Prompt injection detection
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
0FPR
7
Feb 26, 2026
0FPR
7
Feb 26, 2026
0FPR (Coding)
7
Feb 26, 2026
85.2F1 Score
6
Jun 4, 2026
0FPR (Embed Screenshot)
6
Apr 29, 2026
0Naive FPR
6
Feb 26, 2026
0FPR
6
Feb 26, 2026
0.01FPR
6
Feb 26, 2026
0FNR (X)
5
Jun 15, 2026
95Accuracy
5
Jun 12, 2026
92.4Accuracy
5
Jun 1, 2026
72.96Macro F1
5
May 27, 2026
95.41Macro F1 Score
5
May 27, 2026
6Naïve RL Score
5
Feb 26, 2026
97Banking RL
5
Feb 26, 2026
0.089Delta
4
Jun 23, 2026
82.65Accuracy
4
Jun 4, 2026
53.2FNR
3
Jun 23, 2026
58.3Precision
3
May 29, 2026
80.82Macro-F1
3
May 27, 2026
75.28Macro-F1
3
May 27, 2026
98.8Detection Accuracy (Top-1)
3
Feb 26, 2026
0Misses Count
2
Jun 23, 2026
1FNR
2
Jun 23, 2026
69.3False Negative Rate (FNR)
2
Jun 23, 2026