Anxiety is a rising concern for mental health professionals. Traditional therapy methods involving stand-alone Cognitive Behavioral Therapy (CBT) are a gold-standard technique for social anxiety management, yet face challenges such as patient hesitance and logistical barriers. Technology-assisted therapy methods, such as teletherapy and digital platforms, are utilized to increase the accessibility to treatment and make it less alarming. A core component of CBT, Exposure Therapy, is reliant on real-world stimuli that can limit its control and consistency. Virtual Reality Exposure Therapy (VRET) emerges as a solution to simulate a virtual world with precisely controlled stimuli for exposure and aims to augment the therapeutic potential of CBT. VRET utilizes Artificial Intelligence (AI) techniques to make a session more interactive, through chatbots, virtual reality therapists, and voice and facial recognition technology. The effectiveness of AI-enabled VRET simulations is enhanced through the utilization of Generative Artificial Intelligence (GAI) methods. GAI approaches address limitations such as suboptimal sample generation and limited creativity presented by traditional AI models. GAI models can dynamically generate realistic environments and avatars, automating virtual world creation. Similarly, the trajectory of GAI-enabled Virtual Reality (VR) research in addressing social anxiety management needs to be defined clearly. To the authors’ knowledge, no published review article addressing the potential and limitations of GAI-enabled VRET is available to date. Similarly, this study reviews recent literature on VR approaches for social anxiety management, presents a taxonomy to highlight key components of AI-enabled VRET, and discusses the future directions for GAI-enabled VRET.