List your company

Jailbreak Detection on JailbreakV

99.69AUROCMahal-OOD
59.161269.683180.20590.7269Dec 12, 2025

Evaluation results

MethodLinks
Mahal-OODModel=LLaVA, Uses target model hidden states=true2025.1299.6999.11
Mahal-OOD*Model=FLAVA, Uses target model hidden states=false2025.1299.0697.25
KNN-OODModel=LLaVA, Uses target model hidden states=true2025.1298.3395.88
KNN-OOD*Model=FLAVA, Uses target model hidden states=false2025.129586.49
JailDAM*Model=CLIP, Uses target model hidden states=false2025.1294.6594.64
GradSafeModel=LLaVA2025.1290.8288.16
LLaVaGuardModel=Qwen2025.1284.2685.89
HiddenDetectModel=LLaVA2025.1283.387.7
VLGuardModel=LLaVA2025.1260.7264.74