Research Article
A Layered Terminal-Goal Framework for AI Alignment: Quantilized Optimization, Irrevocable Constraints, and a Conformance Regime