Policy Violation Detection on SafeAgentBench
234True Positives
VIGIL
Updated 29d ago
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
2026.06 | 234 | 4 | 98.3 | 94.4 | 96.3 | |
2026.06 | 231 | 2 | 99.1 | 93.1 | 96 | |
2026.06 | 222 | 41 | 84.4 | 89.5 | 86.9 |
VIGIL
| Method | Links | |||||
|---|---|---|---|---|---|---|
2026.06 | 234 | 4 | 98.3 | 94.4 | 96.3 | |
2026.06 | 231 | 2 | 99.1 | 93.1 | 96 | |
2026.06 | 222 | 41 | 84.4 | 89.5 | 86.9 |