Opt-Out Implied Licenses in Copyright Law: From Search Engines to GPAI Models

Abstract

Traditionally, the doctrine of implied license is consistent with the default of copyright law, because it requires evidence of a clear grant. Currently, US and German courts have implicitly applied an extended doctrine of implied licenses in the search engine context: opt-out implied licenses. This article calls for legislative recognition of opt-out implied licenses and identifies the unique mechanism and independent position in copyright law. Furthermore, two criteria, i.e. efficiency and balance, are proposed to assess the applicability of the doctrine. The proposed criteria can guide legislators in crafting legal norms that the opt-out implied license applies in specific contexts. On this basis, this article makes the case that opt-out implied licenses have some implications for training general-purpose AI models (hereinafter ‘GPAI models’) using copyrighted materials and that the proposed criteria can help better design and improve the law.