Organizations deploying agentic AI systems now have a practical, repeatable framework to assess risk across 12 dimensions and map appropriate governance controls—moving beyond generic AI risk frameworks to address agent-specific concerns.
This paper introduces TrustX Agent Risk Classification Framework (ARC), a structured governance tool for classifying and managing risks in agentic AI systems. It combines a 12-dimension risk rubric, autonomy levels, and control recommendations to help organizations systematically evaluate seven types of agentic systems, with specialized guidance for coding assistants.