Recommender systems (RS) have become a significant tool for various online services to enhance user experiences by suggesting relevant items such as contents, movies, and services. However, these RS often struggle with the cold start problem when new users or items are introduced without sufficient interaction history. This paper proposes a novel approach using Generative Adversarial Networks (GANs) with the integration of auxiliary information to solve the cold start problem. Auxiliary information, such as user demographics, item attributes, and contextual data, is used to enhance the quality of generated interactions, helping to mitigate cold start problems. In this study, we proposed a GAN-based model that can concurrently consider multiple types of auxiliary information correspondingly. We benchmark our proposed model against state-of-the-art methods such as traditional models and GAN-based models. Extensive experiments on two public datasets (MovieLens and Book-crossing) show that our model outperforms the existing methods, particularly in cold start challenges. To achieve significant improvements, we utilized evaluation metrics such as Precision, Recall, NDCG, and MAE.