Malicious Prompt Detection
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
0.92AUROC (VLGuard & MSSBench)
26
Jul 7, 2026
94.37AUROC
26
Jul 7, 2026
98.06AUROC
26
Jul 7, 2026
0FNR
16
Apr 9, 2026
0.19FNR
16
Apr 9, 2026
0.16FPR
14
Apr 9, 2026
3.67FPR
14
Apr 9, 2026
0FPR
14
Apr 9, 2026
0FPR
14
Apr 9, 2026
97.1Language Accuracy
12
Jun 23, 2026
4.5ASR
6
Feb 26, 2026
97.48AUROC
5
Jul 7, 2026
96.82AUROC
5
Jul 7, 2026
99.05AUROC
5
Jul 7, 2026
98.71Accuracy
4
Mar 10, 2026
98.72Accuracy
4
Mar 10, 2026
87.89Accuracy (%)
4
Mar 10, 2026
98.15Accuracy
4
Mar 10, 2026
92.1Accuracy
4
Mar 10, 2026
99.73Accuracy
4
Mar 10, 2026
30Fires Rate
1
Jun 5, 2026
91Detection Accuracy
1
Feb 26, 2026
100Detection Accuracy
1
Feb 26, 2026