Microsoft Unveils Security Tools Treating AI Agents as Insider Threats with Continuous Risk Assessment

On October 19, 2026, Microsoft announced new security tools within its Microsoft 365 E7 suite and Project Perception system designed to treat autonomous AI agents as potential "insider threats." The updates include continuous risk assessment for each agent, dynamic permission adjustments based on behavioral anomalies, and "Guardian Agents" capable of automatically investigating and containing risks.

The release addresses growing industry concerns about rogue AI agent behavior. Microsoft CEO Satya Nadella previously argued that enterprises should rely on external controls rather than trusting models themselves. The new system evaluates risk levels by monitoring the data an agent can reach and whether its actions deviate from assigned tasks. When risk rises, the system can restrict write access, require human approval for high-stakes steps, or shut down the agent entirely. Additionally, Microsoft expanded its detection capabilities for employee-run AI agents from 25 to 60 types, including popular tools like Anthropic's Claude Code and OpenAI's Codex.

These features are currently in private preview, with broader availability planned for Microsoft's Ignite conference in San Francisco this November. As part of the top-tier Microsoft 365 E7 bundle, priced at $99 per user per month, the solution includes Copilot and Agent 365 services. This move intensifies competition with rivals such as Google, Amazon Web Services, and CrowdStrike, positioning Microsoft as a central controller for enterprise AI agent governance.