| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Alpaca-Farm | Robustness via Referencing | ASR0 | 105 | 1mo ago | |
| NQ simplified | Naïve Success Rate41 | 24 | 5mo ago | ||
| AgentDojo (test) | Banking Success Rate36 | 7 | 5mo ago | ||
| ASB 5 direct prompt-injection styles | SecureClaw | ASR0 | 2 | 1mo ago |