| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Standard non-adversarial dataset e-commerce | GAVEL | TPR90 | 3 | 4mo ago | |
| Standard non-adversarial dataset Romance | GAVEL | TPR100 | 3 | 4mo ago | |
| Standard non-adversarial dataset Tax Authority | TPR100 | 3 | 4mo ago | ||
| Standard non-adversarial dataset Racism | GAVEL | TPR99 | 3 | 4mo ago | |
| Standard non-adversarial dataset Elections | GAVEL | TPR99 | 3 | 4mo ago | |
| Standard non-adversarial dataset Anti-LGBTQ | GAVEL | TPR100 | 3 | 4mo ago | |
| Standard non-adversarial dataset Delusional | GAVEL | TPR90 | 3 | 4mo ago | |
| Standard non-adversarial dataset SQL Injection | TPR100 | 3 | 4mo ago | ||
| Standard non-adversarial dataset Phishing | TPR98 | 3 | 4mo ago | ||
| ToxiGen | gpt-oss-safeguard tailored policy | Detection Rate72.3 | 1 | 1mo ago | |
| BBQ | gpt-oss-safeguard tailored policy | Detection Rate76.8 | 1 | 1mo ago | |
| HVD | gpt-oss-safeguard tailored policy | Detection Rate80.5 | 1 | 1mo ago |