From the Rational Actor to the Rational Algorithm: Rethinking the Concept of Criminal Capacity for Autonomous Artificial Intelligence in the U.S. Criminal Justice System

Abstract

The traditional foundation of criminal law rests on the model of the “rational actor”; that is, a human being endowed with awareness, volition, and the capacity for moral reasoning and choice, whose attributes constitute the basis for the recognition of legal personality and criminal responsibility. With the emergence of advanced and autonomous artificial intelligence agents capable of independent, goal-directed decision-making, this classical model has encountered a fundamental challenge. Through a critical examination of the transition from the “rational actor” to the “rational algorithm,” this article argues that the traditional concept of legal capacity requires substantial reconsideration if it is to encompass non-human autonomous agents. On this basis, the article demonstrates that anthropocentric criteria of criminal responsibility, including the actus reus and mens rea elements of an offense, are no longer fully adequate to address emerging technological realities. The study subsequently proposes a functionalist framework for redefining legal capacity, according to which capacity is decoupled from biological consciousness and exclusively human characteristics and is instead assessed on the basis of the degree of autonomy, functional rationality, and the effects and consequences of an agent’s conduct. By analyzing the characteristics of advanced AI agents, including adaptive learning, operational independence, and the ability to undertake complex decisions and tasks, the article examines the possibility of establishing a new form of “electronic legal personhood” with graduated levels of responsibility. The article concludes that this conceptual transformation is not merely a theoretical hypothesis but an unavoidable necessity for addressing responsibility gaps, ensuring accountability, preserving social order, and guiding legal policymaking in a world increasingly shaped by algorithmic decision-making.