Microsoft Entra security for AI overview
In brief
The documentation now distinguishes the Agent identity platform’s authentication and authorization role from the Agent 365 registry’s inventory role. It also limits Entra logging guidance to identity activity and states that permission inheritance for fleets requires consent and explicit configuration.
What Entra admins need to know
Administrators managing agent fleets must provide consent and configure permission inheritance; application and downstream-service activity must be monitored separately.
This editorial summary was generated by AI from the documentation changes. Verify important details in the full Microsoft Learn article.
Documentation change
The comparison below shows only the changed extract. Use the full-page view for complete context.
Register and manage agents: Create and manage agent identity blueprints as templates and agent identities as individual instances with parent-child relationships, enabling centralized metadata management and automatic organization into security collections.
Assign secure, scalable identities: The
Microsoft Entra Agent identity platform enables you to assign identities to agents, autodiscover them across your organization, and manage all agent metadata in one place including capabilities, tasks, and protocols. ItMicrosoft Entra Agent identity platform providesagent-to-agent discoverythe authentication and authorizationbased on standard protocols such as Model Context Protocol (MCP)foundation for agents. The Microsoft Agent 365 registry provides the broader agent inventory andagent-to-agent (A2A).metadata experience.Log and monitor
agentidentity activity:AllUse Microsoft Entra sign-in and directory audit logs to investigate authentication and identity-management operations. Instrument the agent and its downstream services separately to observe application actionsperformed by agents are logged in Microsoft Entra IDandviewable through the Microsoft Entra admin center for compliance and audit purposes.results. See Sign-in and audit logs for agents.
For more information about agent identities as identity constructs, including how they differ from application and user identities, see What are agent identities?What are agent identities?.
Control AI access with Zero Trust
- Govern agent identities at scale, from deployment to expiration.
- Ensure sponsors and owners are assigned and maintained for each agent identity, preventing orphaned agent identities.
- Enforce that agent access to resources is intentional, auditable, and time-bound through access packages.
- Apply
Conditional Access rules, permissions,supported policy and governance controls at theblueprint level so all currentblueprint level. API permission inheritance is separate: fleets require consent andfuture agent instances inherit them automatically, with the ability to disable an entire class ofexplicit inheritance configuration, while single-instance agentsin a single operation.receive grants directly. - Maintain a complete inventory of all agent identities through centralized discovery and management in the Microsoft Entra admin center and Microsoft Graph, preventing shadow AI and enabling organizations to track agents across the full lifecycle — from registration and credential management through deactivation and decommissioning.
For more information, see Identity governance for agents.
Related content
@@ -2,7 +2,7 @@ title: Microsoft Entra security for AI overview titleSuffix: Microsoft Entra Agent ID description: Learn how Microsoft Entra provides identity-based security controls for AI agents, applications, and services through authentication, governance, and Zero Trust policy enforcement.-ms.date: 05/08/2026+ms.date: 11/05/2026 ms.topic: concept-article ai-usage: ai-assisted @@ -66,11 +66,11 @@ Microsoft Entra Agent ID enables organizations to: - **Register and manage agents**: Create and manage agent identity blueprints as templates and agent identities as individual instances with parent-child relationships, enabling centralized metadata management and automatic organization into security collections. -- **Assign secure, scalable identities**: The [Microsoft Entra Agent identity platform](identity-platform/what-is-agent-id-platform.md) enables you to assign identities to agents, autodiscover them across your organization, and manage all agent metadata in one place including capabilities, tasks, and protocols. It provides agent-to-agent discovery and authorization based on standard protocols such as Model Context Protocol (MCP) and agent-to-agent (A2A).+- **Assign secure, scalable identities**: The [Microsoft Entra Agent identity platform](what-is-agent-id-platform.md) provides the authentication and authorization foundation for agents. The [Microsoft Agent 365 registry](agent-registry-convergence.md) provides the broader agent inventory and metadata experience. -- **Log and monitor agent activity**: All authentication and actions performed by agents are logged in Microsoft Entra ID and viewable through the Microsoft Entra admin center for compliance and audit purposes.+- **Log and monitor identity activity**: Use Microsoft Entra sign-in and directory audit logs to investigate authentication and identity-management operations. Instrument the agent and its downstream services separately to observe application actions and results. See [Sign-in and audit logs for agents](sign-in-audit-logs-agents.md). -For more information about agent identities as identity constructs, including how they differ from application and user identities, see [What are agent identities?](identity-platform/what-are-agent-identities.md).+For more information about agent identities as identity constructs, including how they differ from application and user identities, see [What are agent identities?](what-are-agent-identities.md). ## Control AI access with Zero Trust @@ -103,7 +103,7 @@ As organizations deploy more AI agents, identity governance becomes critical to - Govern agent identities at scale, from deployment to expiration. - Ensure sponsors and owners are assigned and maintained for each agent identity, preventing orphaned agent identities. - Enforce that agent access to resources is intentional, auditable, and time-bound through access packages.-- Apply Conditional Access rules, permissions, and governance controls at the [blueprint level](identity-platform/agent-blueprint.md) so all current and future agent instances inherit them automatically, with the ability to disable an entire class of agents in a single operation.+- Apply supported policy and governance controls at the [blueprint level](agent-blueprint.md). API permission inheritance is separate: fleets require consent and [explicit inheritance configuration](concept-inheritable-permissions.md), while single-instance agents receive grants directly. - Maintain a complete inventory of all agent identities through centralized discovery and management in the Microsoft Entra admin center and [Microsoft Graph](/graph/api/resources/agentidentity), preventing shadow AI and enabling organizations to track agents across the full lifecycle — from registration and credential management through deactivation and decommissioning. For more information, see [Identity governance for agents](/entra/id-governance/agent-id-governance-overview).@@ -133,4 +133,4 @@ For more information, see [Secure Web and AI Gateway for agents](/entra/global-s ## Related content - [What is Microsoft Entra Agent ID?](what-is-microsoft-entra-agent-id.md)-- [What is the Microsoft Entra Agent identity platform?](identity-platform/what-is-agent-id-platform.md)+- [What is the Microsoft Entra Agent identity platform?](what-is-agent-id-platform.md) 