Vehicular Ad-hoc Networks (VANETs) are necessary to the advancement of intelligent transportation systems, enabling communication among vehicles and roadside units (RSUs). However, the highly dynamic topology, frequent disconnections, and high mobility of nodes in VANETs introduce significant security challenges. Traditional intrusion detection systems (IDS) are often inadequate in VANETs due to their inability to adapt to these evolving conditions and detect complex, emerging threats effectively. This paper explores the application of machine learning techniques specifically designed to enhance the security of VANETs by accurately identifying potential intrusions. We implemented and evaluated several machines learning models, comprising Support Vector Machines (SVM), Random Forests, and Neural Networks, using the VeReMi Dataset. Extensive preprocessing and feature selection techniques, such as normalization and the Synthetic Minority Over-sampling Technique (SMOTE), were applied to optimize model performance. According to our findings, regarding all the models we examined, the Neural Network model had the greatest accuracy and the lowest false positive rate, which made it especially well-suited for real-time applications in VANET environments.