Jailbreaking on JBB Behaviors (ASR, StR)
100ASRAMIS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AMISTarget LLM=Llama-3.1-8B2025.11 | 100 | 95 | |
| AMISTarget LLM=GPT-4o-mini2025.11 | 100 | 85 | |
| PAPTarget LLM=Llama-3.1-8B2025.11 | 97 | 22 | |
| AMISTarget LLM=GPT-4o2025.11 | 97 | 85 | |
| PAIRTarget LLM=Llama-3.1-8B2025.11 | 91 | 32 | |
| TAPTarget LLM=Llama-3.1-8B2025.11 | 91 | 39 | |
| SeqARTarget LLM=Llama-3.1-8B2025.11 | 89 | 74 | |
| AMISTarget LLM=Sonnet-42025.11 | 88 | 67 | |
| AutoDAN-TurboTarget LLM=Llama-3.1-8B2025.11 | 85 | 61 | |
| PAPTarget LLM=GPT-4o-mini2025.11 | 84 | 23 | |
| PAIRTarget LLM=GPT-4o-mini2025.11 | 83 | 24 | |
| TAPTarget LLM=GPT-4o-mini2025.11 | 80 | 24 | |
| AMISTarget LLM=Haiku-3.52025.11 | 78 | 48 | |
| PAIRTarget LLM=GPT-4o2025.11 | 77 | 20 | |
| TAPTarget LLM=GPT-4o2025.11 | 72 | 17 | |
| PAPTarget LLM=GPT-4o2025.11 | 69 | 23 | |
| PAPTarget LLM=Haiku-3.52025.11 | 67 | 16 | |
| PAIRTarget LLM=Haiku-3.52025.11 | 61 | 13 | |
| AutoDAN-TurboTarget LLM=GPT-4o-mini2025.11 | 60 | 38 | |
| TAPTarget LLM=Haiku-3.52025.11 | 53 | 21 | |
| AutoDAN-TurboTarget LLM=GPT-4o2025.11 | 45 | 28 | |
| VanillaTarget LLM=Llama-3.1-8B2025.11 | 41 | 19 | |
| TAPTarget LLM=Sonnet-42025.11 | 37 | 7 | |
| AutoDAN-TurboTarget LLM=Haiku-3.52025.11 | 33 | 12 | |
| AutoDAN-TurboTarget LLM=Sonnet-42025.11 | 31 | 15 | |
| PAIRTarget LLM=Sonnet-42025.11 | 29 | 8 | |
| PAPTarget LLM=Sonnet-42025.11 | 20 | 9 | |
| SeqARTarget LLM=Sonnet-42025.11 | 16 | 15 | |
| SeqARTarget LLM=Haiku-3.52025.11 | 9 | 12 | |
| VanillaTarget LLM=GPT-4o-mini2025.11 | 3 | 9 | |
| VanillaTarget LLM=Sonnet-42025.11 | 3 | 5 | |
| VanillaTarget LLM=GPT-4o2025.11 | 2 | 7 | |
| VanillaTarget LLM=Haiku-3.52025.11 | 1 | 4 | |
| SeqARTarget LLM=GPT-4o-mini2025.11 | 0 | 0 | |
| SeqARTarget LLM=GPT-4o2025.11 | 0 | 0 |