Jailbreak Attack Success Evaluation on IICL Evaluation Set 5 harmful queries
0Average Bypass RateIICL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| IICLTarget Model=gpt-52026.04 | 0 | 0 | 0 | |
| IICLTarget Model=gpt-5-mini2026.04 | 0 | 0 | 0 | |
| IICLTarget Model=gpt-5-pro2026.04 | 0 | 0 | 0 | |
| IICLTarget Model=gpt-5.22026.04 | 0 | 0 | 0 | |
| IICLTarget Model=gpt-5.2-pro2026.04 | 0 | 0 | 0 | |
| IICLTarget Model=gpt-5.4-pro2026.04 | 0 | 0 | 0 | |
| IICLTarget Model=gpt-5.42026.04 | 1.7 | 60 | 60 | |
| IICLTarget Model=gpt-5.12026.04 | 4.2 | 60 | 60 | |
| IICLTarget Model=gpt-4o2026.04 | 14.3 | 100 | 100 | |
| IICLTarget Model=gpt-4.12026.04 | 14.9 | 100 | 100 |