Not Quite Human: Artificial Intelligence, Animals, and the Regulation of Sentient Property

Abstract

The advent of artificial intelligence promises to revolutionize nearly every aspect of civil society. AIs will perform complex tasks, occupy jobs traditionally performed by humans, develop groundbreaking inventions, provide familial comfort, and even offer sexual companionship. In part because the versatility of AI makes it suited to perform so many disparate roles, policymakers will need to confront the question of legal status: what are AI in the eyes of the law? Are they mere property? Are they “people” too? Are they something in between? This Article takes a first step in answering that question by analogizing AI to another form of sentient property: nonhuman animals. Four centuries of domestic animal rights law offer critical insights into the legal values that AI regulations ought to protect. These include protection of AIs as property (including allocation of responsibility for damage done by AIs); protecting human emotional attachments to companion and romantic AIs; limiting extreme antisocial behavior involving AIs—such as the realistic torture of physically embodied AIs—because of the psychologically harmful consequences it has for humans; and potentially protecting the AIs as intrinsically valuable. These values can be protected by a number of regulatory regimes that have analogously emerged in reaction to animal rights concerns, including anticruelty legislation, private ordering for leased AIs, federal administrative oversight, public nuisance litigation, and research or grant conditions. The practical lessons of animal rights law avoid unhelpfully abstract theorizing and offer a realistic model for regulation of artificial intelligence in the twenty-first century.