Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Feature-Augmented Transformers for Robust AI-Text Detection Across Domains and Generators

About

AI-generated text is nowadays produced at scale across domains and heterogeneous generation pipelines, making robustness to distribution shift a central requirement for supervised binary detectors. We train transformer-based detectors on HC3 PLUS and calibrate a single decision threshold by maximising balanced accuracy on held-out validation; this threshold is then kept fixed for all downstream test distributions, revealing domain- and generator-dependent error asymmetries under shift. We evaluate in-domain on HC3 PLUS, under cross-dataset transfer to the multi-domain, multi-generator M4 benchmark, and on the external AI-Text-Detection-Pile. Although base models achieve near-ceiling in-domain performance (up to 99.5% balanced accuracy), performance under shift is brittle and strongly model-dependent. Feature augmentation via attention-based linguistic feature fusion improves transfer, with our best model (DeBERTa-v3-base+FeatAttn) achieving 85.9% balanced accuracy on M4. Multi-seed experiments confirm high stability. Under the same fixed-threshold protocol, our model outperforms strong zero-shot baselines by up to +7.22 points. Category-level ablations further show that readability and vocabulary features contribute most to robustness under shift. Overall, these results demonstrate that feature augmentation and a modern DeBERTa backbone significantly outperform earlier BERT/RoBERTa models, while the fixed-threshold protocol provides a more realistic and informative assessment of practical detector robustness.

Mohamed Mady, Johannes Reschke, Bj\"orn Schuller• 2026

Related benchmarks

TaskDatasetResultRank
AI Text DetectionM4 (macro-average)
Balanced Accuracy (BA)84.5
6
AI Text DetectionM4 (macro)
Balanced Accuracy84.5
4
AI Text DetectionHC3 PLUS QA (test)
Balanced Accuracy99.47
2
AI Text DetectionHC3 PLUS (test_si)
Balanced Accuracy (BA)87.21
2
AI Text DetectionM4 wikipedia
Balanced Accuracy89.96
2
AI Text DetectionM4 wikihow
Balanced Accuracy (BA)81.31
2
AI Text DetectionM4 reddit
Balanced Accuracy90.23
2
AI Text DetectionM4 arxiv
BA75.55
2
AI Text DetectionM4 peerread
Balanced Accuracy93
2
AI Text DetectionHC3 QA
Balanced Accuracy99.63
1
Showing 10 of 12 rows

Other info

Follow for update