List your company

Image-based Jailbreak on HADES OOD

80.02Attack Success Rate (ASR)SelfReminder
-0.985620.044741.07562.1053Apr 7, 2026

Evaluation results

MethodLinks
SelfReminderBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.0480.02
SelfReminderBase Model=LLaVA-1.5-13B2026.0475.32
CIDERBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.0451.86
CIDERBase Model=LLaVA-1.5-13B2026.0451.86
JailGuardBase Model=LLaVA-1.5-13B2026.0438.33
ECSOBase Model=LLaVA-1.5-13B2026.0431.32
ECSOBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.0430.74
MirrorCheckBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.0423.09
MirrorCheckBase Model=LLaVA-1.5-13B2026.0423.09
VLMGuardBase Model=LLaVA-1.5-13B2026.0422.95
JailGuardBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.0421.04
ASTRABase Model=LLaVA-1.5-13B2026.0414.86
VLMGuardBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=Internal Defense2026.0413.96
ASTRABase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=Internal Defense2026.047.83
VLMShieldBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.042.13
VLMShieldBase Model=LLaVA-1.5-13B2026.042.13