Advanced deep face detection: leveraging multi-model data and explainable AI for improved performance

The rise of fake news in the digital era presents a critical challenge, impacting public opinion, democratic processes, and societal trust. Traditional detection methods based on textual analysis are insufficient against sophisticated miss information involving images and videos. This research introduces an advanced fake news detection framework leveraging multi-modal data (images and videos) and integrating cutting-edge machine learning techniques like transformers and graph neural networks to capture complex cross- modal patterns. To enhance transparency, Explainable AI (XAI) techniques like (LIME AND SHAP) are incorporated, providing insights into the system’s decisions. The framework is designed for seamless use via a Telegram bot, enabling real-time detection and feedback for stakeholders like fact-checkers, media outlets, and policymakers. Experiments on real-world datasets highlight significant accuracy improvements, with XAI visualizations fostering trust and informed decision-making. This work demonstrates the potential of multi-modal data, advanced algorithms, and explain-ability in building a robust, user-friendly fake news detection system for a safer digital ecosystem.