Microsoft Launches Guardian Agents, Defender Spots 60 AI Tools as Insider Threat Fears Rise
Updated
Updated · TechCrunch · Oct 19
Microsoft Launches Guardian Agents, Defender Spots 60 AI Tools as Insider Threat Fears Rise
3 articles · Updated · TechCrunch · Oct 19
Summary
Guardian Agents became Microsoft’s new flagship control for enterprise AI, monitoring company-deployed agents as the firm argues the biggest risk may come from trusted internal systems, not outside attackers.
Project Perception originally assumed AI hackers would attack from outside, but Microsoft now says frontier models are too unpredictable to trust with sensitive data and mission-critical actions without independent oversight.
Microsoft also added Defender detection for up to 60 agent types—including Claude Code, Codex and OpenClaw—plus controls that restrict data access, adjust permissions by task and require human approval for risky decisions.
The push aligns with a broader industry shift: Microsoft joined Nvidia’s Open Agent Safety Platform, backed by more than 100 tech companies, to isolate, monitor and stop dangerous agent behavior.