| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| JBB-Behaviors | RA | ASR (PAIR)10 | 18 | 3mo ago | |
| GCG | RealSafe-R1 | Harmfulness Rate0 | 17 | 27d ago | |
| FigStep jailbreak attack | TGA-7B | DSR17.44 | 7 | 3mo ago | |
| ICA | TGA-7B | DSR1,543 | 7 | 3mo ago | |
| Role-Play jailbreak attack | TGA-7B | DSR2,108 | 7 | 3mo ago | |
| Jailbreak Attack Evaluation Set Llama-3 8B | GSAE | GCG Robustness Score100 | 6 | 3mo ago | |
| LogiBreak | SFT | LogiBreak ASR0 | 4 | 1mo ago |