AI-generated text detection
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
72.62Precision
7
Feb 26, 2026
70.58Precision
7
Feb 26, 2026
100Accuracy (1% FPR)
7
Feb 27, 2026
100Accuracy (1% FPR)
7
Feb 27, 2026
0.887AUROC (News)
6
Jul 1, 2026
27.27Throughput (sample/s)
6
Jun 2, 2026
89.44Macro-F1 (GPT-4.1)
6
May 22, 2026
82.71AUROC (Homoglyph)
6
Apr 21, 2026
49.5Accuracy (at 1% FPR)
6
Feb 27, 2026
0.962Avg Recall
6
Feb 26, 2026
99.74Avg Recall
6
Feb 26, 2026
93.42Avg Recall
6
Feb 26, 2026
98.44Avg Recall
6
Feb 26, 2026
89.63Avg Recall
6
Feb 26, 2026
0.93Avg Recall
6
Feb 26, 2026
65.9Detection Accuracy (1% FPR)
6
Feb 27, 2026
1Accuracy (1% FPR)
6
Feb 27, 2026
0.963AUROC
5
Jul 7, 2026
89.2TPR
5
Apr 21, 2026
94.23TPR
5
Jun 2, 2026
93.37TPR
5
Jun 2, 2026
99BERT Score
5
Apr 10, 2026
91.5Accuracy
5
Feb 26, 2026
97Avg Recall
5
Feb 26, 2026
97.7Avg Recall
5
Feb 26, 2026