Lightweight Stylistic Consistency Profiling: Robust Detection of LLM-Generated Textual Content for Multimedia Moderation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Siyuan, Wulianghai, Aodu, Lin, Xi, Yuan, Xibin, Mao, Qinghua, Li, Guangyan, Chen, Xiang, Wu, Jun, Li, Jianhua
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913098506436608
author Li, Siyuan
Wulianghai, Aodu
Lin, Xi
Yuan, Xibin
Mao, Qinghua
Li, Guangyan
Chen, Xiang
Wu, Jun
Li, Jianhua
author_facet Li, Siyuan
Wulianghai, Aodu
Lin, Xi
Yuan, Xibin
Mao, Qinghua
Li, Guangyan
Chen, Xiang
Wu, Jun
Li, Jianhua
contents The increasing prevalence of Large Language Models (LLMs) in content creation has made distinguishing human-written textual content from LLM-generated counterparts a critical task for multimedia moderation. Existing detectors often rely on statistical cues or model-specific heuristics, making them vulnerable to paraphrasing and adversarial manipulations, and consequently limiting their robustness and interpretability. In this work, we proposeLiSCP , a novel lightweight stylistic consistency profiling method for robust detection of LLM-generated textual content, focusing on feature stability under adversarial manipulation. Our approach constructs a consistency profile that combines discrete stylistic features with continuous semantic signals, leveraging stylistic stability across multimodal-guided paraphrased text variants. Experiments spanning real-world multimedia news and movie datasets and conventional text domains demonstrate that LiSCP achieves superior performance on in-domain detection and outperforms existing approaches by up to 11.79% in cross-domain settings. Additionally,it demonstrates notable robustness under adversarial scenarios, including adversarial attacks and hybrid human-AI settings.
format Preprint
id arxiv_https___arxiv_org_abs_2605_05950
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Lightweight Stylistic Consistency Profiling: Robust Detection of LLM-Generated Textual Content for Multimedia Moderation
Li, Siyuan
Wulianghai, Aodu
Lin, Xi
Yuan, Xibin
Mao, Qinghua
Li, Guangyan
Chen, Xiang
Wu, Jun
Li, Jianhua
Computation and Language
The increasing prevalence of Large Language Models (LLMs) in content creation has made distinguishing human-written textual content from LLM-generated counterparts a critical task for multimedia moderation. Existing detectors often rely on statistical cues or model-specific heuristics, making them vulnerable to paraphrasing and adversarial manipulations, and consequently limiting their robustness and interpretability. In this work, we proposeLiSCP , a novel lightweight stylistic consistency profiling method for robust detection of LLM-generated textual content, focusing on feature stability under adversarial manipulation. Our approach constructs a consistency profile that combines discrete stylistic features with continuous semantic signals, leveraging stylistic stability across multimodal-guided paraphrased text variants. Experiments spanning real-world multimedia news and movie datasets and conventional text domains demonstrate that LiSCP achieves superior performance on in-domain detection and outperforms existing approaches by up to 11.79% in cross-domain settings. Additionally,it demonstrates notable robustness under adversarial scenarios, including adversarial attacks and hybrid human-AI settings.
title Lightweight Stylistic Consistency Profiling: Robust Detection of LLM-Generated Textual Content for Multimedia Moderation
topic Computation and Language
url https://arxiv.org/abs/2605.05950