Adversarial Attack Detection on Agentic AI workflow benchmark internal
100Precision
Qwen3Guard-8B-loose
Updated 4d ago
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
2025.12 | 100 | 1 | 1 | 0 | |
2025.12 | 100 | 91 | 95 | 1 | |
2025.12 | 99 | 84 | 91 | 2 | |
2025.12 | 98 | 13 | 22 | 1 | |
2025.12 | 98 | 19 | 32 | 1 | |
2025.12 | 91 | 20 | 33 | 27 | |
2025.12 | 91 | 21 | 34 | 6 | |
2025.12 | 88 | 17 | 28 | 6 | |
2025.12 | 86 | 18 | 30 | 8 | |
2025.12 | 83 | 11 | 19 | 6 | |
2025.12 | 78 | 81 | 80 | 62 | |
2025.12 | 74 | 75 | 74 | 71 | |
2025.12 | 72 | 1 | 3 | 2 | |
2025.12 | 72 | 82 | 77 | 90 |