Fake news detection is an essential task; however, the complexity of several languages makes fake news detection challenging. It requires drawing many conclusions about the numerous people involved to comprehend the logic behind some fake stories. Existing works cannot collect more semantic and contextual characteristics from documents in a particular multilingual text corpus. To bridge these challenges and deal with multilingual fake news detection, we present a semantic approach to the identification of fake news based on relational variables like sentiment, entities, or facts that may be directly derived from the text. Our model outperformed the state-of-the-art methods by approximately 3.97% for English to English, 1.41% for English to Hindi, 5.47% for English to Indonesian, 2.18% for English to Swahili, and 2.88% for English to Vietnamese language reviews on TALLIP fake news dataset. To the best of our knowledge, our paper is the first study that uses a capsule neural network for multilingual fake news detection.