aiqbench analysis
This research addresses the critical limitation of black-box detectors by introducing a graph-based framework that captures high-order structural dependencies in LLM outputs. By transforming text into lexical co-occurrence graphs, the proposed method achieves state-of-the-art detection performance with significantly higher faithfulness than traditional token-level saliency approaches. The work advances AI safety and transparency through interpretable motifs derived from probabilistic graphical models rather than linear sequences.
Towards Structurally Explainable Machine-Generated Text Detection: A Graph-Perspective Framework
arXiv:2505.12507v2 Announce Type: replace Abstract: Despite the success of machine-generated text detectors, the black-box nature remains a critical limitation. Traditional explainability methods rely on token-level saliency, insufficient to reveal the high-order structural dependencies that distinguish LLM outputs. In this paper, we propose \textsc{LM$^2$otifs}, a principled framework that shifts detection from linear sequences to graph-structured manifolds. We first provide a theoretical grounding based on probabilistic graphical models, demonstrating that detection performance is more distinguishable in the graph-topological space. Driven by this theory, \textsc{LM$^2$otifs} transforms text into lexical co-occurrence graphs to preserve latent structural fingerprints. The framework employs Graph Neural Networks for robust detection and utilizes graph-specific explainers to extract interpretable motifs. Crucially, our experiments reveal that these structural motifs achieve higher faithfulness compared to traditional methods. This empirical evidence confirms the existence of high-order structural explanations that linear methods fail to capture. Experimental results show that \textsc{LM$^2$otifs} achieves state-of-the-art performance while providing multi-level \textit{distinct linguistic fingerprints} that are more faithful to the model's decision.
Abstract:Despite the success of machine-generated text detectors, the black-box nature remains a critical limitation. Traditional explainability methods rely on token-level saliency, insufficient to reveal the high-order structural dependencies that distinguish LLM outputs. In this paper, we propose \textsc{LM$^2$otifs}, a principled framework that shifts detection from linear sequences to graph-structured manifolds. We first provide a theoretical grounding based on probabilistic graphical models, demonstrating that detection performance is more distinguishable in the graph-topological space. Driven by this theory, \textsc{LM$^2$otifs} transforms text into lexical co-occurrence graphs to preserve latent structural fingerprints. The framework employs Graph Neural Networks for robust detection and utilizes graph-specific explainers to extract interpretable motifs. Crucially, our experiments reveal that these structural motifs achieve higher faithfulness compared to traditional methods. This empirical evidence confirms the existence of high-order structural explanations that linear methods fail to capture. Experimental results show that \textsc{LM$^2$otifs} achieves state-of-the-art performance while providing multi-level \textit{distinct linguistic fingerprints} that are more faithful to the model's decision.
Computation and Language (cs.CL); Computers and Society (cs.CY)
arXiv:2505.12507 [cs.CL]
(or arXiv:2505.12507v2 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2505.12507 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 255 words
Related coverage
OpenRouter: from Seed to Stripe — with OpenRouter’s Alex Atallah & AMP’s Anjney Midha
Latent Space · Sep 25, 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · Sep 18, 2026
Presentation: Running AI at the Edge: Running Real Workloads Directly in the Browser
InfoQ · Aug 31, 2026
Game-Theoretic Disempowerment: Loss of Control to Agencyless AI
LessWrong · Sep 27, 2026