Consulting services / ITKoncept

AI governance for business

AI governance means knowing which AI services your business uses, who owns them, what company data they can access and how people sign in: through Entra ID and SSO or personal accounts. An AI policy defines approved use. We map the current setup and help put controls and an approval process in place.

Start with a Microsoft 365 assessment

Talk to us about AI governance

Example: what an inventory can look like

EXAMPLE — fictitious company. Roles, user counts and assessments are fictional, not customer data.
AI serviceOwnerUsersCompany data?Sign-inStatus
Microsoft 365 CopilotIT42Yes, through Microsoft 365 permissionsEntra IDApproved
Copilot Studio (agents)IT/Business6Yes, the agent’s connected sourcesEntra IDUnder review
ChatGPT EnterpriseMarketing12Yes, uploaded files and connectorsSSO via Entra IDUnder review
Claude / Perplexity with personal accountsUnknownUnknownUnknownPersonal accountsBeing mapped
Free AI tools on the webNoneUnknownRisk of company data being pasted inNo sign-in / personal accountUnapproved

Approved: authorised for the stated use. Under review: risks and conditions are being assessed. Being mapped: ownership or usage is unknown and needs review. Unapproved: must not be used for company work.

What we do

  • AI service inventory, including shadow AI: We review enterprise apps, sign-in logs and app consents/OAuth grants in Microsoft Entra ID. These logs only show apps that sign in through Entra. Cloud Discovery in Microsoft Defender for Cloud Apps has a generative AI category. Network data from sources such as Defender for Endpoint can also reveal use of these services, including with personal accounts, on onboarded devices. Coverage depends on licences, configuration and available data.

  • Identity and access: We configure SSO through Entra ID for approved services that support it. We restrict or block user consent to apps, review Conditional Access and define who may use what.

  • Data protection: We review Microsoft Purview sensitivity labels, oversharing in SharePoint and OneDrive, and DLP. Microsoft Purview DSPM for AI can provide visibility into AI interactions with sensitive data where supported and configured.

  • AI policy and approval process: We document approved services, who approves new ones and which data may be used where. We also define how decisions are followed up.

  • Copilot readiness: We assess identity, permissions and data protection before Copilot rollout. The Microsoft 365 assessment includes Copilot readiness.

Read about Microsoft 365 Copilot for business

See what the Microsoft 365 assessment covers

How it works

  1. Inventory

    We collect services, owners, users, data access and sign-in methods. Unknown details are flagged for further review.

  2. Assessment

    Each service is assessed as approved, under review or unapproved. Services with unknown usage remain marked as being mapped until there is enough information.

  3. Actions

    We implement agreed changes to SSO, consent settings and sensitivity labels. The AI policy sets out rules and responsibilities.

  4. Follow-up

    We review new services, changes in data access and previous decisions on a recurring basis.

We work remotely across Sweden and on site in the Stockholm area.

Common questions

AI governance FAQ

What is AI governance?

AI governance means controlling company AI services, ownership, data access and sign-in. A policy and approval process define how services may be used and reviewed.

What is shadow AI?

Shadow AI is AI used for work without the company’s knowledge or approval. This can include web tools or services with personal accounts where company data is entered.

How do we see which AI services are being used?

Entra ID shows apps and sign-ins that go through Entra. Defender for Cloud Apps can add Cloud Discovery. Network data from sources such as Defender for Endpoint can also reveal use of these services, including with personal accounts, on onboarded devices. Discussions with the business are also needed. No single log shows all AI use.

Do we need an AI policy?

If AI is used for work, staff need clear rules. The policy should define approved services, permitted data, responsibilities and how new services are reviewed. It needs updating as usage changes.

How does AI governance relate to Copilot?

Copilot is one of the AI services that need governance. Copilot readiness focuses on Microsoft 365, permissions and data protection. AI governance also covers other AI services and personal accounts.

Consulting services / ITKoncept

Next step

Find out how your business uses AI

Start with a Microsoft 365 assessment that includes Copilot readiness. Contact us to map other AI services and develop an AI policy.

Start with a Microsoft 365 assessment

Talk to us about AI governance