Abstract: Safety monitoring is critical for ensuring the long-term performance of earth-rock dams. This paper systematically investigates three key components: data processing and management techniques, early-warning prediction, and safety evaluation frameworks, while reviewing global research progress and critical technical bottlenecks. For data processing and management techniques, the current status of gross error detection and missing data imputation is analyzed, with comparisons of the advantages and limitations of statistical methods, machine learning, and deep learning algorithms. For early warning and forecasting, the evolution from single models to hybrid models is summarized, and the role of dynamic weight allocation strategies in improving prediction accuracy is discussed. Regarding safety evaluation, systematic deficiencies and insufficient standardization are identified, particularly in monitoring data interpretation and multi-method integration. Significant progress has been made in research on dam safety monitoring technologies concerning data governance, prediction and early-warning systems, and safety evaluation frameworks for earth-rock dams. However, critical bottlenecks remain that require urgent resolution. (1) Insufficient data governance efficiency and generalization capability: Traditional statistical methods exhibit limited adaptability to complex, non-stationary data. While machine learning and deep learning techniques have enhanced the accuracy of outlier detection and missing data imputation, they face challenges such as high computational complexity and strong parameter dependency. Future efforts should integrate digital twin technology to construct dynamic data governance platforms, enabling full-process automation. (2) Weak coupling between early-warning models and mechanistic models: Hybrid models have improved prediction robustness through dynamic weighting strategies. Nevertheless, early-warning indicators predominantly rely on statistical thresholds and show weak correlation with the structural-mechanical characteristics of the dam body. There is an urgent need to integrate data-driven approaches, mechanistic analysis, and knowledge reasoning to enhance the completeness of multi-driver early-warning theories for dam safety behavior. (3) Inadequate standardization and dynamic fusion in safety evaluation systems: Existing methods often focus on static, single-dimensional analyses. Frameworks for dynamically fusing multi-source data are constrained by data quality limitations and deficiencies in monitoring systems. It is recommended to strengthen capabilities for multi-physics coupling analysis and to promote cross-physical-field data collaborative verification and mutual validation of results, thereby improving the comprehensiveness and accuracy of evaluation outcomes. Future studies on embankment dam safety monitoring in future should enhance interdisciplinary integration, fully utilize the potential applications of artificial intelligence and digital twin technologies, and advance intelligent and precise solutions through systematic technological innovation and standardized frameworks.