6G networks can flexibly arrange network resources in multiple areas through integration with edge computing and cloud computing. However, devices at the edge of the network generate large quantities of multimodal data (e.g., user behaviour records, audio, and video), creating new challenges for data security and privacy, which has led to research and practical applications of privacy protection methods based on information hiding. In this article, we propose an information hiding method based on multidimensional feature fusion for privacy protection in 6G networks. This method is based on the strong attack resistance of carriers carrying private information; we study multidimensional feature fusion arbitration methods and construct an efficient and accurate feature-based information hiding technology. In this paper, we first perform private information preprocessing to encrypt private information and make it more secure. Then, we perform multidimensional feature extraction on the carrier and fuse these features to obtain the final feature region. Finally, the step of hiding or extracting private information is performed. A large number of experiments show that the performance of the proposed method is superior to those of existing methods.