Fake News and Image Detection Using Text Analysis and AI Generated Image Identification

Nowadays,the rapid proliferation of digital news and social media has amplified the spread of misinformation, often supported by artificially generated or manipulated images. This project introduces a multimodal fake news detection system that evaluates both textual content and visual media to improve reliability and transparency. The text analysis branch employs advanced Natural Language Processing (NLP) techniques using transformer-based models such as BERT to classify news headlines and articles as real or fake. The image analysis branch leverages state-of-the-art deepfake and AI-generated image detection models (e.g., ResNet, EfficientNet) to determine whether associated images are authentic or artificially generated. Unlike traditional approaches that fuse text and image into a single prediction, the proposed system outputs separate results for textual and visual content, enabling users to identify whether misinformation arises from text, image, or both. Users can input news text, upload an image, or provide both, and the system will return predictions with probability scores.