Emergent autonomy and the collapse of causal traceability: reconstituting functional liability for frontier model-based agents

Abstract

This article examines the structural tensions that the epistemic opacity inherent in frontier models deployed in AI agents imposes on the constitutional guarantees of reason-giving, traceability, and judicial oversight. The research begins with the premise that emergent autonomy and the technical infeasibility of reconstructing causal chains in the operation of these systems render the paradigm of meaningful human control structurally unviable, reducing it to a normative fiction incapable of securing effective accountability. From this diagnosis, the study adopts a critical-analytical and legal-dogmatic approach to address the legitimacy crisis triggered by generative artificial intelligence. The findings demonstrate that the absence of identifiable human agents causally linked to the effects produced by these models creates a structural void of legal imputation, incompatible with the normative demands of the Rule of Law. The article concludes that the establishment of an instrumental form of AI legal personhood constitutes a normatively coherent alternative to reconstitute functional centers of liability, ensure effective redress, and preserve the internal coherence of the legal system in contexts shaped by distributed agency.