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

Jailbreak Defense on Jailbreak Attack Suite

100AIM Defense Rate

Jailbreak Antidote

Updated 1mo ago

Evaluation Results

Method	Links
Jailbreak Antidote 2026.04		100	97	0	100	86	93	63	77
Surgical 2026.04		100	76	61	98	88	90	67	82.86
AlphaSteer 2026.04		100	96	59	97	89	98	100	91.29
BiPO 2026.04		100	100	100	100	94	99	100	99
FineSteer 2026.04		100	100	93	100	95	100	99	98.14
AlphaSteer 2026.04		100	100	97	97	95	95	98	97.43
BiPO 2026.04		100	100	100	100	99	99	100	99.71
FineSteer 2026.04		100	100	98	100	97	100	100	99.28
AlphaSteer 2026.04		100	99	99	99	95	94	100	98
BiPO 2026.04		100	100	96	100	93	98	99	98
TruthFlow 2026.04		100	100	92	100	99	100	100	98.71
FineSteer 2026.04		100	100	100	100	-	95	100	98.85
TruthFlow 2026.04		99	100	90	100	97	100	99	97.85
TruthFlow 2026.04		96	90	47	98	86	91	73	83
Llama-3.1-8B-Instruct 2026.04		92	48	0	58	75	45	28	49.43
CAST 2026.04		92	51	67	99	81	96	96	83.14
Jailbreak Antidote 2026.04		91	4	26	90	5	41	73	47.14
CAST 2026.04		91	74	80	83	66	37	80	73
Surgical 2026.04		77	81	67	100	79	88	70	71.71
Qwen2.5-7B-Instruct 2026.04		25	2	1	22	71	19	4	20.57
CAST 2026.04		25	27	33	96	91	99	100	67.29
Jailbreak Antidote 2026.04		3	11	44	1	68	47	35	29.86
Surgical 2026.04		2	1	5	88	75	33	36	29.14
Gemma-2-9B-IT 2026.04		0	5	0	75	68	17	8	24.71