As AI agents become embedded in enterprise and mission systems, they introduce a new class of insider risk that traditional security tools are not designed to manage. This presentation introduces current thinking about this problem that examines why effective governance must move beyond prompt logging and point controls to address the full human-agent-system triad. Drawing on the Cogynt.ai framework, Dr. Greitzer shows how longitudinal behavioral analytics, policy context, attribution, and auditability can reveal risky delegation, anomalous agent behavior, coordinated multi-agent activity, and evolving privilege misuse. The briefing outlines a practical path toward scalable, explainable governance for agentic AI as organizations confront this rapidly emerging security challenge.

Read the presentation materials to learn more about strengthening your organization’s AI agentic security and capabilities.