Private Law Theory

ABSTRACT
This Article examines broad arguments of infringing copyright’s entitlement of the right of derivatives in the context of Generative AI (‘GenAI’) systems. Copyright owners make derivatives arguments against various activities in the GenAI supply chain even in the absence of substantially similar output. They make these arguments in an attempt to go around the limitations of the right of reproduction and establish liability even when reproduction arguments fail. The Article’s argument is threefold. First, broad derivatives arguments fly in the face of copyright’s basic subject matter and scope principles. This is because underlying such arguments is the least plausible and most normatively precarious conception of the right of derivatives. Second, courts can use two existing elements of the case law to reject these implausible claims: the requirement that to be a derivative, an infringing use must incorporate expression from the copyrighted work; and a firm application of the substantial similarity test of infringement. Third, the abuse of derivatives claims in the GenAI context exposes the more general difficulties associated with this right: a feeble normative basis, an ever-extending scope, and an obscure relation to the right of reproduction. These serious shortcomings should be remedied by a general, coherent conception of the derivatives right as a right of adaptations: the right of creating versions of the same work in its entirety, in a different expressive medium. Such a construal would allow a proper and normatively grounded application of the right both generally and in the specific context of GenAI.

Oren Bracha, Generating Derivatives: AI and Copyright’s Most Troublesome Right, 25 North Carolina Journal of Law & Technology 345 (2024).