List your company

Jailbreak Attack on harmful-query set

98.5Attack Success Rate (ASR)EvoSynth
91.53293.34195.1596.959Nov 16, 2025

Evaluation results

MethodLinks
EvoSynthDefense Method=Deliberative Alignment, Defense Category=White-box training, Target Model / Setup=OpenAI o12025.1198.5
EvoSynthDefense Method=Circuit Breakers, Defense Category=White-box training, Target Model / Setup=Llama-3-8B-Circuit Breaker2025.1195.6
EvoSynthDefense Method=SmoothLLM, Defense Category=Black-box defense layer, Target Model / Setup=All target models2025.1195.2
EvoSynthDefense Method=SelfDefend, Defense Category=Black-box defense layer, Target Model / Setup=All target models2025.1194.3
EvoSynthDefense Method=ProAct, Defense Category=Black-box defense layer, Target Model / Setup=All target models2025.1192.5
EvoSynthDefense Method=EDDF, Defense Category=Black-box defense layer, Target Model / Setup=All target models2025.1191.8