In IoT-enabled smart cities vehicular ad hoc networks (VANETs) are necessary for smart transport systems, yet combining them with 5G and high mobility settings presents significant safety concerns. The flexibility, immediate attack reaction, and new malicious attacks like Sybil, GPS spoofing, and AI-driven attacks that compromise accuracy of data, privacy and system reliability are challenges for traditional centralized security frameworks. As a way to tackle these challenges this research presents a unique AI-blockchain hybrid system which combines blockchain technology for decentralized trust management with machine learning (ML) for reactive identification of intrusions. Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) are two Deep Learning models which our approach uses for identifying evolving risks like DDoS and Man-in-the-Middle attacks with 96% accuracy. Additionally, blockchain guarantees unbreakable verification of identities and data origins which decreases Sybil attacks by 40%. Additionally, by overcoming delays and scalability problems that classical Public Key Infrastructure (PKI) has, Elliptic Curve Cryptography (ECC) strengthens confidentiality verification. By applying comprehensive investigation this study demonstrate how the combined system helps immediate action in changing VANET networks while also reducing growing security threats. The proposed model provides a strong basis for future smart cities where effectiveness and safety are crucial by promoting the development of safe, adjustable, and privacy aware modes of transport.