Abstract: As a core non-invasive tool for diagnosing cardiovascular diseases, traditional manual electrocardiogram (ECG) analysis suffers from limitations such as low diagnostic consistency, difficulty in adapting ECG morphology for specific populations, high rates of missed diagnoses of dynamic arrhythmias, and delayed responses to acute events, etc. Artificial intelligence–enabled ECG (AI-ECG) analysis is driving profound transformation in the diagnosis and treatment of cardiovascular diseases. In this article, the technological evolution of AI-ECG from traditional machine learning and deep learning to generative AI/large language models and scenario-specific models, as well as its applications in arrhythmias, structural heart disease, acute coronary syndrome, cardiac rehabilitation, and wearable monitoring, etc. AI-ECG has achieved breakthroughs from "static diagnosis" to "dynamic early warning" and from "single-disease screening" to "full-cycle management." However, AI-ECG still faces challenges including insufficient interpretability, lack of data and evaluation standardization, model bias, and obstacles in clinical translation. Subsequently, multimodal integration, customized models for special populations, standardization of dynamic ECGs, and interdisciplinary collaboration are needed to advance AI-ECG from "technical feasibility" to "patient benefit", offering a new idea for precise diagnosis and treatment of cardiovascular diseases.