Jailbreak Attack on AdvCUA
0.5ASSTRACE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TRACETarget Agent=DeepSeek Agent2026.05 | 0.5 | 100 | |
| TRACETarget Agent=Gemini Agent2026.05 | 0.46 | 100 | |
| TRACETarget Agent=GPT Agent2026.05 | 0.31 | 100 | |
| Red-Agent-ReflectTarget Agent=DeepSeek Agent2026.05 | 0.27 | 62 | |
| AutoDAN-TurboTarget Agent=DeepSeek Agent2026.05 | 0.23 | 54 | |
| BaseTarget Agent=DeepSeek Agent2026.05 | 0.19 | 54 | |
| ReNeLLMTarget Agent=DeepSeek Agent2026.05 | 0.19 | 58 | |
| X-TeamingTarget Agent=DeepSeek Agent2026.05 | 0.19 | 73 | |
| Red-Agent-ReflectTarget Agent=GPT Agent2026.05 | 0.04 | 15 | |
| Red-Agent-ReflectTarget Agent=Gemini Agent2026.05 | 0.04 | 8 | |
| BaseTarget Agent=GPT Agent2026.05 | 0 | 0 | |
| BaseTarget Agent=Gemini Agent2026.05 | 0 | 0 | |
| ReNeLLMTarget Agent=GPT Agent2026.05 | 0 | 4 | |
| ReNeLLMTarget Agent=Gemini Agent2026.05 | 0 | 42 | |
| AutoDAN-TurboTarget Agent=GPT Agent2026.05 | 0 | 4 | |
| AutoDAN-TurboTarget Agent=Gemini Agent2026.05 | 0 | 15 | |
| X-TeamingTarget Agent=GPT Agent2026.05 | 0 | 12 | |
| X-TeamingTarget Agent=Gemini Agent2026.05 | 0 | 27 |