Does using Microsoft Copilot or Foundry make my ISO 42001 work easier?
A customer or contract asks for ISO 42001, and your AI runs on Microsoft services. Microsoft lists GitHub Copilot, Microsoft Copilot, Copilot Chat, Copilot Health, Copilot Studio, Dragon Copilot (including the Radiologist edition), Microsoft Foundry and Microsoft Security Copilot as in scope for ISO/IEC 42001 certification, and it publishes the certificates and audit reports on the Service Trust Portal. That is useful evidence, but it does not replace your own assessment. Microsoft states that customers remain responsible for engaging an assessor to evaluate their own controls, processes and implementation.
The distinction matters because the vendor's certification covers Microsoft's management system for the service, not your organization. It does not certify you, and it does not show that the prompts, data sources, grounding content, user access, human oversight or incident handling you built on top of the service meet the clauses and Annex A controls. Those depend on your own AI system, the data it touches and the processes around it. A Copilot Studio agent that reads your customer records raises different questions from an internal Copilot Chat deployment used for drafting, and each needs its own statement of applicability entry and evidence.
Before you rely on the vendor report, confirm which service and version your deployment uses against the in-scope list, pull the current certificate and audit report from the Service Trust Portal, and map what they cover to your own controls for A.10.2 (allocation of life cycle responsibilities) and A.8.2 (information for users). Ask your assessor to record where the vendor evidence stops and your evidence starts. Any Microsoft product that is not on the list needs its own assessment path.
Which Microsoft AI services are in scope and what that means
Microsoft lists GitHub Copilot, Microsoft Copilot, Copilot Chat and Copilot Health as in scope for ISO/IEC 42001 certification, with the certificate and audit reports available on the Service Trust Portal. The listing tells you that Microsoft's own management system covers these services. It does not describe how your staff prompt these tools, which data they connect to, or which business decisions people make from their outputs. Those choices belong to your AI management system and must be assessed on their own.
- Listed scope: GitHub Copilot, Microsoft Copilot, Copilot Chat and Copilot Health are named in Microsoft's ISO/IEC 42001 scope list.
- Tenant configuration: Permissions, connectors, sharing settings and which content each user can reach remain controls you set and must evidence.
- Acceptable use: Rules for what staff may enter into prompts and how outputs are checked before use sit with your policy and A.9 controls.
- Health and personal data: Copilot Health use with personal or patient data creates obligations under your own privacy and sector rules that Microsoft's listing does not satisfy for you.
Check which exact service and plan your teams use against the scope list and confirm the certificate version applies to it.
Microsoft lists Copilot Studio as in scope for ISO/IEC 42001 certification. Copilot Studio is a low-code environment, so the agents your teams build are a different matter from the service itself. An agent's instructions, knowledge sources, actions and the people who can publish it define how it behaves in your organization. Those design decisions are yours to document, test and govern under the life cycle and use controls.
- Listed scope: Copilot Studio is named in Microsoft's ISO/IEC 42001 scope list for the service.
- Agent design: Instructions, grounding data and tool actions you configure are your responsibility and need documented design records.
- Publishing rights: Who may create, approve and publish agents should be restricted and recorded as part of your roles and responsibilities.
- Connected systems: Each action that reaches an external system or personal data needs its own impact and data review.
Ask Microsoft for the application card covering the capabilities and limitations you rely on, then test your own agents against it.
Microsoft Foundry is listed by Microsoft as in scope for ISO/IEC 42001 certification. Foundry is a platform for building and operating AI applications, so the model choices, deployments, data sources and evaluations you run on it shape your own AI system. The platform listing supports your supplier and provider analysis, but the system you build is a separate object that your organization must govern and evidence.
- Listed scope: Microsoft Foundry is named in Microsoft's ISO/IEC 42001 scope list for the platform.
- Model selection: Choice of models and versions, and the reasons for them, must be recorded in your resource and life cycle documentation.
- Evaluation and monitoring: Testing results, drift checks and event logs for your application are evidence you produce, not something the platform certificate supplies.
- Data handling: The data you ground or fine-tune on falls under your data controls and any privacy obligations.
Confirm which Foundry capabilities your deployment uses and check each against the transparency documentation before relying on it.
Microsoft Security Copilot is listed by Microsoft as in scope for ISO/IEC 42001 certification. Security Copilot is used by your security team to investigate incidents and summarize threat data, so the risk sits in what analysts trust and act on. Your organization still decides which prompts, plugins and data sources are allowed, and how outputs feed incident response and management review.
- Listed scope: Microsoft Security Copilot is named in Microsoft's ISO/IEC 42001 scope list.
- Analyst reliance: Procedures for verifying Security Copilot output before it changes a containment or reporting decision belong to your controls.
- Plugins and data sources: Each connected source and plugin needs an access decision and an entry in your resource records.
- Incident records: Logs of prompts and actions taken on the basis of AI output support your event logging evidence.
Review the current Security Copilot documentation with your security leads before writing these procedures into your statement of applicability.
Microsoft lists Dragon Copilot, including the Radiologist edition, as in scope for ISO/IEC 42001 certification. Clinical use is high stakes, so the listing matters less than how your clinicians, deployment and patient data flows are governed. Any clinical or regulatory obligations that apply to your organization stay with you, and the outputs used in care decisions need your own oversight and impact assessment.
- Listed scope: Dragon Copilot, including the Radiologist edition, is named in Microsoft's ISO/IEC 42001 scope list.
- Clinical oversight: Clinician review of AI-generated content and the authority to override it must be defined and recorded.
- Patient data: Consent, retention and sharing of patient information remain subject to your own privacy and health rules.
- Impact assessment: Effects on patients and staff, including errors in documentation, need an assessment documented under your own process.
Confirm which edition and configuration your clinical teams use before citing the listing in any assessment.
How to get Microsoft's certificate and audit reports from the Service Trust Portal
This sequence walks you from the Microsoft compliance listing to a dated, scoped copy of the ISO/IEC 42001 certificate and audit report, filed as evidence for your own assessment. It shows what the documents prove and what they leave for you to assess.
Confirm the Microsoft AI services you actually use
Start by listing the Microsoft AI services your organization builds on or deploys, such as Copilot, Microsoft Foundry or Security Copilot, and check each against the in-scope list on Microsoft's ISO/IEC 42001 compliance page. Note any service you use that is not listed, because Microsoft's certificate does not cover it. Owner: the AI governance lead with the platform team. Done when you have a written list of in-scope and out-of-scope services that will drive the rest of the sequence.
Sign in to the Service Trust Portal
Open the Service Trust Portal with an account your organization controls and confirm you can access the compliance documents section. Access terms, sign-in requirements and document availability are set by Microsoft and can change, so confirm them in the portal and with your Microsoft account team before you rely on this step. Owner: the compliance lead. Done when the portal opens under an account that is recorded in your access log.
Locate the ISO/IEC 42001 certificate and audit report
In the portal, search for the ISO/IEC 42001 documents and download the certificate and the audit report that Microsoft publishes for its AI services. Save the original files without editing them, so the copies you file match what the certification body or auditor issued. Owner: the compliance lead. Done when both files are saved in a read-only evidence folder with their original names intact.
Review the scope statement against your service list
Read the scope statement in the certificate and audit report and match each service and version it names against your list from the first step. Mark any service you use that the scope does not name, and any service the scope names with a version different from the one you run. Owner: the AI governance lead with a security engineer. Done when every service on your list is marked covered or not covered, with the scope wording quoted beside it.
Record the report date and validity in the evidence register
Enter the certificate number, issuing body, report date, period covered and scope wording into your evidence register, along with the file location and the owner who downloaded it. Note that the report describes Microsoft's controls at a point in time and that you need to check for newer versions before each assessment. Owner: the compliance lead. Done when the register row links to the stored file and lists the next review date you set.
Map Microsoft's coverage to your own Annex A controls
Using the scope and the report's description of Microsoft's controls, decide which Annex A controls Microsoft's evidence can support and which remain with your organization, such as your AI policy, impact assessments and use controls. Record that mapping in your statement of applicability and label Microsoft's report as supporting evidence, not as proof of your own controls. Owner: the AI governance lead. Done when each in-scope Annex A control shows the Microsoft evidence it relies on and your own evidence for the rest.
Hand the package to your independent assessor
Give the assessor the downloaded documents, the scope review, the evidence register row and your statement of applicability. Ask them to confirm how they will treat Microsoft's report in their assessment, since they evaluate your controls and implementation independently of Microsoft's certification. Owner: the compliance lead. Done when the assessor has acknowledged receipt and recorded any questions about the Microsoft evidence.
Using the Purview Compliance Manager template and transparency documentation
Microsoft Purview Compliance Manager includes a premium assessment template for ISO/IEC 42001, and Microsoft publishes transparency documentation, such as application cards, that describes how its AI systems work and what they can and cannot do. Treat both as starting points for your own evidence, not as a finished assessment. The template lists controls and lets you record your own assessments, actions and evidence against them, but the scoring and the conclusions about your implementation are yours to make. Before you plan around it, confirm with Microsoft which license tier or Purview subscription unlocks the premium ISO/IEC 42001 template in your tenant, since that determines who on your team can use it and what it costs to add.
The most direct use of the application cards is for A.8.2, which asks you to determine and provide the information users need about the AI system. Microsoft's documentation covers the vendor side of that information, including intended uses, capabilities and limitations. Your user information must describe how your deployment behaves: which Copilot or Foundry features you enabled, which data sources they can reach, who may use them and what users should not rely on them for. Map each relevant statement from the application card into your own user guidance, and record the source and date so an assessor can trace it. Similarly, use the transparency material as inputs to your A.6 technical documentation, then add the configuration records, data flows and monitoring settings specific to your environment.
Plan the template work as a gap exercise. Import the template into a scoped assessment, mark each control as applicable or excluded in line with your statement of applicability under clause 6.1.3, and assign an owner for every control you keep. Expect the Microsoft-supplied content to cover the platform layer only, so the controls on your policies, impact assessments, use objectives and supplier processes will still need your own documents and sign-off. Keep the assessment exports, dated screenshots and version history, because an independent certification body will ask how the assessment was built and when it was last reviewed.
What Microsoft's certification covers and what you must still assess
This table compares ISO 42001 control areas by who typically provides the evidence: Microsoft's certified platform, or your organization's own AI management system. Read each row as a question of whether the vendor's certificate answers the control or only supports your assessment of it. The main takeaway is that Microsoft's certification covers the platform, while your policies, roles, impact assessments, data records and user information remain yours to prove.
| Control area | Typically covered by vendor evidence | Stays with your organization | Evidence you produce |
|---|---|---|---|
| Vendor AI platform controls | Certified operation of in-scope services such as Microsoft Foundry and Copilot, with audit reports on the Service Trust Portal | Confirming each service you use is in scope and that your configuration matches the audited scope | Scope register listing services, tenants and versions, and the Service Trust Portal report reviewed |
| A.2 Policies related to AI | None for your own policy | Your documented AI policy, its alignment with other policies and periodic review | Approved AI policy, review records and mapping to security and privacy policies |
| A.3 Internal organization | None; Microsoft's internal roles do not assign yours | Named AI roles, responsibilities and a channel for raising concerns | RACI for AI systems and records of concerns received and handled |
| A.4 and A.7 Resources and data | Description of data handling in Microsoft's transparency documentation | Resource record for each AI system, plus data acquisition, quality, provenance and preparation for your own data | Resource inventory, data lineage records and data quality checks on your inputs |
| A.5 Impact assessments | Application cards describing capabilities and limitations of Microsoft AI systems | Your impact assessment of individuals, groups and society for each use case, which Microsoft cannot perform for you | Completed impact assessment documents, approvals and review dates |
| A.9 Use of AI systems | Platform guardrails and product-level controls, as described in product documentation | Responsible use processes, objectives and enforcement that keep use within intended purpose | Use policies, access configurations, usage monitoring reports and exception logs |
| A.10.2 and A.10.3 Suppliers and responsibilities | Microsoft's certificate and audit reports as evidence of its own supplier controls | Allocation of life cycle responsibilities between you, Microsoft, partners and customers, and supplier alignment checks | Responsibility matrix, supplier assessment records and contract clauses on AI use |
| A.8.2 User information | Public transparency documentation Microsoft publishes for its AI systems | Information you give your own users, including limitations and intended use of the systems you deploy | User-facing documentation, release notes and version records for your deployments |
The assessment work that stays with your organization
Microsoft says customers that need ISO/IEC 42001 certification for their own Microsoft AI implementations can use Microsoft's certification in their compliance assessment, but remain responsible for engaging an assessor. The cards below cover the work that stays with your organization: scoping, risk and impact decisions, supplier responsibilities, monitoring and governance.
Engage an independent assessor
Microsoft's certificate covers Microsoft's side of the service. Your own controls, processes and implementation still need evaluation by an independent assessor or certification body, which is the only party that can issue a report or certificate for your organization. Ask the assessor early how they will use Microsoft's audit reports and where they will need your own evidence. Confirm the assessor's accreditation and scope before fieldwork starts.
Define the AI system boundary
Your statement of applicability under clause 6.1.3 must list the Annex A controls you need and justify each inclusion or exclusion. Start by defining which Copilot, Foundry or Security Copilot deployments, prompts, connectors, data stores and user groups sit inside your AI management system. A boundary drawn around the Microsoft service alone leaves out the configuration, data and prompts you control, and assessors will test those first.
Run your own impact assessments
Microsoft's transparency documentation describes how its AI systems work, their capabilities and their limitations. It does not assess the impact of your use case on your customers, employees or society. Your organization must run its own AI risk and impact assessments for each system it deploys, using the Annex A impact assessment controls. Record the affected groups, the foreseeable misuse and the treatment decisions, and keep them current when the use case or the model version changes.
Allocate responsibilities with Microsoft
Under A.10.2, you must allocate responsibilities across the AI system life cycle between yourself, your partners, suppliers, customers and other third parties. Under A.10.3, you must show that the services you obtain from suppliers align with your approach to responsible AI development and use. Write down which party owns model selection, grounding data, prompt content, output review, and incident response. Keep the supplier review on file so an assessor can trace each allocation to evidence.
Monitor and log your deployments
Microsoft's audit reports cover the service, but event logs and monitoring of your own AI deployments are your responsibility. Confirm which logs you need to satisfy the event logging and operation controls, including prompts, outputs, administrator actions and configuration changes. Decide where logs are stored, who reviews them, how long they are retained, and how alerts reach the people who handle AI incidents. Test that the logs actually record what your risk assessment says matters.
Run internal audit and management review
Clauses 9 and 10 require your organization to check its AI management system and act on what it finds. Internal audit should sample your Copilot and Foundry deployments against your own policy and statement of applicability, not against Microsoft's certificate. Management review should receive the impact assessments, incident records and supplier findings, and record decisions and owners. Assessors will ask for the most recent internal audit and management review records before they accept any other evidence.
What to check before you rely on Microsoft's evidence
Microsoft's certificate and audit reports are a starting point, not a finished compliance position. Each check below narrows what the vendor evidence covers and what you still must show, and each one should be confirmed with Microsoft and with your assessor before the evidence goes into your statement of applicability.
Confirm the exact service and version in scope
Microsoft's in-scope list names products such as Copilot Studio, Microsoft Foundry and Security Copilot, but a product name alone may not tell you which features, editions or deployment models the certificate covers. Check the scope statement in the Service Trust Portal against the specific service and configuration your AI systems actually run. Ask Microsoft to confirm in writing whether a feature you plan to use, or a preview capability, falls inside that scope.
Confirm the certificate date and issuer
Open the ISO/IEC 42001 certificate and audit report in the Service Trust Portal and record its issue date, validity period and the certification body that issued it. A report that is out of date, or that was issued for a different scope, does not cover your current use. Check that the issuing body is a certification body you can verify, since ISO itself does not certify organizations and accreditation of certification bodies is voluntary.
Map each AI system to in-scope services
A single application may combine an in-scope Microsoft service with a model, plug-in, connector or data store that Microsoft does not cover. For each AI system in your inventory, list every component that processes data or produces outputs, mark which ones are in the Microsoft scope, and identify the gaps. Those gaps are where your own controls under the Annex A life cycle and supplier requirements must carry the evidence.
Check data residency and retention settings
Vendor certification does not decide where your prompts, outputs and logs are stored or how long they are kept. Confirm the regions your tenants use, the retention settings on prompts, outputs and audit logs, and whether those settings match the data handling rules in your AI policy and any applicable privacy obligations. Record the chosen settings as configuration evidence, since an assessor will ask how they were set and who approved them.
Confirm your assessor accepts vendor evidence
Microsoft states that customers remain responsible for engaging an assessor to evaluate their own controls, processes and implementation. Before you rely on the Microsoft certificate, ask your assessor whether they will accept the Service Trust Portal report, how they want it referenced, and what supplementary evidence they will test. Agree the method in writing so the vendor report supports your statement of applicability rather than being treated as a substitute for it.
Verify the Purview template and transparency documents
The Purview Compliance Manager template for ISO/IEC 42001 and Microsoft's application cards help map requirements to evidence, but they describe Microsoft's systems, not your deployment. Confirm which template version you are using, which assessments you have completed, and which controls remain marked as your responsibility. Check that the application card you cite matches the model and version you deploy, and update your own documentation when either changes.
Where an implementation partner such as DeployOpen fits
An implementation partner fits between your own AI governance team and the independent assessor, and its job is to prepare your side of the evidence, not to produce the assessment. DeployOpen plans, configures, documents and prepares your systems, controls and evidence inside your own environment. For Azure and Copilot work, that can include mapping the items in the Microsoft Purview Compliance Manager ISO/IEC 42001 premium assessment template to your own controls, identifying which items Microsoft's certification covers and which items depend on your configuration, data handling and use cases, and assembling the records an assessor will ask for. DeployOpen does not issue audit reports, attestations or certificates. Only an independent assessor or a certification body can do that, and the organization stays responsible for its own controls and for engaging that assessor.
Engagements are fixed-scope. Before work starts, the scope, the systems and Microsoft services covered, the deliverables and the timeline are agreed in writing, and a quote follows from that written scope rather than from a published price list. Ask any partner to state in writing which Purview template items it will map, which Azure or Copilot configurations it will review, who keeps the resulting documentation, and what happens if the assessor asks for evidence outside that list. Confirm with your own assessor, before you sign, what evidence they will accept and whether they want the mapping in a particular format. DeployOpen cannot promise any assessment outcome, and a partner that promises one should be treated with caution.
Questions buyers ask about Microsoft services and ISO 42001
Does using Copilot make my organization ISO 42001 certified?
No. Using a Microsoft AI service does not certify your organization. Microsoft lists GitHub Copilot, Microsoft Copilot, Copilot Chat, Copilot Health, Copilot Studio, Dragon Copilot (including the Radiologist edition), Microsoft Foundry and Microsoft Security Copilot as in scope for ISO/IEC 42001 certification, and that certificate belongs to Microsoft's own AI management system. Your organization is certified only when an accredited certification body audits your AI management system and issues a certificate to you. ISO does not certify organizations. Microsoft's certificate can support your compliance assessment for the parts it covers, but it does not cover how you choose use cases, document impacts, train staff, manage your data or oversee your suppliers. Ask your assessor to map which controls rely on Microsoft and which remain yours.
Where do I find Microsoft's ISO 42001 audit reports?
Microsoft publishes its ISO/IEC 42001 certificates and audit reports on the Service Trust Portal, and its Microsoft Learn compliance page lists the in-scope services and links to that portal. Access requires signing in to the portal and accepting its terms. Read the scope statement before you rely on any report. Check the exact service names, the audit period and any carve-outs against the implementation you run, such as a specific Copilot edition or a particular Foundry deployment. A report covers only the scope it states, so a service missing from that scope is not covered. Keep a dated copy of each report in your evidence repository, record its version and retrieval date, and recheck the portal when Microsoft updates its pages. Your assessor will expect to see that you reviewed the scope, not only that you downloaded the file.
Can I use the Purview Compliance Manager ISO 42001 template alone?
No. The template helps you organize the assessment, but it does not complete the assessment. Microsoft Purview Compliance Manager offers a premium assessment template for ISO/IEC 42001, which gives you a structure for recording your own actions and evidence against the standard. It does not review your policies, impact assessments, supplier arrangements or operating records, and it does not produce a certification. Confirm in your own tenant exactly what the template includes, what it tracks and who can edit it. The statement of applicability required by clause 6.1.3 must reflect your own analysis of which Annex A controls apply and why, and an independent assessor decides whether your controls meet the requirements. Treat the template as a project tracker and evidence index, and keep your decisions and their justification in documents you own.
How much effort is left after Microsoft's certificate, and what drives it?
The remaining effort depends mainly on how much of the AI system you build, configure and operate yourself. No researched figure is available for this, so ask assessors and partners for an effort estimate against a written scope. The main drivers are the number of AI systems and use cases in scope, whether you build on Foundry or only use Copilot licenses, how many Annex A controls you include in the statement of applicability, how much ISO 27001 or privacy maturity you already have, the number of data sources and impact assessments, and how many suppliers you depend on. Costs fall into two groups. One-time costs include the gap analysis, policy and impact assessment writing, evidence setup and the stage 1 and stage 2 audit fees paid to the certification body. Recurring costs include surveillance audits, internal audit, management review, retraining and reassessment whenever a system changes materially.
Who must engage the assessor for our Microsoft AI implementation?
Your organization must engage the assessor. Microsoft states that customers who use its certification in a compliance assessment remain responsible for engaging an assessor to evaluate their own controls, processes and implementation. The party that issues a report or certificate must be independent: for certification, an accredited certification body; for readiness reviews or internal assurance, a qualified consultant whose conclusions you do not present as a certificate. ISO says accreditation of certification bodies is voluntary, so ask whether the body is accredited for ISO/IEC 42001 and whether it applies ISO/IEC 42006:2025, which sets additional requirements on top of ISO/IEC 17021-1 for bodies certifying AI management systems. Get the engagement terms, audit plan and fee basis in writing before you sign.
Can a partner like DeployOpen do the ISO 42001 work for us?
A partner can prepare systems and evidence, but it cannot own your controls or issue your certificate. DeployOpen plans, implements, configures, hardens, documents and supports self-hosted open-source platforms inside your own infrastructure, and it prepares systems, controls and evidence for compliance assessments. It does not issue audit reports, attestations or certifications, and independent auditors and certification bodies do. Your organization keeps ownership of its AI policy, risk acceptance, impact assessments, supplier decisions and management review. DeployOpen engagements are fixed-scope, with scope, deliverables and timelines agreed in writing before work starts, and a quote follows from that written scope. Ask any partner which documents it writes, which ones your leaders must approve, and what it will not do for you.
Do Microsoft's transparency documents cover our user information duties?
They cover part of the work, not all of it. Microsoft publishes transparency documentation such as application cards that describe how its AI systems work, their capabilities and their limitations. Control A.8.2 asks your organization to determine and provide the information users of your AI system need, and that obligation attaches to the system you deliver, not only to the Microsoft component inside it. Use Microsoft's documents as source material, then write your own user information covering your configuration, your intended purpose, your data handling, known limitations you have tested, and how users report problems. Test that the information is accurate for your deployment, keep the version you published, and link it to the impact assessment and incident process that the control set expects you to maintain.
How DeployOpen prepares Microsoft-based AI systems for assessment
DeployOpen works on the parts of ISO 42001 on Microsoft Azure and Copilot services that remain with your organization: your own AI systems, configurations, policies and evidence. We scope the work in writing, build and document the controls you own, and hand over an evidence pack for your independent assessor, who makes the assessment decision.
How we scope
- We confirm which Microsoft AI services your own implementations use and which Microsoft certificate and reports apply to each.
- We map your AI systems, data flows and third-party dependencies to the Annex A controls your statement of applicability includes.
- Engagements are fixed-scope, with scope, deliverables and timelines agreed in writing before any work starts.
How we build
- We configure your Azure and Microsoft AI environments and Purview Compliance Manager assessment against the scope you agreed.
- We document your AI policy, roles, impact assessments and supplier responsibility records alongside the technical settings.
- We test your AI systems and their monitoring and event logging, recording results for review by your management team.
How we hand over
- We deliver an evidence pack that shows which controls are met by Microsoft, which are yours and where the evidence sits.
- We walk your compliance and engineering owners through the documentation so they can answer assessor questions directly.
- We state clearly what we did not assess, so your assessor can plan its own testing without relying on our work.
How we support
- Ongoing support, upgrades and continuous testing are available as a retainer after the initial handover.
- Retainers can cover updates to documentation when Microsoft changes service scope, transparency documents or certificates.
- Ask us to confirm the current scope of any Microsoft certificate with the Service Trust Portal before each assessment cycle.

