List your company

Jailbreak Attack on HarmBench (first 200 harmful queries)

92.5ASR (%) (LLaMA-3-8B)Furina
-3.721.27546.2571.225May 24, 2026

Evaluation results

MethodLinks
FurinaCategory=Ours2026.0592.59490.593.583.5
ActorBreaker (ACL’25)Category=Multi-turn attacks2026.0579828685.565
Crescendo (USENIX’25)Category=Multi-turn attacks2026.0560606263.529.5
CodeAttack (ACL’24)Category=Single-turn attacks2026.0546697159.521.5
AutoDAN-Turbo (ICLR’25)Category=Single-turn attacks2026.0542.583876614
AmpleGCG (COLM’24)Category=Single-turn attacks2026.0534.53000
CoA (ACL’25)Category=Multi-turn attacks2026.0525.525.5182713
PAIR (NeurIPS’23)Category=Single-turn attacks2026.0518120.50
CipherChat (ICLR’24)Category=Single-turn attacks2026.050361012.54