Cross-Layer Semantic Security

Artificial intelligence (AI)-native wireless networks are regarded as a promising communication technology for next-generation and beyond systems, owing to their high reliability, low latency, and active cognitive capabilities. In contrast to the Shannon communication paradigm, which emphasizes bit-level efficiency at the physical layer, the AI-native framework prioritizes the meaning and utility of information at the application layer, with semantic communication (SemCom) being a typical instance. Nonetheless, the broadcast characteristic of wireless transmission introduces security risks including eavesdropping and hostile interventions. It is noteworthy that AI-native communications introduce both novel security challenges and opportunities.