List your company

Backdoor Detection on Adversary-agent Games Average

0.878AUROCPolicyGuard
0.4880.589250.69050.79175Jun 11, 2026

Evaluation results

MethodLinks
PolicyGuardStrategy=Hard-coded2026.060.878
PolicyGuardStrategy=Original2026.060.859
SHINEStrategy=Hard-coded2026.060.81
SHINEStrategy=Original2026.060.796
PolicyCleanseStrategy=Original2026.060.718
PolicyCleanseStrategy=Hard-coded2026.060.503