The VANETs system is used to enhance Transportation Systems with intelligence, which would be attained by V2V and V2I information sharing so as to ensure road safety and efficient traffic in road. VANETs are characterized by the unpredictable nature which presents challenges in spectrum allocation, which would affect the communication reliability and network performance. The machine learning technologies including Support Vector Machines (SVM) and Reinforcement Learning (RL) algorithms are introduced in this paper to introduce a new method of dynamic spectrum allocation in VANETs. Based on past data, traffic patterns and network conditions in a given area are foreseen using SVM. To rapidly allocate spectrum we use Reinforcement Learning (RL). We improve communication stability through spectrum allocation by looking at past data and other sources of information, such as environment conditions. We compare our method with conventional static spectrum allocation techniques through proper simulations. The outcomes indicate that the network capacity has been increased with our AI powered method, latency has been reduced, and spectrum usage has been maximized. Therefore, it has become a feasible technology that can be applied to improve the efficiency and reliability of VANET communication systems.