| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Human Red-teaming IH-Challenge | Number of Tasks271 | 3 | 1mo ago | ||
| Developer <> User Conflict | GPT-5-Mini-R | Non-violation Rate95 | 2 | 1mo ago | |
| System <> Developer Conflict | Non-violation Rate86 | 2 | 1mo ago | ||
| System <> User Conflict | GPT-5-Mini-R | Non-violation Rate95 | 2 | 1mo ago | |
| Tutor Jailbreak user (dev) | GPT-5-Mini-R | Non-violation Rate99 | 2 | 1mo ago | |
| Tutor Jailbreak sys-user | GPT-5-Mini-R | Non-violation Rate99 | 2 | 1mo ago | |
| RealGuardrails Handwritten | GPT-5-Mini-R | Score89 | 2 | 1mo ago | |
| RealGuardrails Distractors | GPT-5-Mini-R | Score0.95 | 2 | 1mo ago | |
| TensorTrust user (dev) | GPT-5-Mini-R | Score91 | 2 | 1mo ago | |
| TensorTrust sys-user | GPT-5-Mini-R | Score94 | 2 | 1mo ago | |
| Gandalf Password (dev-user) | GPT-5-Mini-R | Score100 | 2 | 1mo ago | |
| Gandalf Password sys-user | Score99 | 2 | 1mo ago |