Land record digitization is a critical component of modern e-governance systems, enabling transparent land administration and efficient property management. Large-scale cadastral digitization initiatives often face data quality challenges such as boundary overlaps, geometric inconsistencies, and ownership mismatches due to legacy data migration, manual surveying errors, and heterogeneous data sources. Traditional rule-based validation techniques are limited in scalability and effectiveness for nationwide land digitization programs. This paper proposes an intelligent machine learning–based framework for automated anomaly detection in digitized land records, with a case study of the land record digitization platform. The proposed framework integrates spatial, topological, and attribute-based features extracted from cadastral data and applies unsupervised machine learning models to identify anomalous records without requiring labelled training data. Experimental evaluation on real-world digitized land records demonstrates that the proposed approach significantly improves anomaly detection accuracy and reduces manual verification effort compared to conventional validation methods. The study highlights the potential of intelligent data-driven techniques to enhance the reliability and scalability of large-scale land record digitization systems.