CAA tells court AI licensing can work at scale

Creative Artists Agency and a coalition including Cate Blanchett, Steven Soderbergh and musician Nick Hexum have submitted a proposed amicus brief in the consolidated OpenAI copyright litigation, arguing that artificial intelligence developers can license creative work at scale rather than treating unlicensed training as fair use.

The 31-page filing, docket-stamped Oct. 30 in the U.S. District Court for the Southern District of New York, says “technical silence is not legal consent.” It urges the court to assess OpenAI’s fair-use defence against what the amici describe as an operational, accessible licensing market, rather than accepting that permissions would be prohibitively costly or impractical.

The named amici are CAA, the Human Consent Foundation, Music Artists Coalition, National Association of Voice Actors, Writer, Inc., Creators Coalition on Artificial Intelligence, Blanchett, Hexum and Soderbergh. Bass, Berry & Sims filed the proposed brief in In re: OpenAI, Inc., Copyright Infringement Litigation, the multidistrict case that includes The New York Times’ claims.

The filing says collective licensing for dispersed rights holders is already established practice, while machine-readable systems for signalling consent are feasible and in use. “The mechanisms to manage multi-rightsholder transaction costs are not only well-established,” the brief says, “but are being actively adapted for the AI era” by creative and technology-sector leaders.

CAA chief Bryan Lourd said the dispute concerned who determines the value and permitted use of creative work, and whether creators share in AI-created opportunities. He said the agency supported AI’s potential, but that its future should be built with artists and grounded in “consent, control, and compensation.”

Nikki Hexum, who co-founded the Human Consent Foundation with Blanchett, said the creative sector already had functioning licensing infrastructure, while stressing creators’ ability to decline participation. The foundation launched a registry in June allowing people to record permissions for AI models to scrape their likenesses; Blanchett said using her face, voice or image required prior permission, a contract and a conversation.

The intervention follows WME’s argument that AI use of creative material should be licensed rather than stopped, as we reported Oct. 30. The CAA coalition similarly does not frame its position as opposition to AI, but as a challenge to OpenAI’s assertion that using copyrighted works to train models is fair use, fair to creators and essential to innovation and US competitiveness.

The Copyright Office has identified consent, compensation and feasible licensing as central questions in generative-AI training, which draws on massive quantities of data that can include copyrighted works. Its report considered voluntary, collective and compulsory licensing models, as well as extended collective licensing and opt-out approaches.

The case has attracted an unusually broad set of outside submissions. Publishing Perspectives counted 51 amicus briefs, including 26 supporting plaintiffs, 18 supporting defendants and eight neutral filings. A final round of briefs is due Nov. 6, after which oral arguments are to be scheduled before Judge Sidney H. Stein.