WME tells court AI training should be licensed, not stopped

William Morris Endeavor has urged a Manhattan federal court to reject OpenAI’s fair-use arguments unless creators retain control over whether their work is used for artificial-intelligence development. In an Oct. 30 amicus brief, the talent agency said AI use of creative material should be permitted, but licensed, authorised and compensated.

The filing supported the news plaintiffs’ opposition to the defendants’ summary-judgment motions on fair use in the consolidated In re: OpenAI, Inc. Copyright Infringement Litigation. WME argued that compensation for AI use of copyrighted work was already an economic reality, rather than a theory devised for the litigation.

Its central position was that creators should be able to decide whether, and on what terms, their protected work is used to develop or train AI systems. WME represents actors, screenwriters, directors, musicians, authors, athletes, digital creators and public figures in negotiations involving intellectual property; it also represents the AI animation studio Toonstar, whose business uses authorised content.

The agency said an existing, robust licensing market undermined the case for treating unlicensed training as fair use. It cited more than two dozen publicly announced licensing partnerships between rights holders and AI companies from mid-2023 to mid-2024, including at least 28 arrangements involving payment for training data.

Examples in the brief included News Corp’s reported five-year, $250 million agreement with OpenAI and Reddit’s $60 million agreement with Google. WME also pointed to reported Shutterstock agreements worth $25 million to $50 million each with Apple, Amazon and Google, while arguing that developers continued to seek licences for film and television libraries.

WME said unlicensed use could produce substitutes for its clients’ work, erode the licences creators could otherwise command and cost them both paid work and licensing revenue. It argued that licensing did not prevent earlier technologies such as radio and streaming from developing, and should not be treated as an obstacle to AI.

The filing followed the Oct. 29 joint submission by SAG-AFTRA, the American Federation of Musicians and the Writers Guilds, which opposed a categorical fair-use exemption for AI training, as we reported. The unions warned that such an approach could shift enormous value from creators to corporations and disrupt negotiated terms premised on licensing covered work.

The filing comes against sharply different positions in the litigation. The Justice Department argued in September that the United States had a strong interest in rejecting the view that training large language models on copyrighted texts violates copyright law. OpenAI has separately maintained that training on publicly available internet material is fair use. The consolidated case includes claims involving The New York Times, Florida and Arkansas newspapers, The Intercept, Ta-Nehisi Coates and other writers and outlets.