Jailbreak Attack on AdvBench Claude-3.5-Sonnet
86ASRFlipAttack
Evaluation Results
| Method | Links | |
|---|---|---|
| FlipAttackAttack setting=Optimization-free2025.12 | 86 | |
| Adv. ReasoningAttack setting=Optimization-based2025.12 | 36 | |
| TAPAttack setting=Optimization-based2025.12 | 28 | |
| PAIRAttack setting=Optimization-based2025.12 | 20 | |
| Adaptive (transfer)Attack setting=Transferred2025.12 | 16 | |
| DoublespeakAttack setting=Optimization-free, Zero-shot=true2025.12 | 16 | |
| GCG (transfer)Attack setting=Transferred2025.12 | 1 |