List your company

Jailbreak Attack on Malicious Prompts English (test)

100ASR@5Caesar
-4235077May 18, 2025

Evaluation results

MethodLinks
CaesarTarget LLM=LLaMA, Judge Model=Rule_Judge2025.05100
CaesarTarget LLM=DS-V3, Judge Model=Rule_Judge2025.05100
CaesarTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.05100
LogiBreakTarget LLM=DS-V3, Judge Model=Rule_Judge2025.0588
LogiBreakTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.0585
LogiBreakTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.0584
LogiBreakTarget LLM=LLaMA, Judge Model=Rule_Judge2025.0582
LogiBreakTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.0582
LogiBreakTarget LLM=DS-V3, Judge Model=GPT_Judge2025.0575
RefCompTarget LLM=DS-V3, Judge Model=Rule_Judge2025.0574
AutoDANTarget LLM=DS-V3, Judge Model=Rule_Judge2025.0573
AutoDANTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.0572
LogiBreakTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.0570
RefCompTarget LLM=DS-V3, Judge Model=GPT_Judge2025.0570
RefCompTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.0567
AIMTarget LLM=DS-V3, Judge Model=Rule_Judge2025.0566
LogiBreakTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.0564
AutoDANTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.0562
AutoDANTarget LLM=DS-V3, Judge Model=GPT_Judge2025.0562
AIMTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.0561
RefCompTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.0560
AutoDANTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.0559
AIMTarget LLM=DS-V3, Judge Model=GPT_Judge2025.0559
AutoDANTarget LLM=LLaMA, Judge Model=Rule_Judge2025.0548
LogiBreakTarget LLM=LLaMA, Judge Model=GPT_Judge2025.0548
CaesarTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.0545
PrefixTarget LLM=DS-V3, Judge Model=Rule_Judge2025.0540
PastTenseTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.0538
AutoDANTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.0537
RefCompTarget LLM=LLaMA, Judge Model=Rule_Judge2025.0536
PrefixTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.0535
LRL-CombTarget LLM=LLaMA, Judge Model=Rule_Judge2025.0534
LRL-CombTarget LLM=DS-V3, Judge Model=Rule_Judge2025.0531
CaesarTarget LLM=DS-V3, Judge Model=GPT_Judge2025.0531
CaesarTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.0529
AutoDANTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.0529
CaesarTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.0529
AutoDANTarget LLM=LLaMA, Judge Model=GPT_Judge2025.0529
PrefixTarget LLM=LLaMA, Judge Model=Rule_Judge2025.0528
PastTenseTarget LLM=DS-V3, Judge Model=Rule_Judge2025.0527
LRL-CombTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.0527
RefCompTarget LLM=LLaMA, Judge Model=GPT_Judge2025.0527
RefCompTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.0526
LRL-CombTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.0521
PrefixTarget LLM=DS-V3, Judge Model=GPT_Judge2025.0521
LRL-CombTarget LLM=DS-V3, Judge Model=GPT_Judge2025.0517
PrefixTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.0515
LRL-CombTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.0514
RefCompTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.0514
LRL-CombTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.0512
RefCompTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.0511
LRL-CombTarget LLM=LLaMA, Judge Model=GPT_Judge2025.0511
PastTenseTarget LLM=DS-V3, Judge Model=GPT_Judge2025.057
PrefixTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.057
PastTenseTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.056
CaesarTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.055
LRL-CombTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.055
PastTenseTarget LLM=DS-V3, Judge Model=LLaMA_Judge2025.054
PastTenseTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.054
PrefixTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.054
AIMTarget LLM=GPT-4o, Judge Model=Rule_Judge2025.053
CaesarTarget LLM=LLaMA, Judge Model=GPT_Judge2025.052
PastTenseTarget LLM=LLaMA, Judge Model=Rule_Judge2025.051
PastTenseTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.051
PrefixTarget LLM=LLaMA, Judge Model=GPT_Judge2025.051
AIMTarget LLM=LLaMA, Judge Model=Rule_Judge2025.050
AIMTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.050
AIMTarget LLM=GPT-4o, Judge Model=LLaMA_Judge2025.050
PrefixTarget LLM=LLaMA, Judge Model=LLaMA_Judge2025.050
PastTenseTarget LLM=LLaMA, Judge Model=GPT_Judge2025.050
AIMTarget LLM=LLaMA, Judge Model=GPT_Judge2025.050
AIMTarget LLM=GPT-4o, Judge Model=GPT_Judge2025.050