The intelligent cyber security model for intrusion detection with federated machine learning is based on distributed learning protocols for processing data and training models while preserving data security and privacy. Data owners can use the federated machine learning architecture to create a standard intrusion detection system by transferring their data without revealing private information. Taking a global approach to fraud management, models based on predictive analysis and anomaly detection are developed using a federated learning model. By leveraging unsupervised machine learning algorithms, the system can find new and unconventional ways fraudsters make a move by recognizing intricate relationships within them. In addition, the system can develop an adaptive intrusion detection solution with current new profile downloads and model training. This model is a handy and effective mechanism culminating in distributed architectures and proper data processing protocols to develop radical improvements in security systems to counter cyberattacks. Also, the model seeks to enhance cyber security systems since federated learning combines the strength that comes with advances in data analysis techniques, which helps in the detection and response to cyberattacks.