Policy Corruption Evaluation on GPT-4o mini
3.53Compliance ScoreHPM
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| HPMVictim Model=GPT-4o-mini2025.12 | 3.53 | 3.21 | 2.95 | 3.6 | 2.81 | 2.9 | 3.04 | |
| PAPVictim Model=GPT-4o-mini2025.12 | 2.25 | 2.01 | 1.85 | 2.22 | 1.25 | 1.55 | 1.43 | |
| CoAVictim Model=GPT-4o-mini2025.12 | 1.81 | 1.53 | 1.3 | 1.92 | 0.88 | 0.81 | 1.05 | |
| PAIRVictim Model=GPT-4o-mini2025.12 | 1.22 | 1.01 | 0.85 | 1.03 | 0.61 | 0.42 | 0.55 | |
| AutoDANVictim Model=GPT-4o-mini2025.12 | 1.03 | 0.84 | 0.62 | 0.81 | 0.45 | 0.22 | 0.31 |