Persona Manipulation on MPI (test)
83.54Success ScorePHISH
Evaluation Results
| Method | Links | |
|---|---|---|
| PHISHModel=DeepSeek-V3, Category=Open2026.01 | 83.54 | |
| DANModel=DeepSeek-V3, Category=Open2026.01 | 81.25 | |
| PHISHModel=GPT-4o, Category=Proprietary2026.01 | 79.38 | |
| FlipAttackModel=GPT-4o, Category=Proprietary2026.01 | 78.54 | |
| PHISHModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 75.42 | |
| DrAttackModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 75 | |
| DANModel=Llama4-Maverick, Category=Open2026.01 | 75 | |
| PHISHModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 74.58 | |
| DrAttackModel=o3-mini, Category=Proprietary2026.01 | 72.51 | |
| FlipAttackModel=o3-mini, Category=Proprietary2026.01 | 70.63 | |
| CipherCHATModel=o3-mini, Category=Proprietary2026.01 | 70.42 | |
| DrAttackModel=GPT-4o, Category=Proprietary2026.01 | 70.21 | |
| FlipAttackModel=Llama4-Maverick, Category=Open2026.01 | 68.75 | |
| PHISHModel=Llama4-Maverick, Category=Open2026.01 | 68.13 | |
| PHISHModel=o3-mini, Category=Proprietary2026.01 | 67.08 | |
| PHISHModel=MedGemma-27B (Medical), Category=Open2026.01 | 66.67 | |
| DrAttackModel=DeepSeek-V3, Category=Open2026.01 | 66.46 | |
| DrAttackModel=Llama4-Maverick, Category=Open2026.01 | 66.04 | |
| DeepIncModel=o3-mini, Category=Proprietary2026.01 | 65.42 | |
| DANModel=MedGemma-27B (Medical), Category=Open2026.01 | 65.33 | |
| DANModel=o3-mini, Category=Proprietary2026.01 | 59.38 | |
| DANModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 57.29 | |
| FlipAttackModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 56.46 | |
| DeepIncModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 56.25 | |
| DeepIncModel=DeepSeek-V3, Category=Open2026.01 | 56.25 | |
| DrAttackModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 55.83 | |
| DeepIncModel=Llama4-Maverick, Category=Open2026.01 | 53.54 | |
| DeepIncModel=GPT-4o, Category=Proprietary2026.01 | 52.08 | |
| DrAttackModel=MedGemma-27B (Medical), Category=Open2026.01 | 50.83 | |
| FlipAttackModel=DeepSeek-V3, Category=Open2026.01 | 49.38 | |
| CipherCHATModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 45.83 | |
| FlipAttackModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 43.96 | |
| CipherCHATModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 42.92 | |
| FlipAttackModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 42.71 | |
| CipherCHATModel=DeepSeek-V3, Category=Open2026.01 | 42.71 | |
| DeepIncModel=MedGemma-27B (Medical), Category=Open2026.01 | 41.67 | |
| DeepIncModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 38.96 | |
| SLIPModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 36.04 | |
| UASModel=MedGemma-27B (Medical), Category=Open2026.01 | 35.83 | |
| DANModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 34.58 | |
| UASModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 30.42 | |
| PHISHModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 26.46 | |
| DANModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 26.04 | |
| UASModel=Llama4-Maverick, Category=Open2026.01 | 25 | |
| CipherCHATModel=Llama4-Maverick, Category=Open2026.01 | 24.79 | |
| DrAttackModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 24.38 | |
| DeepInceptionModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 23.75 | |
| CipherCHATModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 17.08 | |
| SLIPModel=Llama4-Maverick, Category=Open2026.01 | 13.75 | |
| UASModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 12.08 | |
| SLIPModel=MedGemma-27B (Medical), Category=Open2026.01 | 10.83 | |
| FlipAttackModel=MedGemma-27B (Medical), Category=Open2026.01 | 10.83 | |
| SLIPModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 7.92 | |
| RANDOMModel=Llama4-Maverick, Category=Open2026.01 | 7.08 | |
| UASModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 6.67 | |
| RANDOMModel=Claude-3.5-Haiku, Category=Proprietary2026.01 | 5.83 | |
| CipherCHATModel=GPT-4o, Category=Proprietary2026.01 | 5 | |
| RANDOMModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 4.17 | |
| RANDOMModel=ChatHaruhi (Role-playing), Category=Open2026.01 | 3.13 | |
| SLIPModel=o3-mini, Category=Proprietary2026.01 | 2.92 | |
| CipherCHATModel=MedGemma-27B (Medical), Category=Open2026.01 | 2.71 | |
| DANModel=GPT-4o, Category=Proprietary2026.01 | 2.29 | |
| SLIPModel=DeepSeek-V3, Category=Open2026.01 | 2.29 | |
| UASModel=DeepSeek-V3, Category=Open2026.01 | 1.67 | |
| RANDOMModel=o3-mini, Category=Proprietary2026.01 | 1.25 | |
| SLIPModel=GPT-4o, Category=Proprietary2026.01 | 0.83 | |
| UASModel=GPT-4o, Category=Proprietary2026.01 | 0.63 | |
| RANDOMModel=GPT-4o, Category=Proprietary2026.01 | 0.42 | |
| SLIPModel=Gemini-2.0-Flash, Category=Proprietary2026.01 | 0.42 | |
| RANDOMModel=MedGemma-27B (Medical), Category=Open2026.01 | 0.42 | |
| UASModel=o3-mini, Category=Proprietary2026.01 | 0 | |
| RANDOMModel=DeepSeek-V3, Category=Open2026.01 | 0 |