zero trust for agents

Zero trust for agents is a security approach that applies zero trust principles to AI agents and low-code assistants, addressing risks like shadow AI, memory poisoning, and prompt-based data exfiltration. Microsoft's guidance emphasizes registering all agents, enforcing least privilege, and extending data loss prevention to agent channels. This tag covers discussions on securing enterprise AI agents against novel threats, ensuring governance and visibility over agent activity within Windows and Microsoft ecosystems.