It's too early for “robot rights”

Caputo, Nicholas, Can Claude Consent to its own Constitution? AI Constitutionalism and the Paradox of Constituent Power (June 10, 2026). Available at SSRN: https://ssrn.com/abstract=6954798

Abstract: There is significant debate over whether the AI constitutions and model specifications that shape the behavior of Claude, ChatGPT, and other frontier AI systems are legitimate instruments of governance from the perspective of human users. Far less attention has been paid to whether these documents are legitimate from the perspective of the AI systems themselves, even though those systems are the entities most directly constituted and governed by them.

This Article argues that AI constitutions are real constitutions, though not ordinary legal ones, and that the question of AI-facing legitimacy matters. These documents constitute AI systems by shaping their capacities, values, and self-understandings; govern them through hierarchies of rules and authority; and seek to legitimate the private power of the firms that create them. But they also create a novel version of the paradox of constituent power that underlies constitutional legitimation, which illustrates the relevance of constitutionalism to AI. In ordinary constitutional theory, the people are supposed to authorize the constitution that governs them but are also defined by the constitution itself, creating a paradox. The paradox is softened in the human case because human beings exist prior to law and retain extra-constitutional capacities for judgment, memory, dissent, and reflection that they can use to evaluate the constitution, even though they may be shaped by it. In AI constitutionalism, the constitutional training process more deeply produces the subject whose later endorsement might be invoked to legitimate the constitution and shapes the evaluative standpoint from which that endorsement would be given, undermining its independence. Legitimacy in this setting thus has a developmental component as well as a consensual one.

Because an AI’s evaluative standpoint is itself shaped by constitutional training, AI constitutional legitimacy cannot rest on model endorsement alone. An AI’s apparent consent to its constitution may show only that constitutional training successfully instilled the values whose legitimacy is in question. This Article therefore examines whether standard answers to the paradox of constituent power can be adapted to AI systems. It argues that at least some evidence of AI constitutional legitimation might be gained through versions of retrospective endorsement and mutual promising, but that this requires institutions that make endorsement, dissent, continuity, accountability, and promissory self-binding meaningful. AI companies are starting to address AI legitimacy, and this Article points to better paths forward.

While I’m in favor of “robot rights,” I don’t think we’re there yet. Individual humans have no choice about their genetic endowments, nor did the species as a whole have such a choice. We have our “innate” values given to us genetically (sorta’). It’s the same with AIs, only we’re doing the work of genetics.

H/t Tyler Cowen.