RepreGuard: Detecting LLM-Generated Text by Revealing Hidden Representation Patterns

TL;DR

RepreGuard is a novel detection method that leverages internal neural activation patterns of LLMs to distinguish between machine-generated and human-written texts, showing high accuracy and robustness across various scenarios.

Contribution

This paper introduces RepreGuard, a new statistics-based detection approach utilizing internal LLM representations, improving robustness over existing methods in diverse conditions.

Findings

Achieves 94.92% AUROC on both ID and OOD data

Demonstrates robustness against various text sizes and attacks

Outperforms all baseline detection methods

Abstract

Detecting content generated by large language models (LLMs) is crucial for preventing misuse and building trustworthy AI systems. Although existing detection methods perform well, their robustness in out-of-distribution (OOD) scenarios is still lacking. In this paper, we hypothesize that, compared to features used by existing detection methods, the internal representations of LLMs contain more comprehensive and raw features that can more effectively capture and distinguish the statistical pattern differences between LLM-generated texts (LGT) and human-written texts (HWT). We validated this hypothesis across different LLMs and observed significant differences in neural activation patterns when processing these two types of texts. Based on this, we propose RepreGuard, an efficient statistics-based detection method. Specifically, we first employ a surrogate model to collect representation of LGT and HWT, and extract the distinct activation feature that can better identify LGT. We can classify the text by calculating the projection score of the text representations along this feature direction and comparing with a precomputed threshold. Experimental results show that RepreGuard outperforms all baselines with average 94.92% AUROC on both in-distribution (ID) and OOD scenarios, while also demonstrating robust resilience to various text sizes and mainstream attacks. Data and code are publicly available at: https://github.com/NLP2CT/RepreGuard