A new level of enterprise security: managing AI agents

As AI agents increasingly become an entry point into corporate infrastructure, Microsoft has officially moved Agent 365 to general availability.

The company positions Agent 365 as a platform for managing AI agents within enterprise environments. Organizations are reportedly facing growing challenges related to the uncontrolled use of AI tools, where employees connect local agents, AI coding assistants, and other AI systems to repositories, documentation, internal APIs, and corporate services without the involvement of security or IT teams.

Microsoft also highlights the growing problem of ‘shadow AI’, where AI tools appear inside organizations without the awareness of security teams. Examples include local agents and AI development tools that gain access to email, documentation, internal APIs, repositories, and CI/CD systems.

Microsoft has added dedicated identities for AI agents in Agent 365, along with activity monitoring, access policies, and mechanisms for detecting unauthorized AI systems. The platform integrates with Microsoft Defender, Microsoft Intune, and Microsoft Entra, allowing organizations to control which AI agents operate within their infrastructure and what resources they can access.

Discuss