The proliferation of social media has revolutionized information dissemination, leading to an exponential increase in the spread of news. However, this digital transformation has also given rise to the pervasive issue of fake news, which poses significant threats to society. This paper addresses the critical challenge of detecting fake news on social media platforms. We present an innovative approach leveraging deep learning techniques to identify and quantify the falseness of news statements. Our model integrates contextual word embeddings with attention mechanisms, enriched by relevant metadata to enhance detection accuracy. By employing this sophisticated architecture, we aim to distinguish between authentic and fraudulent news with greater precision. The proposed model's performance is evaluated on multiple real-world datasets, including the LIAR dataset, where it achieved a notable accuracy of 66%. Our findings underscore the potential of advanced machine learning methods to mitigate the impact of fake news and contribute to more reliable information consumption in the digital age.