsoft specifically, the move complements its broader multi‑model strategy — integrating Anthropic, advancing in‑house models, and layering publisher content — and makes Copilot a more defensible choice for healthcare customers.

---

Independent verification and what remains unverified​

Multiple outlets reported on the Harvard license and the Copilot integration, with the Wall Street Journal and Reuters among the first to publish the core claims. These independent reports corroborate the existence of a licensing arrangement and the intent to surface Harvard content in Copilot, while noting that Microsoft and Harvard had not publicly disclosed contract specifics at the time of reporting.
Unverified claims that need confirmation:

• Exact licensing fee amount and payment structure (undisclosed).
• Whether the license permits fine‑tuning/model training on Harvard content or is restricted to retrieval and display.
• The specific product surfaces (consumer Copilot, Copilot in Windows, Copilot Studio, Dragon Copilot in EHRs) and the rollout schedule. Early reporting suggested an October timeframe but did not confirm dates or geographies. Treat “as soon as October” as provisional until vendor documentation or product posts confirm rollout.

Where public reporting is silent or ambiguous, cautious language is warranted: the presence of licensed content in Copilot can improve reliability but does not guarantee clinical safety or regulatory compliance by itself.

---

Recommended best practices for Microsoft to make this a meaningful safety improvement​

• Deterministic citations: force the assistant to quote retrieved Harvard passages or display them prominently in answer cards, rather than paraphrasing without provenance.
• Fact‑checker ensembles: implement secondary verification steps that cross‑check generated summaries against the retrieved passage and an alternate trusted source to detect conflicts or omissions.
• Temporal safeguards: expose last‑updated timestamps and flag potentially stale guidance.
• Human‑in‑the‑loop gating: ensure high‑risk outputs are routed to clinicians or moderated workflows, especially for medication changes, triage, and crisis responses.
• Independent audits and transparency: publish third‑party evaluation plans and performance benchmarks for health queries that rely on Harvard content so enterprise buyers and regulators can assess real‑world safety.

---

Conclusion​

Microsoft’s reported licensing of Harvard Health Publishing for Copilot is a pragmatic and strategically sensible move: it pairs the fluency of large language models with an editorially curated medical knowledge base, addressing a visible failure mode of generative AI in healthcare. When implemented with visible provenance, strict update cadences, deterministic citation, human‑in‑the‑loop safeguards, and contractual clarity, this can be a real step toward safer, more defensible AI health answers.
At the same time, licensing is not a panacea. Critical details remain undisclosed — monetary terms, usage rights for training versus retrieval, and the operational rollout plan — and those details will determine whether the integration meaningfully reduces clinical risk or merely gives conversational AI a veneer of authority. Until Microsoft and Harvard publish implementation specifics and provide enterprise‑grade documentation and auditability, organizations and clinicians should treat the reports as promising directionality rather than a completed solution.
For IT leaders, clinical governance teams, and product managers, the immediate priorities are clear: demand transparency about scope and update cadence, require provenance and audit trails in every medically actionable response, and pilot cautiously with clinician validation and contractual protections in place. These steps will determine whether licensed publisher content transforms Copilot from a helpful information tool into a reliable, auditable partner for healthcare workflows.

---

Source: Windows Report Microsoft Taps Harvard Medical School for Copilot's Health-related Answers