Paper Overview
Field: NLP Authors: Ben Glickenhaus, Katherine Thai, Jenna Russell, Elyas Masrour, Yue Han, Max Spero, Bradley Emi Published: 2026-07-29 arXiv: 2607.27183
Summary
This paper presents Pangram 4, the latest deep-learning-based AI-text classification model from Pangram Labs. The authors achieve an AUROC of 0.9916 with a false positive rate of 0.0041% and a false negative rate of 0.3396%.
Key Contributions
- Improved accuracy: Higher overall detection accuracy compared with Pangram 3.
- Robustness: Superior out-of-distribution generalization and stronger resistance to adversarial attacks.
- Fine-grained and mixed-text detection: A novel capability to distinguish fine-grained edits and mixed AI-human co-authored text, with improvements on both boundary detection tasks and detection of interleaved AI assistance.
- State-of-the-art results: Metrics on standard AI detection benchmarks show Pangram 4 achieves state-of-the-art performance on the AI text detection task across a wide variety of settings and domains.
Original Abstract (excerpt)
> We present Pangram 4, the latest deep-learning-based AI-text classification model from Pangram Labs. We achieve an AUROC of 0.9916 with a false positive rate of 0.0041% and a false negative rate of 0.3396%. In addition to its increased overall accuracy compared with Pangram 3, Pangram 4 exhibits superior out-of-distribution generalization and adversarial attack robustness. Another novel contribution of Pangram 4 is its improved ability to distinguish fine-grained edits and mixed AI-human co-authored text. We demonstrate improvements to both boundary detection tasks and the detection of interleaved AI assistance. Finally, we report metrics on standard AI detection benchmarks showing that Pangram 4 achieves state-of-the-art performance on the AI text detection task across a wide variety of settings and domains.
Full paper: arXiv:2607.27183
--- *Auto-collected on 2026-07-31*