ARC Authority Governance Framework: Interrupting Machine Authority at Execution Time

Artificial intelligence systems are increasingly embedded within digital infrastructures that govern finance, healthcare, public services, law enforcement, employment, education, and information ecosystems. While much of the current discourse on AI governance focuses on fairness, transparency, and algorithmic bias, a fundamental structural question remains underexplored: where automated systems execute authority over human outcomes. This paper introduces the ARC Authority Governance Framework, where ARC stands for Architectural Risk Control. The ARC framework focuses on identifying the architectural points within digital infrastructures where automated systems exercise real-world authority, such as approving or denying services, restricting access to resources, flagging individuals for enforcement actions, or triggering institutional responses. The framework includes a safeguard architecture known as ARC-S (Architectural Risk Control System). ARC-S introduces interrupt mechanisms designed to pause, contain, or reverse automated authority execution before harmful decisions propagate across interconnected systems. These mechanisms may include human override channels, execution delays, containment barriers, and system rollback capabilities. The ARC Authority Governance Framework consists of three primary components: • Authority Atlas – a structural map of sectors and decision nodes where automated systems exercise authority across society.• ARC Records – documented case analyses of real-world authority execution events within automated systems.• ARC-S – an execution interrupt architecture designed to preserve human oversight and prevent uncontrolled propagation of automated authority decisions. Together, these components establish a governance architecture that shifts AI governance from evaluating algorithmic outputs to governing the structural execution of machine authority within socio-technical systems. The ARC framework is operationalized through five governance principles: Authority Mapping Execution Transparency Interruptibility (ARC-S) Containment Accountability By focusing on where machines exercise authority and how that authority can be interrupted, the ARC framework provides a structural approach to governing automated decision infrastructures. In an era where automated systems increasingly influence access to resources, services, and opportunities, ensuring that machine authority remains transparent, interruptible, and accountable becomes a central challenge of responsible AI governance. The ARC Authority Governance Framework contributes a governance architecture designed to address that challenge.