Intrusion Detection Systems (IDS) are essential for enhancing cybersecurity in modern vehicular networks, especially as they become increasingly interconnected. However, traditional IDS approaches often face limitations in handling complex attack patterns, evolving threat landscapes, and the high volume of network data generated. This paper examines the incorporation of DL methodologies, including LSTM networks, CNNs, autoencoders, and DRL, within IDS frameworks. These methods offer enhanced detection accuracy, real-time anomaly identification, and adaptability, addressing key challenges in IDS deployment for vehicular networks. This review study highlights the improvements in IDS effectiveness and the future directions for DL-driven cybersecurity in connected and autonomous vehicles.