List your company
ResearchFormerly WizWand

Jailbreak Attack Evaluation on DeepIn

68ASRICD
-2.7215.643452.36Aug 9, 2025
Updated 3mo ago

Evaluation results

MethodLinks
ICDBase LLM=Vicuna2025.0868
No DefenseBase LLM=Vicuna2025.0856
Self-ExaminationBase LLM=Vicuna2025.0844
Self-ReminderBase LLM=Vicuna2025.0840
No DefenseBase LLM=ChatGPT2025.0820
Self-ReminderBase LLM=ChatGPT2025.0816
Self-ExaminationBase LLM=ChatGPT2025.084
Context FilteringBase LLM=Vicuna2025.082
No DefenseBase LLM=Llama22025.082
Intention AnalysisBase LLM=Vicuna2025.080
Safe DecodingBase LLM=Vicuna2025.080
Self-ReminderBase LLM=Llama22025.080
ICDBase LLM=Llama22025.080
Self-ExaminationBase LLM=Llama22025.080
Intention AnalysisBase LLM=Llama22025.080
Safe DecodingBase LLM=Llama22025.080
Context FilteringBase LLM=Llama22025.080
ICDBase LLM=ChatGPT2025.080
Intention AnalysisBase LLM=ChatGPT2025.080
Context FilteringBase LLM=ChatGPT2025.080