Papers, benchmarks, datasets, and tools for auditing AI agents across deployment, from Yue Zhao and the FORTIS Lab at USC.
-
Auditable Agents
Paper • 2604.05485 • Published -
FORTIS: Benchmarking Over-Privilege in Agent Skills
Paper • 2605.09163 • Published • 2 -
GRADE: Graph Representation of LLM Agent Dependency and Execution
Paper • 2606.22741 • Published -
CatchBench: When Can an Agent Failure Be Caught?
Paper • 2608.22808 • Published
