TB
The Two Boundaries: Why Behavioral AI Governance Fails Structurally
arXiv (cs.PL) · 2026-03
Formal analysis of the structural gap in AI effect governance. Every effectful system has two boundaries (expressiveness and governance) that are almost never identical. Rice's theorem proves the gap is permanent for Turing-complete systems. Proposes coterminous governance as testable criterion.
CP
Certified Purity for Cognitive Workflow Executors: From Static Analysis to Cryptographic Attestation
arXiv (cs.PL/cs.CR) · 2026-02
Certified purity architecture converting governance enforcement from runtime convention to structural capability boundary. Four mechanisms: restricted WASM compilation, purity certificates, runtime verification gate, remote attestation.
CRP
Cryptographic Registry Provenance: Structural Defense Against Dependency Confusion in AI Package Ecosystems
arXiv (cs.CR/cs.SE) · 2026-02
Cryptographic distribution provenance system addressing dependency confusion attacks. Three components: registry identity (Ed25519), dual-signature model, authoritative namespace binding. Comparison across eight ecosystems.
MFSG
Mechanized Foundations of Structural Governance: Machine-Checked Proofs for Governed Intelligence
arXiv (cs.PL/cs.LO) · 2026-03
Five foundational results mechanized in Coq 8.19 with zero admitted lemmas. 9,942 lines across 30 modules, 349 theorems. Establishes Governed Cognitive Completeness as capstone. Extraction to OCaml NIF for production governance kernel.
GCC
Effect-Transparent Governance for AI Workflow Architectures: Semantic Preservation, Expressive Minimality, and Decidability Boundaries
arXiv (cs.AI/cs.PL/cs.LO) · 2026-03
Machine-checked formalization proving effect-level governance can be imposed without reducing computational expressivity. Establishes seven properties including governed Turing completeness and subsumption asymmetry.
AS
Algebraic Semantics of Governed Execution: Monoidal Categories, Effect Algebras, and Coterminous Boundaries
OOPSLA 2027 / arXiv (cs.PL/cs.LO) · 2026-03
Algebraic semantics for governed AI execution. Defines GovernanceAlgebra (three axioms) and shows it gives rise to a symmetric monoidal category with verified coherence. All mechanized in Coq with extraction to verified OCaml NIF running in production.
GH
Governed Metaprogramming for Intelligent Systems: Reclassifying Eval as a Governed Effect
arXiv (cs.PL) / OOPSLA 2027 · 2026-03
Introduces governed homoiconicity: code-as-data manipulation is pure computation, data-as-code materialization is a governed effect. Machine forms are first-class values subject to structural inspection before materialization. Preserves coterminous boundary property while enabling governed self-modification.