Model Drift

The loss of accuracy in a deployed AI model as the conditions it works in change. The ICO defines concept or model drift as the case in which "the domain in which an AI system is used changes over time in unforeseen ways leading to the outputs becoming less statistically accurate". The Bank of England's 2022 discussion paper on AI and machine learning lists "model or concept drift" among model-related risks, and notes that systems able to learn continuously can be more susceptible to drift. The model itself can also change underneath a system: Kainos chief executive Brendan Mooney told BBC News NI in November 2026 that an AI system "runs on top of a model, that model drifts, either through changes to the model or changes to how the model interprets the data". For businesses, drift means an AI system needs monitoring after launch, not just testing before it.