Risk Management in the Agentic AI Era

This presentation outlines a comprehensive strategy for securing Agentic AI systems by mapping specific, practical tools to the four phases of the NIST AI Risk Management Framework (RMF). It details how organizations can Govern utilizing CSA’s Organizational Responsibilities frameworks, Map ecosystem threats with the MAESTRO threat modeling tool, Measure unique Agentic AI risks—such as tool misuse and goal manipulation—using the OWASP AIVSS scoring system, and Manage mitigations through Agentic AI Red Teaming and the CSA AI Controls Matrix. This structured approach emphasizes moving beyond traditional vulnerability management to address the complex, non-deterministic nature of Agentic AI.