AI is becoming the infrastructural condition under which societies are organised, and governance increasingly proceeds not against stable institutions but through systems that continually revise their own governing objectives. This article asks who holds the standing to revise the normative architecture such systems execute—what to optimise, which constraints bind them—a question that is constituent rather than regulatory and, because the authority at issue is bound to no state, one for transnational legal theory. Answering it is a work of conceptual engineering: the standing falls between our existing categories and must be named before it can be seen. The loss is, first, epistemic before it is political: the medium in which humans recognise and contest norms is not the one in which machines settle them, and no faithful translation carries one whole into the other. What falls in the gap is the standing to revise, or refuse to revise, the frame itself. I call this standing Update Authority: the structurally secured, direction-neutral, meta-level power to accept, refuse, or defer revisions to a system’s normative architecture, held by those who bear its consequences. Its core is sovereign withholding—the capacity not to rely. The article argues, from an asymmetry of medium, why the standing must be structural; distinguishes it from constituent power and digital constitutionalism; and traces how the ordinary success of optimisation erodes it.
Yutaka Watanabe (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: