The widespread growth of social and traditional media in the last decade has led to the spread of information. Nowadays, these platforms are playing a vital role in everyone's life. Many use these platforms to share and discuss ideas and issues in almost every field. Getting the most recent news at their fingers has become simpler through the internet. However, in this dissemination of information, it has been noted that some misinformation and fake news are circulating with no relevance to reality, which can cause serious problems in political and social aspects. The main goal of this study is to find the ideal learning model for achieving high accuracy and performance. In this regard, this study offers a state-of-the-art dataset that extends fake and real news datasets called the Misleading Content dataset. To analyze the performance of the dataset, three different machine learning algorithms, Random Forest, Naïve Bayes (NB) and Support Vector Machine (SVM), and three different deep learning, Bidirectional Long Short-Term Memory (BiLSTM), Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) models are being used. BiLSTM performs well and provides promising results along with 88.54% accuracy.