Enregistré dans:
| Auteur principal: | |
|---|---|
| Format: | Recurso digital |
| Langue: | anglais |
| Publié: |
Zenodo
2026
|
| Accès en ligne: | https://doi.org/10.5281/zenodo.19261587 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
Table des matières:
- <p>Sovereignty in multi-stakeholder AI systems is not a property of execution integrity or policy compliance. It is a structural condition defined by two orthogonal but inseparable halves: governance geometry, which identifies the surfaces that hardware and execution boundaries cannot govern, and admissibility conditions, which define the invariants that must hold before any execution is permitted. This paper unifies these two halves into a theoretical framework for structural sovereignty and proposes them as the foundation of a governance discipline distinct from behavioral alignment, execution integrity, and policy audit. The framework is model-agnostic, vendor-agnostic, and temporal.</p>