"Model drift" refers to a situation where a machine learning model becomes less accurate or effective over time because the data it is trained on changes or evolves. It means that the model no longer accurately represents the current patterns or relationships in the data it was designed to predict.
Full definition