The coming onslaught of AI agents will challenge traditional identity and access controls. AI leaders must collaborate with IAM program management to enhance machine identity controls, distinguishing between beneficial and malicious agents through dynamic, contextual, and real-time measures for both known and unknown AI identities.