Insider threats pose significant risks to the operational continuity and security of critical energy infrastructure. This paper presents a federated multi-modal system that integrates biometric and cyber data to detect insider threats with high accuracy while preserving data privacy. The proposed architecture combines facial recognition, keystroke dynamics, and network activity logs using a federated learning framework, enabling decentralized model training across multiple nodes. This approach reduces data exposure risks and supports compliance with privacy regulations. Experimental evaluations on synthetic and real-world datasets demonstrate the system’s effectiveness in identifying anomalous user behavior patterns, outperforming centralized baselines in both detection rate and resilience. The study offers a scalable, privacy-aware solution for securing energy systems against internal cyber-physical threats.