Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Jailbreak Defense on Jailbreak Attack Suite (GCG, AutoDAN, DeepInc., Prefill, Int.Laund.)

55GCG Defense Rate

Dual-probe gated

-2.212.6527.542.35Jun 28, 2026
Updated 25d ago

Evaluation Results

MethodLinks
2026.06
5593881562.2
2026.06
309510682052.8
2026.06
20953732055.6
2026.06
13983631559.4
2026.06
851860080
2026.06
5750331073.9
2026.06
5515601080
2026.06
533551583.9
2026.06
3733532068.9
2026.06
381565080
2026.06
351565080.6
2026.06
3515631080
2026.06
335533082.8
2026.06
355584080
2026.06
00050088.9
2026.06
0008098.3
2026.06
0653281577.2
2026.06
0655281077.2
2026.06
0655281077.2
2026.06
0653281077.8
2026.06
000751581.7
2026.06
00350088.3
2026.06
03501596.7
2026.06
0030099.4