FactLense: An AI-Driven Comparative Approach to Detecting Online News Deception

The quick growth of online news sources has been matched by a worrying rise in the amount of false online news, such as misinformation and disinformation. This research introduces FactLense, an AI-driven framework for detecting online news deception through a comparative analysis of advanced deep learning and conventional machine learning methodologies. Using preprocessing, TF-IDF vectorization, and transformer-based embeddings (like BERT), the system combines linguistic, semantic, and statistical features to give a better understanding of the context. A hybrid classification system is developed, integrating gradient boosting and neural networks to maximize accuracy. The system is tested and trained on well-balanced sets of labeled true and false news, with results measured using F1-score, recall, and precision. Experiments have produced results showing that FactLense is more accurate than the baseline models and offers useful AI insights for enhanced transparency. The model is designed to be scalable for application in journalism, social media management, and digital forensics to counter misinformation on the internet.