Model governance is the control of which models may be used, for what purposes and with what data — covering approval, version pinning, provider and residency constraints, evaluation, deprecation and the evidence that these controls were applied.
What is AI model governance?
Model choice is a risk decision as much as a performance one, because it determines where data is processed, under whose terms, and with what documented behaviour. Model governance maintains an approved catalogue with the conditions attached to each entry — permitted data sensitivity, permitted purposes, residency, retention terms — and enforces those conditions at the point of use rather than in a policy document. It also covers change: a silently upgraded model version can alter behaviour that was previously evaluated and signed off.