This paper establishes a public reference formulation of Structural Sovereignty and Execution-Bound Governance as a federated architecture for consequence-bearing AI-supported systems. It defines the Governance-Execution Gap as the set of non-permitted transitions that nevertheless remain reachable inside a declared boundary, and introduces a boundary-relative Federated Permit whose mandatory predicates may include formal admissibility, control feasibility, externalized human or institutional authority, runtime mediation, physical enforcement where required, current actor/device/case/object context, version and freshness, fault-envelope validity, evidence readiness, and remedy. The contribution is not the isolated invention of supervisory control, reference-monitor concepts, runtime assurance, control barrier functions, provenance, human oversight, hardware interlocks, or regulatory risk management. The claimed contribution is their explicit cross-layer federation around a typed reachability invariant; the separation of proposal, permission, human commit, execution, consequence, and evidence; the derivation of authority from operational context external to model confidence; and the rule that composite maturity cannot exceed the weakest mandatory dependency. A bounded KWI-LIAN-Kyber route and a specified downstream CoChip architecture are presented as reference instantiations. Physical silicon validation, complete real-field closure, independent end-to-end reproduction, legal compliance determination, and certification remain outside the claims of this release. Protected implementation details are intentionally excluded.
Alexanja Senke (Mon,) studied this question.