Fake News Detection

The swift influx of false information through social media has made the verification of content prior to publishing a necessity. In this paper, we propose Spottix, a real-time fake news detection framework that continuously assesses news content before it goes live, thus impeding the flow of misleading information. The system comprises an input of text and/or image news from the user, a comparison with verified sources, and an output of the classification decision with explanations based on Explainable AI (XAI). The framework employs a finely tuned BERT-based model for text verification and OCR-supported extraction along with the same model for discerning false news in photos. Experimental results show 95.8% accuracy on text classification and 100% accuracy on verifiable historical claims with sub-3-second response times in real-world testing. The system justifies each prediction with evidence, making it appropriate for eventual integration into social media platforms.