Sarcopenia, a condition characterized by age-related muscle mass and function decline, poses significant risks including falls, fractures, gait disorders, and even mortality. This study aimed to develop an AI motion sensor system utilizing micro inertial measurement units (μIMUs) to screen sarcopeniaprone elderly subjects. Subjects within the age range of 65–84 years performed single sit-to-stand and 5-time chair stand while wearing two μIMUs. K-Nearest-Neighbours (KNN) algorithms were employed to collect and analyze motion data from the tests. The 53 subjects were categorized as either healthy or sarcopeniaprone, with the sarcopenia-prone group further classified into three levels based on their condition severity. The highest classification accuracy achieved was 94.64% for distinguishing between healthy and sarcopenia-prone subjects, and 90.44% for differentiating various sarcopenia-prone risk levels. This AI motion sensor system demonstrates potential as a cost-effective and accessible approach for large-scale sarcopenia screening. Further refinement of this method could enable remote health monitoring and telerehabilitation programs catering to older adults.