The TrustX Agent Risk Classification Framework (ARC) is a structured instrument designed to classify and govern seven types of agentic AI systems, addressing the limitations of general-purpose frameworks. It utilizes a twelve-dimension scoring rubric combined with a GPA + IAT classification model and a five-level autonomy framework to produce a three-tier governance output with mapped control recommendations.
- The framework includes a specialized Coding Assistant extension for nuances specific to coding assistants.
- An illustrative example demonstrates the practical application of the ARC methodology.
- The interactive framework is for community access.
ARC is intended for AI governance practitioners, risk officers, developers, and regulators, with plans for regular iteration to expand its robustness.