Abstarct: In recent years, with the proliferation of fake news on news websites and social media platforms, the need to identify such news has become increasingly critical. This study aims to detect fake news within the context of social networks. The main research problem is the effective and accurate detection of fake news amid vast amounts of textual news data. The primary focus of this research is on improving the accuracy of categorizing fake and real news. One of the challenges in this area pertains to the classification labels of the data. Most data in this domain is binary, but the dataset used in this study is multi-class, which we have employed for multi-class classification. Another challenge in this field is data privacy and security, which is addressed in two scenarios examined in this study. In the first scenario, using federated learning and graph neural networks, we tackled the privacy and data security issues and obtained promising results. In the second scenario, which is the main focus of this research, we examined graph neural network models for detecting and identifying fake and real news. In this scenario, we evaluated several graph models, with the best result being an accuracy of 97.34%, which outperforms other graph neural network models explored in this study.