Fake news has become a growing issue in today's globe and One of the key reasons why false news is so hazardous and effective is the difficulty in recognizing it from actual news. Previous research acquired that fake news differed significantly from true bulletin in terms of lexicon and text structure, indicating the possibility of discriminating. Building high-quality fake news datasets is challenging because algorithms that detect false news require publicly accessible data for training and evaluation. Numerous researchers have worked to develop correct benchmark datasets of real and fake news collected on or after social media sites in an effort to find a solution to this issue, and as social media becomes more widely used, more people are likely to get their news from these platforms as opposed to more traditional news media sources. This paper's research focuses on two stages of fake news detection: characterization and disclosure. The first approach is to use social media to reveal the underlying concepts and principles of fake news. Existing solutions for detecting false news using various supervised learning algorithms are evaluated during the discovery process. So, in this research study era, we propose innovative machine learning Algorithms for detecting fake news in social environments and strive to define the subjects that fake news usually covers, discovering that fake news is frequently about counterfeit trolls on social media platforms. Although the model has proven to be useful, we counsel that future enhancements include additional testing on different datasets with a broader range of news sources.