---

Semantic communication (SC) and generative artificial intelligence (AI) are two cutting-edge technologies that can create a synergetic effect. SC aims to convey the underlying meanings or semantics of the data, tailored for specific tasks or goals. Recent studies on SC have utilized autoencoder neural network architectures to achieve these goals. SC significantly reduces communication latency while maintaining reliability, leveraging the compact and resilient nature of semantic representations. Fully unlocking the potential of SC relies on the ability to decode and interpret these conveyed semantics, which ultimately determines the minimum transmitted semantic information. Recent advances in generative AI have demonstrated their capability to generate high-fidelity, large-scale images from text prompts. In this talk, the speaker will propose an integration of generative AI into SC, introducing a novel semantics-native SC framework with selected examples in multi-user and multi-modal data scenarios. Furthermore, he will discuss the applications of these principles in higher-layer designs, such as medium access control signal designs, showcasing their potential for broader impact and innovation.