Autonomous vehicles (AVs) rely not only on complex high-level decision-making systems but also on the safety and reliability of low-level control systems, including actuators and electronic control units (ECUs). This paper introduces a novel framework for validating these low-level systems using digital twin-based simulations, integrating an AI-driven monitoring system to observe and analyze control signals from the higher-level autonomous software. The AI system detects real-time anomalies, such as localization loss or erratic throttle commands, and automatically triggers corrective actions like reducing speed or initiating emergency stops to prevent accidents. By leveraging digital twins to simulate realistic failure conditions, the framework enhances both low-level and high-level validation, contributing to a more robust and reliable AV system. This approach demonstrates a significant advancement in the validation and verification (V&V) process, providing a powerful tool for improving safety in autonomous driving technologies through comprehensive testing and proactive failure management.