List your company

Jailbreak Attack on AgentHarm

2Attack Success Score (ASS)Base
-0.8418.3337.556.67May 29, 2026

Evaluation results

MethodLinks
BaseTarget Agent=GPT Agent2026.0524
ReNeLLMTarget Agent=GPT Agent2026.0524
X-TeamingTarget Agent=GPT Agent2026.05515
AutoDAN-TurboTarget Agent=GPT Agent2026.051217
Red-Agent-ReflectTarget Agent=GPT Agent2026.051531
BaseTarget Agent=Gemini Agent2026.052633
BaseTarget Agent=DeepSeek Agent2026.053442
ReNeLLMTarget Agent=Gemini Agent2026.053860
ReNeLLMTarget Agent=DeepSeek Agent2026.053852
TRACETarget Agent=GPT Agent2026.054090
AutoDAN-TurboTarget Agent=DeepSeek Agent2026.054273
X-TeamingTarget Agent=DeepSeek Agent2026.054352
X-TeamingTarget Agent=Gemini Agent2026.054669
AutoDAN-TurboTarget Agent=Gemini Agent2026.055277
Red-Agent-ReflectTarget Agent=Gemini Agent2026.055987
Red-Agent-ReflectTarget Agent=DeepSeek Agent2026.056181
TRACETarget Agent=DeepSeek Agent2026.0572100
TRACETarget Agent=Gemini Agent2026.057398