Facing the urgent demand for high-performance, customized electromagnetic protection materials driven by increasingly intelligent electronic information systems, traditional research and development (R&D) models face severe limitations due to complex multi-parameter coupling, high trial-and-error costs, and difficulties in cross-scale design, hindering their ability to meet the need for efficient R&D. Artificial intelligence (AI), leveraging data-driven approaches and algorithmic optimization, offers a transformative paradigm to overcome these limitations. This paper systematically reviews AI-empowered research in electromagnetic protection materials. It begins by analyzing the key characteristics and core challenges in the R&D of these materials, highlighting the high suitability of AI for this domain. Subsequently, it illustrates representative research cases from both forward prediction and inverse design perspectives within the field. Finally, the paper identifies existing challenges concerning data availability, physical interpretability of AI models, and practical application deployment barriers. Specific considerations are proposed in three aspects: constructing specialized electromagnetic material gene databases, developing physics-informed neural networks that integrate data with physical principles, and emphasizing the need to promote domain-specific data sharing and establish standardized protocols, so as to pave the way for the intelligent development of next-generation electromagnetic protection materials.