Despite progress in semantic communication (SemCom), research on SemCom security is still in its infancy. To bridge this gap, we propose a general covert SemCom framework for wireless networks, reducing eavesdropping risk. Our approach transmits semantic information covertly, making it difficult for wardens to detect. Given the aim of maximizing covert SemCom performance, we formulate a power control problem in covert SemCom under energy constraints. Furthermore, we propose a learning-based approach based on the soft actor-critic algorithm, optimizing the power of the transmitter and friendly jammer. Numerical results demonstrate that our approach effectively enhances the performance of covert SemCom.

翻译：尽管语义通信（SemCom）取得了进展，但其安全性研究仍处于起步阶段。为弥补这一空白，我们提出了一种适用于无线网络的通用隐蔽语义通信框架，以降低窃听风险。该方法隐蔽地传输语义信息，使得监测方难以察觉。以最大化隐蔽语义通信性能为目标，我们在能量约束下构建了隐蔽语义通信中的功率控制问题。此外，我们提出了一种基于柔性演员-评论家（soft actor-critic）算法的学习方法，用于优化发射端与协作干扰机的功率配置。数值结果表明，所提方法能有效提升隐蔽语义通信的性能。