The paradigm shift: A comprehensive survey on large vision language models for multimodal fake news detection

2026-01-16

Computer Science Review

AI Summary

Multimodal fake news detection has long been constrained by shallow feature fusion, struggling to capture high-level semantics and complex cross-modal interactions. This work systematically reviews the paradigm shift driven by Large Vision-Language Models (LVLMs), tracing the evolution from traditional approaches to end-to-end unified reasoning frameworks. It presents the first structured taxonomy encompassing model architectures, datasets, and evaluation benchmarks. By mapping the trajectory of technical advancements, the study offers an in-depth analysis of critical challenges—including interpretability, temporal reasoning, and domain generalization—and constructs a comprehensive technical landscape for multimodal fake news detection in the LVLM era, providing both theoretical insights and practical guidance for future research.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Computer Vision: Multi-modal Vision

Application Category

• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Problem

• multimodal fake news detection
• large vision-language models
• paradigm shift
• systematic survey
• cross-modal reasoning

Innovation

• Large Vision-Language Models
• Multimodal Fake News Detection
• Paradigm Shift
• Foundation Models
• Cross-Modal Reasoning