Vehicle-to-everything (V2X) communication plays a pivotal role in revolutionizing the transportation sector, providing real-time, highly reliable, and actionable information exchange for safety, mobility, and environmental applications. It facilitates the seamless flow of data, including Basic Safety Messages, between vehicles and infrastructure, ensuring data integrity. This paper introduces an extensive deep neural network (DNN)-based multiclass attack detection model, MADNet, capable of identifying seven different types of BSM attacks, enhancing V2X threat assessment. The DNN architectures are deployed within roadside units (RSUs) to analyze vehicular data and classify message sequences as genuine or malicious. Multiple DNN architectures are employed, and their performance is evaluated using key metrics. Our leading deep learning model achieves an impressive 99.66% accuracy in attack detection.