Social media platforms have become increasingly popular for news consumption, due to their speed, low cost, and accessibility. However, these platforms facilitated the widespread propagation of fake news, making it essential to ensure users received accurate information. Traditional methods are generally unsuccessful, false news is imitated by real news to mislead users. This highlighted the necessity to discover more advanced detection techniques. This research focused on using Graph Neural Networks (GNNs) to enhance fake news detection by incorporating social context into the detection process. Entities such as news, authors, publishers, users, article titles, and tweet_IDs are represented as nodes within a graph structure, with relationships between these entities transformed into numerical representations using techniques such as TF-IDF. The experiments are presented on three real-world datasets, which achieved significant improvements over traditional baseline methods. Finally, the suggested approach achieved accuracies of 0.95 on the BuzzFeed dataset, 0.87 on the PolitiFact dataset, and 0.91 on the Gossipcop dataset. These results performed its effectiveness in detecting fake news.