A governance doctrine for AI systems based on explicit oversight. Externalizes trust and uncertainty into auditable, adversarial, and constrainable layers. A design framework, not an implementation guide.