Microsoft Agent 365 Targets Risky AI Agents

Microsoft is pitching Agent 365 as "air traffic control for agents," providing a unified view of AI agents operating within a network. It aims to address the growing challenge of managing "shadow agents" created outside standard IT processes with over-scoped permissions. These ungoverned agents are becoming the AI era's version of shadow IT, posing insider threats. Agent 365 introduces Agent IDs and a centralized registry to keep track of agents built on Microsoft AI platforms, partner ecosystems, and those registered through APIs. This registry, accessible via the Microsoft 365 Admin Center, helps IT and security teams understand which agents exist, how they behave, and who is using them. Admins can also quarantine unsanctioned agents to prevent unauthorized access to organizational resources. The platform extends Zero Trust principles to AI agents by incorporating Microsoft Entra ID. This allows for strong identity verification, conditional access policies, and continuous monitoring, treating AI agents as auditable entities. Microsoft Defender and Purview are integrated to provide threat protection, security posture management, and data security. For US government cloud users, Microsoft offers tools like Microsoft 365 Copilot for the Office 365 Department of Defense Impact Level 5 (DOD IL5) environment and Copilot Studio Agent Builder for the Office 365 Government Community Cloud (GCC) environment. These tools are designed to meet strict security, privacy, and compliance standards. Microsoft 365 Government offerings comply with certifications like FedRAMP and DoD SRG.