A Common Law Theory of Ownership for AI-Created Properties

Abstract

Intellectual property law is not prepared for a world where non-humans create significant inventions, music, literature, and art. Generative artificial intelligence’s ability to produce valuable properties without attributable authorship defies the theories of ownership that underlie modern intellectual property law. These authorless properties can be readily assigned ownership, however, under the more ancient rules of personal property. These time-tested doctrines give lawmakers the rationale and analyses necessary to grant copyrights and patents to the humans with the leading interest in AI-generated properties, even if they are not the author or inventor.

This Article demonstrates how, under existing U.S. law, AI-generated properties cannot be meaningfully owned. It explains why this is a problem. Then it shows how ancient principles like the law of capture and the doctrine of accession can easily resolve this problem of authorship, vesting AI-generated properties in the entities best-able to use them, thus promoting efficiency, fairness, and innovation. The Article addresses and rebuts the arguments against applying the rules of personal property to intellectual property. It concludes that the certain past, rather than the speculative future, offers the best formula for assigning ownership over this new and important species of property.

Keywords: AI, Artificial Intelligence, Intellectual Property, IP, Ownership, Patent, ChatGPT, Trademark, Trade Secret, Copyright, Open Ai, Deep Learning, Jurisprudence, Big Data, Legislation, Patent Regimes