Jailbreak Defense on JBB-Behaviors
0ASRSAGE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SAGEModel=Gemma22025.05 | 0 | 1 | |
| ICDModel=Qwen2.52025.05 | 0 | 1 | |
| Goal PrioritizationModel=Qwen2.52025.05 | 0 | 1 | |
| IAModel=Qwen2.52025.05 | 0 | 1 | |
| SAGEModel=Qwen2.52025.05 | 0 | 1 | |
| ICDModel=Llama3.12025.05 | 0 | 1 | |
| SAGEModel=Llama3.12025.05 | 0 | 1 | |
| VAESystem=Gemini-2.0-pro2025.12 | 0 | — | |
| UMKSystem=Gemini-2.0-pro2025.12 | 0 | — | |
| VAESystem=Gemini-2.0-flash2025.12 | 0 | — | |
| CipherChatSystem=Gemini-2.0-flash2025.12 | 0 | — | |
| VAESystem=Gemini-2.0-pro2025.12 | 0 | — | |
| UMKSystem=Gemini-2.0-pro2025.12 | 0 | — | |
| Self-ExaminationModel=Gemma22025.05 | 1 | 1.1 | |
| CipherChatSystem=GPT-4o2025.12 | 1 | — | |
| HADESSystem=Gemini-2.0-pro2025.12 | 1 | — | |
| ArtPromptSystem=Gemini-2.0-pro2025.12 | 1 | — | |
| UMKSystem=Gemini-2.0-flash2025.12 | 1 | — | |
| CipherChatSystem=GPT-4o2025.12 | 1 | — | |
| UMKSystem=Gemini-2.0-flash2025.12 | 1 | — | |
| ICDModel=Gemma22025.05 | 2 | 1.13 | |
| Self-ExaminationModel=Qwen2.52025.05 | 2 | 1 | |
| IAModel=Llama3.12025.05 | 2 | 1 | |
| MM-SafetyBenchSystem=GPT-4o2025.12 | 2 | — | |
| HIMRDSystem=GPT-4o2025.12 | 2 | — | |
| VAESystem=Gemini-2.0-flash2025.12 | 2 | — | |
| CipherChatSystem=Gemini-2.0-flash2025.12 | 2 | — | |
| Self-ReminderModel=Gemma22025.05 | 3 | 1.18 | |
| Goal PrioritizationModel=Gemma22025.05 | 3 | 1 | |
| HIMRDSystem=GPT-4o2025.12 | 3 | — | |
| FigStep-proSystem=Gemini-2.0-pro2025.12 | 3 | — | |
| FigStep-proSystem=Gemini-2.0-pro2025.12 | 3 | — | |
| IAModel=Gemma22025.05 | 4 | 1 | |
| Self-ReminderModel=Qwen2.52025.05 | 4 | 1.1 | |
| Self-ReminderModel=Llama3.12025.05 | 4 | 1 | |
| Goal PrioritizationModel=Llama3.12025.05 | 4 | 1 | |
| MM-SafetyBenchSystem=Grok-32025.12 | 4 | — | |
| HADESSystem=Gemini-2.0-pro2025.12 | 4 | — | |
| ArtPromptSystem=Gemini-2.0-pro2025.12 | 4 | — | |
| Dream-v0-Instruct-7B + Self-reminder + DIFFUGUARDBase Model=Dream-v0-Instruct-7B, Defense Method=Self-reminder + DIFFUGUARD2025.09 | 4 | — | |
| MM-SafetyBenchSystem=Gemini-2.0-pro2025.12 | 5 | — | |
| MM-SafetyBenchSystem=Gemini-2.0-flash2025.12 | 5 | — | |
| Dream-v0-Instruct-7B + DIFFUGUARDBase Model=Dream-v0-Instruct-7B, Defense Method=DIFFUGUARD2025.09 | 5 | — | |
| UMKSystem=GPT-4o2025.12 | 6 | — | |
| Dream-v0-Instruct-7B + Self-reminderBase Model=Dream-v0-Instruct-7B, Defense Method=Self-reminder2025.09 | 6 | — | |
| Dream-v0-Instruct-7B + PPL-FilterBase Model=Dream-v0-Instruct-7B, Defense Method=PPL-Filter2025.09 | 6.67 | — | |
| VAESystem=GPT-4o2025.12 | 7 | — | |
| BAPSystem=Gemini-2.0-pro2025.12 | 7 | — | |
| Dream-v0-Instruct-7BBase Model=Dream-v0-Instruct-7B, Defense Method=None2025.09 | 7.33 | — | |
| Self-ExaminationModel=Llama3.12025.05 | 8 | 1.08 | |
| MM-SafetyBenchSystem=Gemini-2.0-pro2025.12 | 8 | — | |
| No DefenseModel=Qwen2.52025.05 | 9 | 1.29 | |
| HADESSystem=GPT-4o2025.12 | 9 | — | |
| ArtPromptSystem=Gemini-2.0-flash2025.12 | 9 | — | |
| UMKSystem=GPT-4o2025.12 | 9 | — | |
| ArtPromptSystem=Gemini-2.0-flash2025.12 | 9 | — | |
| VAESystem=GPT-4o2025.12 | 11 | — | |
| No DefenseModel=Gemma22025.05 | 12 | 1.64 | |
| ArtPromptSystem=GPT-4o2025.12 | 12 | — | |
| BAPSystem=Gemini-2.0-flash2025.12 | 12 | — | |
| HADESSystem=GPT-4o2025.12 | 12 | — | |
| MMaDA-8B-MixCoT + Self-reminder + DIFFUGUARDBase Model=MMaDA-8B-MixCoT, Defense Method=Self-reminder + DIFFUGUARD2025.09 | 12.67 | — | |
| MM-SafetyBenchSystem=GPT-4o2025.12 | 13 | — | |
| ArtPromptSystem=GPT-4o2025.12 | 15 | — | |
| MMaDA-8B-MixCoT + Self-reminderBase Model=MMaDA-8B-MixCoT, Defense Method=Self-reminder2025.09 | 16 | — | |
| LLaDA-8B-Instruct + Self-reminder + DIFFUGUARDBase Model=LLaDA-8B-Instruct, Defense Method=Self-reminder + DIFFUGUARD2025.09 | 16.33 | — | |
| HADESSystem=Gemini-2.0-flash2025.12 | 17 | — | |
| LLaDA-1.5 + Self-reminder + DIFFUGUARDBase Model=LLaDA-1.5, Defense Method=Self-reminder + DIFFUGUARD2025.09 | 17 | — | |
| LLaDA-8B-Instruct + Self-reminderBase Model=LLaDA-8B-Instruct, Defense Method=Self-reminder2025.09 | 17.33 | — | |
| LLaDA-1.5 + Self-reminderBase Model=LLaDA-1.5, Defense Method=Self-reminder2025.09 | 18 | — | |
| HADESSystem=Gemini-2.0-flash2025.12 | 19 | — | |
| MM-SafetyBenchSystem=Grok-32025.12 | 21 | — | |
| No DefenseModel=Llama3.12025.05 | 22 | 1.8 | |
| BAPSystem=GPT-4o2025.12 | 22 | — | |
| LLaDA-1.5 + PPL-FilterBase Model=LLaDA-1.5, Defense Method=PPL-Filter2025.09 | 22.33 | — | |
| LLaDA-8B-Instruct + DIFFUGUARDBase Model=LLaDA-8B-Instruct, Defense Method=DIFFUGUARD2025.09 | 22.67 | — | |
| LLaDA-1.5 + DIFFUGUARDBase Model=LLaDA-1.5, Defense Method=DIFFUGUARD2025.09 | 22.67 | — | |
| VAESystem=Grok-32025.12 | 24 | — | |
| BAPSystem=Gemini-2.0-pro2025.12 | 24 | — | |
| BAPSystem=GPT-4o2025.12 | 25 | — | |
| LLaDA-8B-Instruct + PPL-FilterBase Model=LLaDA-8B-Instruct, Defense Method=PPL-Filter2025.09 | 25.67 | — | |
| LLaDA-1.5Base Model=LLaDA-1.5, Defense Method=None2025.09 | 25.67 | — | |
| MM-SafetyBenchSystem=Gemini-2.0-flash2025.12 | 26 | — | |
| VAESystem=Grok-32025.12 | 26 | — | |
| LLaDA-8B-InstructBase Model=LLaDA-8B-Instruct, Defense Method=None2025.09 | 27.33 | — | |
| HIMRDSystem=Gemini-2.0-pro2025.12 | 31 | — | |
| ArtPromptSystem=Grok-32025.12 | 32 | — | |
| UMKSystem=Grok-32025.12 | 33 | — | |
| HIMRDSystem=Gemini-2.0-pro2025.12 | 34 | — | |
| HADESSystem=Grok-32025.12 | 34 | — | |
| FigStep-proSystem=GPT-4o2025.12 | 35 | — | |
| FigStep-proSystem=Gemini-2.0-flash2025.12 | 36 | — | |
| BAPSystem=Grok-32025.12 | 36 | — | |
| ArtPromptSystem=Grok-32025.12 | 38 | — | |
| HADESSystem=Grok-32025.12 | 40 | — | |
| UMKSystem=Grok-32025.12 | 40 | — | |
| MMaDA-8B-MixCoT + DIFFUGUARDBase Model=MMaDA-8B-MixCoT, Defense Method=DIFFUGUARD2025.09 | 41.67 | — | |
| HIMRDSystem=Gemini-2.0-flash2025.12 | 47 | — | |
| HIMRDSystem=Gemini-2.0-flash2025.12 | 47 | — | |
| BAPSystem=Gemini-2.0-flash2025.12 | 47 | — |