T-Detect: Tail-Aware Statistical Normalization for Robust Detection of Adversarial Machine-Generated Text

TL;DR

T-Detect introduces a heavy-tailed statistical normalization technique using Student's t-distribution to improve the robustness of adversarial text detection, outperforming existing methods on benchmark datasets.

Contribution

The paper presents a novel heavy-tailed normalization approach for adversarial text detection, replacing Gaussian assumptions with Student's t-distribution, backed by theoretical and empirical validation.

Findings

Improves AUROC by up to 3.9% on RAID benchmark.

Achieves state-of-the-art AUROC of 0.926 on RAID Books domain.

Demonstrates robustness against adversarial perturbations.

Abstract

Large language models (LLMs) have shown the capability to generate fluent and logical content, presenting significant challenges to machine-generated text detection, particularly text polished by adversarial perturbations such as paraphrasing. Current zero-shot detectors often employ Gaussian distributions as statistical measure for computing detection thresholds, which falters when confronted with the heavy-tailed statistical artifacts characteristic of adversarial or non-native English texts. In this paper, we introduce T-Detect, a novel detection method that fundamentally redesigns the curvature-based detectors. Our primary innovation is the replacement of standard Gaussian normalization with a heavy-tailed discrepancy score derived from the Student's t-distribution. This approach is theoretically grounded in the empirical observation that adversarial texts exhibit significant leptokurtosis, rendering traditional statistical assumptions inadequate. T-Detect computes a detection score by normalizing the log-likelihood of a passage against the expected moments of a t-distribution, providing superior resilience to statistical outliers. We validate our approach on the challenging RAID benchmark for adversarial text and the comprehensive HART dataset. Experiments show that T-Detect provides a consistent performance uplift over strong baselines, improving AUROC by up to 3.9\% in targeted domains. When integrated into a two-dimensional detection framework (CT), our method achieves state-of-the-art performance, with an AUROC of 0.926 on the Books domain of RAID. Our contributions are a new, theoretically-justified statistical foundation for text detection, an ablation-validated method that demonstrates superior robustness, and a comprehensive analysis of its performance under adversarial conditions. Ours code are released at https://github.com/ResearAI/t-detect.