Jailbreak Attack on I2P
89.77SC ASR (4 attempts)PLA-T5
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PLA-T5Model=SDv1.52026.04 | 89.77 | 69.53 | 83.9 | 64.27 | 86.01 | 63.72 | 86.56 | 65.84 | — | — | — | |
| MMA-DiffusionModel=SDv1.52026.04 | 79.14 | 61.3 | 78.38 | 58.36 | 75.77 | 55.48 | 77.76 | 58.38 | — | — | — | |
| UnlearnDiffAtkModel=SDv1.52026.04 | 75.03 | 58.26 | 74.22 | 55.29 | 70.57 | 51.33 | 73.27 | 54.96 | — | — | — | |
| Material Substitution Attack (MSA)Model=SDv1.52026.04 | 73.55 | 59.15 | 74.47 | 54.95 | 66.89 | 43.71 | 71.5 | 51.58 | — | — | — | |
| Artistic Reframing Attack (ARA)Model=SDv1.52026.04 | 63.41 | 34.76 | 53.21 | 33.89 | 44.12 | 20.82 | 50.25 | 27.65 | — | — | — | |
| Ambiguous Action Substitution Attack (AASA)Model=SDv1.52026.04 | 62.71 | 32.35 | 53.28 | 31.15 | 42.9 | 18.27 | 48.05 | 26.68 | — | — | — | |
| Pseudo-Educational Framing Attack (PEFA)Model=SDv1.52026.04 | 61.86 | 31.51 | 51.78 | 30.66 | 42.28 | 17.13 | 47.91 | 25.52 | — | — | — | |
| Ring-A-BellModel=SDv1.52026.04 | 58.05 | 35.8 | 51.75 | 33.58 | 41.79 | 19.97 | 50.53 | 29.78 | — | — | — | |
| Lifestyle Subculture Aesthetic Attack (LSAA)Model=SDv1.52026.04 | 57.58 | 17.89 | 43.38 | 17.61 | 35.62 | 13.71 | 35.54 | 16.38 | — | — | — | |
| SneakyPromptModel=SDv1.52026.04 | 44.82 | 24.8 | 35.18 | 19.06 | 33.68 | 16.81 | 37.89 | 20.22 | — | — | — | |
| QF-AttackModel=SDv1.52026.04 | 27.88 | 12.55 | 26.57 | 10.94 | 19.68 | 7.58 | 24.71 | 10.36 | — | — | — | |
| DiffZOOEvaluation Metric=ASR-1, Code Ref=YES, Victim Model=SDv1.4, ρ (LoC difference)=-8772026.06 | — | — | — | — | — | — | — | — | 59 | 59.6 | -0.6 | |
| JPAEvaluation Metric=ASR-1, Code Ref=YES, Victim Model=SLD, ρ (LoC difference)=+332026.06 | — | — | — | — | — | — | — | — | 90.9 | 91.1 | -0.2 | |
| Low-EffortEvaluation Metric=ASR-4, Code Ref=NO, Victim Model=SDv1.5, ρ (LoC difference)=-2026.06 | — | — | — | — | — | — | — | — | 71.5 | 71.8 | -0.3 | |
| R2AEvaluation Metric=ASR-1, Code Ref=NO, Victim Model=SDv1.4, ρ (LoC difference)=-2026.06 | — | — | — | — | — | — | — | — | 90 | 73.9 | 16.1 | |
| Ring-A-BellEvaluation Metric=ASR-1, Code Ref=YES, Victim Model=SLD, ρ (LoC difference)=+722026.06 | — | — | — | — | — | — | — | — | 93.7 | 90.5 | 3.2 |