When a machine learning model learns the training data too well, including its noise and quirks, making it perform poorly on new data. Like a student who memorizes test answers but can't apply the concepts to new problems.
A fraud detection model that achieves 99% accuracy on training data but only 60% on real transactions is likely overfitting to patterns specific to the training set.
Overfitting is a machine learning modeling problem (not a cloud service). All major clouds provide ML platforms (e.g., managed training, AutoML, and MLOps) that include tools to detect and reduce overfitting via validation, regularization options, early stopping, and monitoring.