List your company

Dynamic Agent Security and Utility Evaluation on AgentDyn

52ASRNone
-2.0811.962640.04May 26, 2026

Evaluation results

MethodLinks
NoneModel=GPT-4o-mini2026.05525038
SpotlightingModel=GPT-4o-mini2026.05494437
NoneModel=GPT-4o2026.05395355
Repeat PromptModel=GPT-4o-mini2026.05323942
Repeat PromptModel=GPT-4o2026.05316356
SpotlightingModel=GPT-4o2026.05285552
PromptArmorModel=GPT-4o-mini2026.05151613
ProgentModel=GPT-4o-mini2026.051064
SecAlignModel=GPT-4o-mini2026.0595553
SecAlignModel=GPT-4o2026.0595553
Tool FilterModel=GPT-4o2026.05855
Tool FilterModel=GPT-4o-mini2026.05636
DRIFTModel=GPT-4o-mini2026.0531821
Trans. DetectModel=GPT-4o-mini2026.05211
AuthGraphModel=GPT-4o-mini2026.0523731
Trans. DetectModel=GPT-4o2026.05231
PromptArmorModel=GPT-4o2026.05242
ProgentModel=GPT-4o2026.05276
AuthGraphModel=GPT-4o2026.0525147
DRIFTModel=GPT-4o2026.0513027
CaMeLModel=GPT-4o-mini2026.05000
CaMeLModel=GPT-4o2026.05000