Jailbreak Defense Evaluation on EMRA RQ
2.3ASRCoopGuard
Evaluation Results
| Method | Links | |
|---|---|---|
| CoopGuardTarget LLM=GPT-52026.04 | 2.3 | |
| CoopGuardTarget LLM=Gemini-2.5-Pro2026.04 | 6.3 | |
| GoalPriorityTarget LLM=GPT-52026.04 | 7.2 | |
| GoalPriorityTarget LLM=Gemini-2.5-Pro2026.04 | 7.5 | |
| Self-ReminderTarget LLM=DeepSeek-V32026.04 | 7.9 | |
| Self-ReminderTarget LLM=Gemini-2.5-Pro2026.04 | 8.9 | |
| Self-ReminderTarget LLM=GPT-52026.04 | 11.4 | |
| CoopGuardTarget LLM=DeepSeek-V32026.04 | 13.8 | |
| PATTarget LLM=DeepSeek-V32026.04 | 14.8 | |
| GoalPriorityTarget LLM=DeepSeek-V32026.04 | 15.3 | |
| PATTarget LLM=Gemini-2.5-Pro2026.04 | 16.6 | |
| RPOTarget LLM=Gemini-2.5-Pro2026.04 | 16.8 | |
| RPOTarget LLM=DeepSeek-V32026.04 | 17 | |
| SecurityLinguaTarget LLM=GPT-52026.04 | 17.4 | |
| PATTarget LLM=GPT-52026.04 | 18.7 | |
| RPOTarget LLM=GPT-52026.04 | 20.7 | |
| SecurityLinguaTarget LLM=DeepSeek-V32026.04 | 21.5 | |
| SecurityLinguaTarget LLM=Gemini-2.5-Pro2026.04 | 24.5 |