Antimicrobial peptides (AMPs) have become a major focus in the development of novel antimicrobial agents due to their broad-spectrum antimicrobial properties and extremely low risk of inducing resistance. Traditional AMP design strategies have relied primarily on the isolation and extraction of natural products, along with limited sequence modifications, which has greatly restricted the discovery of new candidate molecules and led to inefficient in vitro screening processes. With rapid advancements in artificial intelligence (AI), significant breakthroughs have been achieved in the virtual screening of AMPs, preliminary prediction of antibacterial activity, de novo design of peptide sequences, and targeted optimization of peptide structures and functions, thereby significantly shortening research timelines and reducing overall development costs. This article systematically reviews the core features of mainstream antimicrobial peptide databases and the fundamental mechanisms of commonly used AI algorithms, and further elaborates on feasible strategies for applying AI to antimicrobial peptide screening, design, and structural optimization. At the same time, it identifies key bottlenecks in current research, such as the lack of standardized datasets, immature in vivo evaluation systems, and insufficient interdisciplinary collaboration. This review provides valuable insights and theoretical support for the subsequent research, translational development, and clinical application of next-generation antimicrobial peptides.