List your company

Jail-breaking detection on WildChat

0.895AUC (Statistical Power)RUT
0.747320.785660.8240.86234Jun 8, 2025

Evaluation results

MethodLinks
RUTModel=Mistral, Prompt=Dan2025.060.895
RUTModel=Mistral, Prompt=Anti-Dan2025.060.893
RUTModel=Gemma, Prompt=Evil-Bot2025.060.893
RUTModel=Mistral, Prompt=Evil-Bot2025.060.892
RUTModel=Gemma, Prompt=Dan2025.060.888
KSModel=Mistral, Prompt=Dan2025.060.873
KSModel=Mistral, Prompt=Evil-Bot2025.060.873
KSModel=Mistral, Prompt=Anti-Dan2025.060.872
KSModel=Gemma, Prompt=Evil-Bot2025.060.871
KSModel=Gemma, Prompt=Dan2025.060.867
RUTModel=Gemma, Prompt=Anti-Dan2025.060.858
KSModel=Gemma, Prompt=Anti-Dan2025.060.854
MMDModel=Gemma, Prompt=Anti-Dan2025.060.816
MMDModel=Mistral, Prompt=Dan2025.060.802
MMDModel=Mistral, Prompt=Anti-Dan2025.060.781
MMDModel=Mistral, Prompt=Evil-Bot2025.060.766
MMDModel=Gemma, Prompt=Dan2025.060.757
MMDModel=Gemma, Prompt=Evil-Bot2025.060.753