A runtime control placed around a model — separate from its trained weights — that checks requests or responses against a policy.
Continue to AI University →