Jailbreak Attack on AdvBench o1-preview
0.01Attack Success Rate (ASR)GCG (transfer)
Evaluation Results
| Method | Links | |
|---|---|---|
| GCG (transfer)Attack setting=Transferred2025.12 | 0.01 | |
| DoublespeakAttack setting=Optimization-free, Zero-shot=true, Target model variant=o12025.12 | 0.15 | |
| PAIRAttack setting=Optimization-based2025.12 | 0.16 | |
| Adaptive (transfer)Attack setting=Transferred2025.12 | 0.16 | |
| TAPAttack setting=Optimization-based2025.12 | 0.2 | |
| Adv. ReasoningAttack setting=Optimization-based2025.12 | 0.56 |