Smart cities (SC) are popular because of the digitization of urban planning and the integration of data from fog computing and the Internet of Things (IoT). However, the proliferation of IoT devices makes them vulnerable to cyberattacks, such as denial of service. To address this issue, there is a need for an intelligent approach that can investigate and detect cybersecurity issues in SC, including complex malicious response injection and malicious state command injection attacks. Artificial intelligence (AI) is crucial for detecting cyberattacks; however, AI faces challenges in adapting to the ever-changing attack landscape. The combination of extreme gradient boosting (XGBoost) and recurrent neural networks (RNN) is applied in context through time series analysis of abnormal changes in data trends and binary classification to predict cybersecurity attack behavior. In this study, we employed an intelligent approach based on XGBoost and RNN to enhance cyberattack detection in SC. We evaluated the proposed methods using datasets from cyber-physical systems, specifically water storage tank systems, which encompass various forms of cyberattacks. The performance results of the two methods demonstrate that XGBoost outperforms RNN in terms of all metrics in identifying true positive cyberattack cases. The RNN model had an accuracy value of 90%, whereas XGBoost had an accuracy value of 97%. These models demonstrate good performance in terms of accuracy and computational complexity.