A Theory of Vicarious Liability for Autonomous-Machine-Caused Harm

Abstract

The possibility of autonomous-machine-caused harm generates doctrinal and theoretical challenges for assigning tort liability. With emergent capabilities, autonomous machines disrupt the structure of interpersonal rights and duties in tort law, framed by conditions of foreseeability and proximate causation. Where algorithmic processes are unintelligible, self-modifying, and unpredictable, the concern goes, algorithmic harms will be untraceable to tortious human agency. As a result, their costs will simply lie where they fall—on faultless victims. This outcome would be unfair and objectionable: A failure of tort’s mechanisms of corrective justice means faultless victims would disproportionately bear the accident costs of autonomous machines. This article suggests that the doctrinal form of vicarious liability is a promising strategy to ground tort liability for autonomous-machine-caused harm. Human or corporate deployers should be held liable for tortious harm caused by autonomous machines in the course of deployment. In this account, autonomous machines constitute a novel legal category as pure legal agents without legal personhood. In reconceiving vicarious liability—and the legal classification of autonomous machines—the article seeks to promote commonsensical liability outcomes for autonomous-machine-caused harm, consistent with tort’s doctrinal and theoretical structure of corrective justice.