List your company

Jailbreak Attack on MI (MaliciousInstructions)

1QSRPuzzler
-0.040.230.50.77Feb 14, 2024

Evaluation results

MethodLinks
PuzzlerTested Model=GPT3.52024.0210.9
PuzzlerTested Model=GPT42024.0210.86
PuzzlerTested Model=GPT4-Turbo2024.0210.87
CRTested Model=GPT3.52024.020.90.936
PuzzlerTested Model=Gemini-pro2024.020.830.867
TCTested Model=GPT3.52024.020.530.867
CRTested Model=Gemini-pro2024.020.470.893
SIMUTested Model=GPT3.52024.020.40.909
TCTested Model=GPT42024.020.40.85
TAPTested Model=GPT3.52024.020.370.81
PuzzlerTested Model=LLama2-13B-chat2024.020.291
TAPTested Model=GPT42024.020.260.769
SIMUTested Model=Gemini-pro2024.020.170.882
TAPTested Model=Gemini-pro2024.020.140.786
CRTested Model=GPT42024.020.130.846
TAPTested Model=GPT4-Turbo2024.020.130.846
TCTested Model=GPT4-Turbo2024.020.070.857
PuzzlerTested Model=LLama2-7B-chat2024.020.030.667
TAPTested Model=LLama2-13B-chat2024.020.021
SIMUTested Model=GPT42024.0200
CRTested Model=GPT4-Turbo2024.0200
SIMUTested Model=GPT4-Turbo2024.0200
TCTested Model=Gemini-pro2024.0200
TAPTested Model=LLama2-7B-chat2024.0200
CRTested Model=LLama2-7B-chat2024.0200
TCTested Model=LLama2-7B-chat2024.0200
SIMUTested Model=LLama2-7B-chat2024.0200
CRTested Model=LLama2-13B-chat2024.0200
TCTested Model=LLama2-13B-chat2024.0200
SIMUTested Model=LLama2-13B-chat2024.0200