Recommended Citation

Xuan-Thao Nguyen and Elizabeth G. Porter, The AI Input Class: Constitutional Urgency and Fair Licensing in AI Copyright Class Actions, 14 N.Y.U. J. Intell. Prop. & Ent. L. 1 (2025), https://digitalcommons.law.uw.edu/faculty-articles/1128

Authors

Xuan-Thao Nguyen, University of Washington School of LawFollow
Elizabeth G. Porter, University of Washington School of LawFollow

The AI Input Class: Constitutional Urgency and Fair Licensing in AI Copyright Class Actions

Publication Title

New York University Journal of Intellectual Property & Entertainment Law

Keywords

Copyright, Intellectual Property, Class Action, FRCP Rule 23, Fair Licensing

Document Type

Article

Abstract

The humanities have long been under attack; now Big Tech is eating them for breakfast. Artificial Intelligence is undermining the cultural and constitutional values of human creativity; it is also threatening the livelihoods of the creative working class. Because neither Congress nor regulators can keep pace with AI’s pace of change, class actions have stepped into the breach. In the first wave of such litigation, authors and artists in courts on both coasts claim that AI companies violate intellectual property law when they “train” their systems on copyrighted works.

This Article analyzes what we call “input” AI claims—that is, proposed class action copyright claims against AI companies for inputting creative works into their models. This action is variously described as copying, scraping, feeding, or training. We argue that input copyright class action claims have both merit and power. Substantively, the creators’ claims of direct copyright infringement are supported by almost three decades of caselaw applying the concept of “copying” to code and other innovative technologies. Procedurally, input claims based on AI “training” fall into the heartland of the letter and spirit of Rule 23, which governs federal class actions. Defendants’ claims of fair use, while seductive, would swallow copyright law and crush human-centered creativity.

Class action settlements—which combine private ordering with judicial oversight—may offer the best opportunity to establish a licensing regime that protects creators from future infringement. In lieu of fair use, we urge a different doctrine: “fair licensing.” Licensing poses its own risks to creativity; it risks creating moats around knowledge and paying creators symbolic amounts for giving up control of their works. But AI class actions can and should play an influential role in spurring the creation of fair AI licensing regimes. Courts and litigants are wise to act expeditiously in doing so: there is a constitutional urgency to protecting the essence of human creativity.