Model drift occurs when an AI model becomes less accurate over time because the real-world data it encounters changes from the data it was originally trained on. As conditions evolve, the model’s predictions may become unreliable unless the model is monitored and updated.