The changing era of digital technology and the use of 5G networks have increased the information available to the users in abundance and at a low price. As the use of Internet and online social networks has increased tremendously, which provides a platform to exchange thoughts, ideas, opinions, and feelings with other people. However, it has led to various forms of online problems such as offensive content, hate speech, fake news, racism, trolling, etc. Hence it becomes crucial to check the truth of the fact, which gets propagated ten times faster than the other text and influences or distracts the people from perceiving the correct information. It also affects in making the right decision by oneself. As fake news content is used to mislead people, it may also contain sarcasm. People will find sarcastic news to be funny and if it is fake, it becomes essential to get rid of these texts. To overcome these issues, hand crafted features along with hybrid machine learning and deep learning models are incorporated to solve these problems on news and social media text. Fake news detection using machine learning and deep learning on bench mark dataset resulted 75% accuracy. When it was applied to Indic language using transfer learning it resulted to 96% accuracy. Handcrafted features were used for detecting sarcasm in social media text which resulted into 81% accuracy, whereas GloVe could result achieve 86%. Comparision of proposed methods with existing work is discussed in this chapter.