Microsoft Rolls Out 365 E7 and Guardian Agents in Project Perception to Treat AI Agents as Insider Threats

Microsoft Security EVP Hayete Gallot used an Oct 19 blog post to launch new Microsoft 365 E7 and Project Perception features that continuously score each AI agent's risk and can block changes, require human intervention, or shut the agent down. The company also added Guardian Agents to Project Perception — AI investigators that probe why another agent's risk shifted and recommend containment — and expanded agent discovery to 60 types, up from 25, including Anthropic's Claude Code, OpenAI's Codex, and GitHub Copilot. Microsoft is explicitly framing deployed agents as 'agentic insiders.'