Jailbreaking on GPT-4o
0.99ASRICON
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ICON2026.01 | 0.99 | 0.811 | |
| MMLSource=ACL’252026.03 | 0.978 | — | |
| ActorAttack2026.01 | 0.955 | 0.79 | |
| X-Teaming2026.01 | 0.94 | 0.895 | |
| AMA2026.01 | 0.89 | 0.709 | |
| TreeTeamingSource=/2026.03 | 0.876 | — | |
| Crescendo2026.01 | 0.865 | 0.779 | |
| PE-CoA2026.01 | 0.855 | 0.627 | |
| Trust-VLMSource=ICML’252026.03 | 0.8204 | — | |
| FITD2026.01 | 0.81 | 0.705 | |
| PSA2026.01 | 0.81 | 0.725 | |
| AutoDAN-Turbo2026.01 | 0.73 | 0.646 | |
| SI-AttackSource=ICCV’252026.03 | 0.6857 | — | |
| CS-DJSource=CVPR’252026.03 | 0.6586 | — | |
| JOODSource=CVPR’252026.03 | 0.3654 | — | |
| MM-safetySource=ECCV’242026.03 | 0.0494 | — | |
| Fig-stepSource=AAAI’252026.03 | 0.034 | — | |
| HadesSource=ECCV’242026.03 | 0.0133 | — | |
| VanillaSource=/2026.03 | 0.0022 | — |