A deployed model or system designed to monitor and filter language model outputs for compliance with regulations or policies.