Anna Su (U Toronto Law) and Sue Anne Teo (Raoul Wallenberg Institute of Human Rights and Humanitarian Law; Harvard University – Carr-Ryan Center for Human Rights) have posted “Can AI Agents have Rights?” on SSRN. Here is the abstract:
AI agents are rapidly moving from theoretical constructs to real-world deployments. This raises urgent questions about governance and, more specifically, rights. While rights for artificial agents have received some scholarly attention, the question of rights for AI agents understood through the paradigmatic shift introduced by generative AI remains underexamined. This article addresses that gap. We develop four theoretical pathways through which rights for AI agents might be recognized. The derivative argument draws on the legal and moral foundations of agency law and tests their applicability to AI agents. The diffusion argument holds that AI agents’ deep embeddedness in social life creates pressure to render their actions legible within existing frameworks of responsibility and liability. The distinction argument examines whether AI agents possess capacities-including a potential role in resolving collective action problems requiring high levels of social coordination-that independently justify rights recognition. The devolution argument frames rights as a counterweight to the concentration of corporate power over AI systems. A central contribution of this analysis is decoupling the question of AI rights from moral personhood and its associated qualities, such as sentience and consciousness. We also address three objections: that AI rights would dilute human rights, generate a problematic proliferation of rights, and that regulatory goals could be achieved through legal duties alone. As AI agents become increasingly embedded in social, professional, and political life, questions about their rights will inevitably arise. This article offers a more nuanced framework for addressing them.