: Rapid growth of misleading information on social network has become a major obstacle, affecting people‘s opinion, dispersing misinformation and frightening the authenticity of virtual content. Although there are several ML models that have been proposed to classify the given data but still there is a scope of improvement which has been achieved through fine tuning the proposed methodology. This paper presents an automated model to detect fake news from social networks using a stack classifier of logistic regression, decision tree, support vector machine, and XGBoost. The given model is examined using a massive dataset of news reports obtained from several social media platforms, thereby undergoing a pre-processing pipeline of helping words removal, uppercase transformation achieving a higher accuracy rate of 95.79% in classifying between fake and reliable news which is higher than the state-of-art models. The approach of this paper is to express the potential of machine learning in computerizing the detection process that results in the reduction of manual examination and improves the efficiency of fake news recognition.