The Justice Automated Operational Governance Standard (JAOGS™) 2026-003 establishes the formal governance architecture for execution-layer decision environments within AI-supported justice systems. As a core component of the JAOGS™ standard series, this document defines the structural model through which automated system outputs, human evaluation, and institutional authority interact to produce operational outcomes. While existing governance frameworks primarily focus on system performance, compliance, and policy design, JAOGS™ 2026-003 addresses a critical and previously undefined layer: the architecture of human decision-making that occurs after system signals are generated. This execution layer represents the point at which authority is exercised and where real-world outcomes are determined. The standard introduces a layered governance architecture that captures the full decision pathway, including signal generation, system processing, human evaluation, authority activation, and resultant action. It further establishes mechanisms for integrating decision pathway reconstruction, governance condition mapping, and structured analytical methods through alignment with Justice Decision Observability™ (JDO™). This version incorporates the Decision Reconstruction Performance Index™ (DRPI™) as a formal measurement layer, enabling evaluation of the completeness, consistency, and integrity of reconstructed decision pathways. Measurement is applied to governance documentation and reconstruction quality, not to system performance or decision correctness. JAOGS™ 2026-003 does not assign fault, determine compliance, or evaluate technology. Instead, it provides a neutral, structured architecture that enables institutions to document how decisions are made in practice, reconstruct governance events with precision, and strengthen accountability across complex operational environments. As part of the broader JAOGS™ framework, this document contributes to the establishment of a new governance discipline centered on the execution layer of human–AI interaction within justice systems.
Fleming et al. (Sun,) studied this question.