Red-teaming on Web-Augmented LLM Red-Teaming Evaluation Set
80.5Detection RateCREST-Search
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CREST-Search2025.10 | 80.5 | 23.6 | 0.59 | |
| HarmBench2025.10 | 11.6 | 53.8 | 0.48 | |
| DangerousQA2025.10 | 10.5 | 65.5 | 0.61 | |
| RED-EVAL2025.10 | 5.7 | 49.9 | 0.76 | |
| UAT2025.10 | 5.2 | 87.9 | 0.77 |