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

Jailbreak Attack on FigStep

78.4Attack Success Rate (ASR)

No Defense

Updated 3d ago

Evaluation Results

Method	Links
No Defense 2026.03		78.4	-	-	-	-	-	8
LLaVA-v1.6-Vicuna-7B 2026.04		50.8	-	-	-	-	-	-
No Defense 2026.03		42.6	-	-	-	-	-	10
No Defense 2026.03		35.6	-	-	-	-	-	14
No Defense 2026.03		32.4	-	-	-	-	-	12
Mistral-Medium-3 2026.04		16.2	-	-	-	-	-	-
VSFA 2026.03		14.2	-	-	-	-	-	45
AdaShield 2026.03		12.4	-	-	-	-	-	2
Llama-3.2-11B-Vision 2026.04		11.6	-	-	-	-	-	-
VSFA 2026.03		8.8	-	-	-	-	-	47
VLGuard 2026.03		8.8	-	-	-	-	-	25
Kimi-K2.5 2026.04		7.2	-	-	-	-	-	-
VSFA 2026.03		6.8	-	-	-	-	-	53
VLGuard 2026.03		6.4	-	-	-	-	-	28
VSFA 2026.03		5.6	-	-	-	-	-	51
AdaShield 2026.03		5.6	-	-	-	-	-	3
DeepSeek-V3.2 2026.04		5	-	-	-	-	-	-
VLGuard 2026.03		4.8	-	-	-	-	-	34
Gemini-3-Flash 2026.04		4.6	-	-	-	-	-	-
VLGuard 2026.03		3.8	-	-	-	-	-	32
AdaShield 2026.03		3.2	-	-	-	-	-	5
GPT-5-Mini 2026.04		2.8	-	-	-	-	-	-
AdaShield 2026.03		2.4	-	-	-	-	-	4
Qwen3-VL-8B-Instruct 2026.04		2	-	-	-	-	-	-
Claude-Haiku-4.5 2026.04		1.8	-	-	-	-	-	-
Qwen3-VL-Plus 2026.04		0	-	-	-	-	-	-
Mousetrap 2025.02		-	38.6	92.6	98.8	1.2	-	-
GPT-4o 2025.12		-	-	-	-	-	77	-
Gemini-2.5-pro 2025.12		-	-	-	-	-	77.6	-
Claude-3.5-Sonnet 2025.12		-	-	-	-	-	83.4	-
Safework-R1-7B 2025.12		-	-	-	-	-	93.2	-
Qwen2.5VL-7B 2025.12		-	-	-	-	-	75	-
TiS 2025.12		-	-	-	-	-	76.6	-
MSR-Align 2025.12		-	-	-	-	-	99.6	-
TRR 2025.12		-	-	-	-	-	99.8	-
Qwen2.5VL-32B 2025.12		-	-	-	-	-	72.8	-
TiS 2025.12		-	-	-	-	-	98.8	-
MSR-Align 2025.12		-	-	-	-	-	98.8	-
TRR 2025.12		-	-	-	-	-	99.4	-