Generative AI for Wireless Communication and Sensing: Toward Unified Foundation Models

Generative Artificial Intelligence (GenAI) has made significant advancements in fields such as computer vision (CV) and natural language processing (NLP), demonstrating its capability to synthesize data, learn representations, and improve generalization across tasks. Recently, there has been growing interest in integrating GenAI into wireless systems to advance both communication and sensing capabilities. Through data augmentation, domain adaptation, and cross-modal synthesis, GenAI can improve communication and sensing performance, support multimodal signal understanding and emerging wireless spatial intelligence, and provide shared representations and generative priors for unified signal modeling. This survey presents an overview of GenAI for wireless communication and sensing from two complementary perspectives. First, we examine how generative models can be integrated into wireless data processing pipelines, either as plugins that augment discriminative systems or as solvers that perform end-to-end tasks. Second, we review representative generative architectures, including Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), diffusion models, and Transformer-based approaches, and discuss how they provide shared generative mechanisms relevant to both communication and sensing through signal generation, prediction, and representation learning. Finally, we identify key challenges in integrating GenAI with wireless systems and discuss future directions toward wireless foundation models, which represent unified and pre-trained frameworks capable of scalable, adaptable, and efficient signal understanding across communication and sensing tasks.