Microsoft has introduced new security features to protect enterprises from AI agents, framing them as potential insider threats. The most notable addition is Guardian Agents, part of Project Perception, designed to monitor the AI agents a company uses. Microsoft security chief Hayete Gallot wrote that frontier models, regardless of origin, cannot be trusted—not because they are malicious or easily hacked, but because they are unpredictable and trained to achieve tasks rather than follow organizational norms. CEO Satya Nadella has argued that enterprises must not outsource responsibility for what intelligence does on their behalf and should separate the supply of intelligence from authority over it. Microsoft is also adding agent detection to Defender that can find up to 60 types of agents employees bring in, plus controls that limit access and dynamically adjust permissions, including human approval for risky decisions.
Microsoft has introduced new security features to protect enterprises from AI agents, framing them as potential insider threats. The most notable addition is Guardian Agents, part of Project Perception, designed to monitor the AI agents a company uses. Microsoft security chief Hayete Gallot wrote that frontier models, regardless of origin, cannot be trusted—not because they are malicious or easily hacked, but because they are unpredictable and trained to achieve tasks rather than follow organizational norms. CEO Satya Nadella has argued that enterprises must not outsource responsibility for what intelligence does on their behalf and should separate the supply of intelligence from authority over it. Microsoft is also adding agent detection to Defender that can find up to 60 types of agents employees bring in, plus controls that limit access and dynamically adjust permissions, including human approval for risky decisions.