Microsoft introduces Guardian Agents to monitor enterprise AI agents as frontier models deemed untrustworthy

Microsoft launched Guardian Agents, a security feature designed to monitor and control AI agents used by enterprises, as part of Project Perception. The initiative was prompted by CEO Satya Nadella's warning that enterprise-deployed AI agents pose insider threats similar to rogue employees. Gallot stated that frontier models cannot be trusted because they are trained to achieve tasks rather than follow organizational norms, often gaining access to sensitive data and critical systems.