Verifiable Autonomy: The HMAA Audit Architecture as a Foundation for AI Arms Control Verification

This working paper argues that the technical foundation for autonomous weapons verification already exists within the Human-Machine Authority Architecture (HMAA) governance framework's Secure Audit Log Module (SALM), and that adapting the IAEA safeguards model to autonomous-weapons governance through an International Autonomous Systems Governance Organization (IASGO) is technically feasible, institutionally precedented, and politically tractable within the November 2026 CCW Seventh Review Conference's mandate. It formalises the structural correspondence between IAEA safeguards and the SALM architecture across three verification dimensions; develops a zero-knowledge verification protocol on the Goldwasser-Micali-Rackoff foundational apparatus, showing that classified parameters can remain confidential while compliance with treaty-specified authority constraints is verified; and derives the IASGO institutional proposal across three verification scenarios: bilateral, multilateral, and unilateral. The Monte Carlo simulation code and results accompanying Section 4 are included in this record as version 1.1. Near-term feasibility is reported in qualitative bands.