AI-Powered Fake News Analysis for Enterprise Security

Abstract Disinformation aimed at companies can erode reputation, move markets, and disrupt operations faster than traditional security teams can respond. This article proposes an AI-enabled framework for enterprise-grade fake news analysis that combines multi-modal detection, credibility modeling, and risk-aware response. The approach integrates transformer-based language models for claim and stance detection with image–video forensics to flag manipulated media, and graph learning to identify coordinated inauthentic behavior and propagation paths across social platforms and news aggregators. A source reputation engine fuses historical reliability, network centrality, and provenance signals to generate dynamic trust scores, while a knowledge-graph layer validates claims against authoritative corporate and regulatory data. The framework produces an incident-level risk score aligned.