Atrial fibrillation (AF) is a major cause of heart failure and ischemic stroke, making early detection and intervention critically important. However, timely ECG recording during paroxysmal episodes is often difficult, leading to delayed diagnosis. Recently, an AI-enhanced 12-lead ECG equipped with a hidden AF risk estimation function has been introduced. This technology analyzes sinus rhythm ECGs and stratifies the likelihood of prior AF into four risk categories. Although this novel approach may facilitate earlier AF detection and optimize the timing of therapeutic intervention, its clinical accuracy and real-world utility remain insufficiently validated. Therefore, this multicenter study aims to evaluate the diagnostic performance and clinical usefulness of AI-based AF risk assessment and to clarify its association with subsequent AF incidence and patient outcomes.