AI-Generated Text is Non-Stationary: Detection via Temporal Tomography

TL;DR

This paper reveals that AI-generated text is non-stationary and introduces Temporal Discrepancy Tomography (TDT), a novel detection method that preserves positional information, significantly improving detection accuracy and robustness against adversarial attacks.

Contribution

The paper introduces TDT, a new detection paradigm that models AI-generated text as a time-series signal, capturing non-stationarity and positional anomalies for improved detection.

Findings

TDT achieves 0.855 AUROC on RAID benchmark, outperforming baselines.

TDT improves robustness with 14.1% AUROC gain against adversarial paraphrasing.

TDT maintains practical efficiency with only 13% additional computational overhead.

Abstract

The field of AI-generated text detection has evolved from supervised classification to zero-shot statistical analysis. However, current approaches share a fundamental limitation: they aggregate token-level measurements into scalar scores, discarding positional information about where anomalies occur. Our empirical analysis reveals that AI-generated text exhibits significant non-stationarity, statistical properties vary by 73.8\% more between text segments compared to human writing. This discovery explains why existing detectors fail against localized adversarial perturbations that exploit this overlooked characteristic. We introduce Temporal Discrepancy Tomography (TDT), a novel detection paradigm that preserves positional information by reformulating detection as a signal processing task. TDT treats token-level discrepancies as a time-series signal and applies Continuous Wavelet Transform to generate a two-dimensional time-scale representation, capturing both the location and linguistic scale of statistical anomalies. On the RAID benchmark, TDT achieves 0.855 AUROC (7.1\% improvement over the best baseline). More importantly, TDT demonstrates robust performance on adversarial tasks, with 14.1\% AUROC improvement on HART Level 2 paraphrasing attacks. Despite its sophisticated analysis, TDT maintains practical efficiency with only 13\% computational overhead. Our work establishes non-stationarity as a fundamental characteristic of AI-generated text and demonstrates that preserving temporal dynamics is essential for robust detection.