Cases of fake news are increasing at a high pace, especially through social media portals. Even though there are various fact checker websites and portals that distinguish news from real or fake news and ongoing research have been done to stop the propagation of fake news. , The major setback is timely detection, prevention and reporting of fake news. The proposed model CBFDR (Context-Based Fake News Detection and Reporting) first analyzes the domain of the given news through Amazon web services (AWS) Comprehend, secondly detects the fake news through RNN-LSTM (Recurrent neural network – Long short term memory & XG-Boost algorithm (extreme gradient) and thirdly includes a set of domain-experts who analyze the set of fake news as well as report in Block-chain (using smart contract and proposed algorithm Proof of Evidence (PoE)) to stop the propagation of fake news at an early stage. The given paper has shown the highest success rate of 98% using RNN-LSTM and 93% using the XG-Boost algorithm for fake news detection. This paper covers detection, prevention and reporting in one application with an average 60-70% faster and more accurate than the existing models.