In the context of smart grid security, this paper explores the integration of Federated Learning (FL) with eXplainable Artificial Intelligence (XAI) techniques to address the dual challenges of data privacy and model interpretability. We employ a federated learning framework with a 3-hidden-layer neural network, trained on diverse smart grid datasets. Our approach includes comprehensive data preprocessing strategies, such as the removal of duplicates and handling of missing values, as well as the application of mutual information for rigorous feature selection. To tackle the opacity of complex machine learning models, we incorporate SHapley Additive exPlanations (SHAP) to elucidate the decision-making processes, enhancing transparency and trust. The model’s effectiveness is validated through improved performance metrics, including accuracy, precision, recall, and F1 scores, across different data distributions. Our results underscore the potential of combining FL with SHAP to enhance the interpretability and reliability of AI applications in energy systems, highlighting significant implications for real-world deployments where both security and privacy are paramount.