Model Drift (monitoring)

Model drift occurs when machine learning models in compliance or fraud systems lose accuracy over time due to changing data patterns. For example, fraudsters may adopt new tactics, making old models less effective.Continuous monitoring is required to detect drift and retrain models for accuracy.Regulators expect institutions using AI in compliance to maintain explainability and regularly validate model performance. Banks, fintechs, and regtech providers monitor for drift in transaction monitoring, fraud detection, and customer risk scoring models. By addressing drift proactively, institutions avoid higher false positives, missed risks, and compliance failures. Model governance frameworks help ensure AI systems remain reliable and regulator-approved.