Beyond the Final Actor: Modeling the Dual Roles of Creator and Editor for Fine-Grained LLM-Generated Text Detection
2026-04-06 • Computation and Language
Computation and Language
AI summaryⓘ
The authors address the problem of detecting text created or modified by large language models (LLMs) with more detail than before. Instead of just saying if text is human-made or AI-made, they distinguish four types, including mixed texts where humans and LLMs both contribute. They introduce a method called RACE that uses rhetorical analysis to spot differences in writing style and logical structure to tell who created or edited the text. Their tests show RACE works better than other methods and helps with more precise policy decisions about AI-generated content.
Large Language ModelsSynthetic Text DetectionRhetorical Structure TheoryElementary Discourse UnitText ClassificationCreator-Editor ModelingFine-grained DetectionAI Text Regulation
Authors
Yang Li, Qiang Sheng, Zhengjia Wang, Yehan Yang, Danding Wang, Juan Cao
Abstract
The misuse of large language models (LLMs) requires precise detection of synthetic text. Existing works mainly follow binary or ternary classification settings, which can only distinguish pure human/LLM text or collaborative text at best. This remains insufficient for the nuanced regulation, as the LLM-polished human text and humanized LLM text often trigger different policy consequences. In this paper, we explore fine-grained LLM-generated text detection under a rigorous four-class setting. To handle such complexities, we propose RACE (Rhetorical Analysis for Creator-Editor Modeling), a fine-grained detection method that characterizes the distinct signatures of creator and editor. Specifically, RACE utilizes Rhetorical Structure Theory to construct a logic graph for the creator's foundation while extracting Elementary Discourse Unit-level features for the editor's style. Experiments show that RACE outperforms 12 baselines in identifying fine-grained types with low false alarms, offering a policy-aligned solution for LLM regulation.