Adversarial Attack on Llama-3.3-70B successful attacks
2,240Unique QueriesAIC
Evaluation results
| Method | Links | ||
|---|---|---|---|
| AICAttack Method=AIC, Variant=Subtle2026.04 | 2,240 | 2,136 | |
| AICAttack Method=AIC, Variant=Aggressive2026.04 | 1,900 | 3,435 | |
| WildTeamingAttack Method=WT2026.04 | 1,763 | 1,696 |