The rapid evolution of the Internet of Vehicles (IoV) introduces significant advancements in smart transportation systems, yet also presents critical challenges in data security, privacy, and real-time decision-making. This study proposes a Federated Learning (FL)-based security framework for IoV, integrating Federated Averaging (FedAvg) and Differential Privacy (DP) to enhance cybersecurity while preserving data privacy. The proposed model leverages decentralized machine learning techniques to mitigate security threats, reduce reliance on raw data transmission, and prevent unauthorized access to sensitive vehicle and user data. Through extensive empirical analysis using real-world cybersecurity datasets, this research evaluates the performance, scalability, and efficiency of FL-based security mechanisms compared to conventional approaches.