Scope as a Governance Primitive

This paper argues that Scope should be treated as a first-order governance primitive in AI governance. Rather than treating Scope as implicit context, metadata, or a synonym for authorization, the paper defines it as an explicitly identifiable governed region under a specific governance viewpoint and context. The paper synthesizes Scope-related constructs developed across prior work on Viewpoint-Structured Specification, Boundary, Behavior Rule Architecture, Decision Analysis, Decision Risk, Structural Degradation, and Continuation Readiness. It shows that these specialized constructs can be understood as application-specific projections of a more general Scope primitive. Three representative viewpoints are examined: Visible Scope — what an actor is exposed to by design Authorized Operable Scope — what an actor is authorized to operate on Actual Effect Scope — what is ultimately affected by execution The paper further distinguishes Scope from Boundary and Authorization, explains why relevant Scope states should be externalized and traceable across time, and shows how Scope can serve as a common reference for inference governance, authorization, effect analysis, Decision Analysis, Decision Risk, and governance evidence. A central argument is that the scope of an authorized operation does not determine the scope of its consequences, and that some governance failures—such as Inference Creep, Scope Absorption, and Cumulative Scope Expansion—can only be detected when Scope states are preserved and compared over time.