Real, Fake, or Manipulated? Detecting Machine-Influenced Text
This addresses the challenge of distinguishing between benign and malicious uses of LLMs in text generation for applications like misinformation detection, though it is incremental over prior MGT detection work.
The paper tackles the problem of detecting fine-grained machine-influenced text types, such as human-written, machine-generated, machine-polished, and machine-translated, and introduces HERO, a hierarchical detector that outperforms state-of-the-art methods by 2.5-3 mAP on average across experiments.
Large Language Model (LLMs) can be used to write or modify documents, presenting a challenge for understanding the intent behind their use. For example, benign uses may involve using LLM on a human-written document to improve its grammar or to translate it into another language. However, a document entirely produced by a LLM may be more likely to be used to spread misinformation than simple translation (\eg, from use by malicious actors or simply by hallucinating). Prior works in Machine Generated Text (MGT) detection mostly focus on simply identifying whether a document was human or machine written, ignoring these fine-grained uses. In this paper, we introduce a HiErarchical, length-RObust machine-influenced text detector (HERO), which learns to separate text samples of varying lengths from four primary types: human-written, machine-generated, machine-polished, and machine-translated. HERO accomplishes this by combining predictions from length-specialist models that have been trained with Subcategory Guidance. Specifically, for categories that are easily confused (\eg, different source languages), our Subcategory Guidance module encourages separation of the fine-grained categories, boosting performance. Extensive experiments across five LLMs and six domains demonstrate the benefits of our HERO, outperforming the state-of-the-art by 2.5-3 mAP on average.