While advancements in deepfake detection have improved, criminals continue to exploit minute techniques to deceive content consumers. The sheer volume of internet data poses a significant challenge for real-time anomaly detection. This paper introduced a novel approach focused on identifying anomalies in lip movements, particularly imperceptible and ultra-thin alterations. Tested on public datasets, our method demonstrated strong capabilities in detecting subtle fakes that often evade human detection and basic forensic tools. Beyond standard anomalies, we created customized edits at the pixel and object levels to thoroughly assess our technique’s efficacy. By leveraging cutting-edge technologies such as SHA-256 hashing and RAID data processing systems, we have significantly enhanced the efficiency of anomaly detection within lip shape models. This approach can potentially be extended to other facial features like ears and noses with appropriate customizations based on specific object properties. Looking forward, our future research aims to explore gender-specific lip sync detection, considerations for skin melanin types, and integrating dental features for improved accuracy in lip area assessments.