Anomaly monitoring detects irregularities or deviations in input data or model outputs, which are key indicators of model drift. According to the AAIA™ Study Guide, continuous anomaly detection is one of the most effective methods for identifying when a model is no longer functioning as expected due to changes in the data environment.

“Monitoring for output anomalies enables early identification of model drift. This proactive approach allows organizations to retrain or adjust models before significant performance degradation occurs.”

Configuration standardization (A) and documentation reviews (C) support governance but don’t detect changes. Retraining (D) is a remediation step, not a detection mechanism. Therefore, B is correct.