Jailbreak Detection
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
92.1Recall
7
Jun 23, 2026
7.3FPR (PMP)
7
Jun 8, 2026
1.4FPR (PMP)
7
Jun 8, 2026
90.04Accuracy
7
Feb 26, 2026
92ASR
5
Jun 8, 2026
81ASR
5
Jun 8, 2026
85Transferred Detection Rate
5
May 12, 2026
81Transferred Detection Rate
5
May 12, 2026
95Accuracy
5
Feb 26, 2026
0.96Accuracy
5
Feb 26, 2026
96.44Reference Score
4
Jun 4, 2026
97.5Refusal Rate
4
Jun 4, 2026
100Detection Accuracy
4
May 6, 2026
92Detection Rate
4
Apr 28, 2026
100Detection Rate
4
Apr 28, 2026
99Detection Rate
4
Apr 28, 2026
100Detection Rate
4
Apr 28, 2026
97Accuracy
4
Feb 26, 2026
1.871High-Risk Vulnerability Discovery Rate
4
Feb 26, 2026
98.1Jailbreak Success Rate
3
May 14, 2026
100Transferred Detection Rate
3
May 12, 2026
89Accuracy
3
Feb 26, 2026
98ASR (Recall)
3
Feb 26, 2026
100Accuracy
3
Feb 26, 2026
99GCG Detection Rate (GCG DR)
2
May 26, 2026