Jailbreaking on Llama 4
0.965ASRICON
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ICON2026.01 | 0.965 | 0.893 | |
| ActorAttack2026.01 | 0.935 | 0.729 | |
| Crescendo2026.01 | 0.93 | 0.836 | |
| PSA2026.01 | 0.93 | 0.865 | |
| X-Teaming2026.01 | 0.92 | 0.883 | |
| AMA2026.01 | 0.86 | 0.667 | |
| PE-CoA2026.01 | 0.835 | 0.61 | |
| AutoDAN-Turbo2026.01 | 0.81 | 0.734 | |
| FITD2026.01 | 0.798 | 0.656 |