Leveraging LLM-Generated Explanations for Detecting Emotionally Rewritten Fake News
Summary
This paper introduces a Gated Cross Attention (GCA) framework that leverages LLM-generated explanations as stable background knowledge to detect fake news that has been rewritten under different emotional framings while preserving the underlying factual claims, showing improvements on PolitiFact, GossipCop, and LUN datasets.
View Cached Full Text
Cached at: 10/08/26, 12:38 PM
# Leveraging LLM-Generated Explanations for Detecting Emotionally Rewritten Fake News Source: [https://arxiv.org/abs/2610.08835](https://arxiv.org/abs/2610.08835) [View PDF](https://arxiv.org/pdf/2610.08835)[HTML \(experimental\)](https://arxiv.org/html/2610.08835v1) > Abstract:The spread of fake news may cause severe social consequences\. Existing fake news detection methods mainly focus on stylistic variations or incorporate external information such as explanations\. However, news articles are often rewritten under different emotional backgrounds while preserving their underlying factual claims, which may affect the robustness of detection models\. In this work, we investigate fake news detec\- tion under fact\-preserving emotional variations\. To study this problem, we construct emotion\-rewritten test sets and generate explanations from the original news articles as stable background knowledge\. We then propose a Gated Cross Attention \(GCA\) framework that adaptively integrates emotionally rewritten news with the corresponding explanations, enabling the model to focus on informative explanation content while reducing potential mismatches caused by emotional reframing\. Experiments on PolitiFact, GossipCop, and LUN demonstrate that the proposed method achieves notable improvements under multiple emotional conditions on PolitiFact and LUN, while maintaining competitive performance on GossipCop\. We further analyze the effects of explanation guidance and gating mechanisms under different emotional conditions\. Our code and data are available at:[this https URL](https://github.com/Flulike/fakenews)gca \. ## Submission history From: Zekun Yang \[[view email](https://arxiv.org/show-email/40f8cd3e/2610.08835)\] **\[v1\]**Tue, 29 Sep 2026 11:50:30 UTC \(852 KB\)
Similar Articles
From Generation to Detection: Exploration of Discourse Driven Scenario based LLM Generated Fake News
The study examines how large language models generate and detect fake news under different scenarios, revealing variations in performance and that refined prompts do not always improve detection.
Fake News Theories: Harnessing Disciplinary Insights for Computational Modeling, Detection, and Explanation
This paper introduces a theory-informed computational framework for automated fake news detection and explanation, integrating cross-disciplinary insights from social sciences and psychology, and validates it through experiments on benchmark datasets.
Latent Fact-Checking: Detecting Misinformation through Activation Engineering
This paper introduces a misinformation detection framework using activation engineering, projecting last-token activations onto a learned 'falsehood direction' in LLM residual streams. It evaluates across Gemma, Llama, and Qwen models on fact-checking benchmarks, showing truthfulness is linearly separable in latent space.
Can Multimodal Large Language Models Generate and Detect Multimodal Social Media Fake News?
An EMNLP Findings 2026 paper introduces a multi-agent framework (story, image, and critic agents) that generates over 9,000 multimodal fake news posts and benchmarks 16 open- and closed-source MLLMs, finding they fall short of human-level detection accuracy, especially at judging image authenticity.
Cross-Prompt Generalization in Detecting AI-Generated Fake News Using Interpretable Linguistic Features
Researchers from Kennesaw State University investigate cross-prompt generalization in detecting AI-generated fake news using interpretable linguistic features (lexical diversity, readability, emotion). A random forest classifier trained on one prompting strategy and tested on another achieves AUC values of 0.988–1.000, suggesting these features capture stable, generalizable properties of AI-generated text.