DMLA Joins Creator Organizations in Amicus Brief Addressing AI Training and Market Harm

The Digital Media Leadership Alliance has joined the Authors Guild and a coalition of creator organizations in an amicus brief filed in In re Mosaic LLM Litigation, a closely watched copyright case involving the use of copyrighted works to train generative AI systems.

Filed August 27, 2026 in the U.S. District Court for the Northern District of California, the brief opposes the defendants’ motion for summary judgment and argues that the unauthorized reproduction of copyrighted literary works to train the defendants’ StoryWriter platform does not qualify as fair use.

At the center of the brief is a growing question for creators, licensors and technology companies alike: When generative AI systems are trained on copyrighted works and then used to create competing content, how should courts evaluate the resulting market harm?

AI Training and Transformative Use

The brief argues that StoryWriter’s use of copyrighted literary works was not transformative because the system was designed to generate the same general category of expression as the works on which it was trained: stories intended to be read by people.

Relying on the Supreme Court’s 2023 decision in Andy Warhol Foundation v. Goldsmith, the amici argue that where an original work and a secondary use serve the same or highly similar commercial purposes, the secondary use may be viewed as superseding the original rather than transforming it.

The brief also distinguishes AI training on literary works from earlier “intermediate copying” cases involving software. Those cases often permitted copying where it was necessary to analyze functional elements of computer programs. The amici argue that literary works present a fundamentally different situation because reproducing their protected expression is not necessary to uncover underlying functional code or interoperability information.

Licensing Markets Matter

A second major argument concerns the growing market for AI-training licenses.

The brief notes that authors, publishers and collective licensing organizations are already making copyrighted works available for AI training under negotiated licenses. As a result, the amici argue that unauthorized use can deprive copyright owners of an existing or potential licensing market.

Importantly, the brief also maintains that copyright owners should retain the right to decide whether they wish to license their works for AI training at all. The absence of a license from a particular creator should not mean that the creator has surrendered control over that market.

The Emerging Issue of Market Dilution

Perhaps the most consequential argument in the brief concerns market dilution.

Traditionally, much of the debate around generative AI and copyright has focused on whether an AI system reproduces or “regurgitates” a protected work. The brief argues that market harm can occur even when an AI-generated output is not substantially similar to any single work in the training data.

Generative AI systems can produce enormous quantities of low-cost content that compete with human-created works for sales, attention and discoverability.

The brief relies in part on the reasoning in Kadrey v. Meta Platforms, where the court recognized that AI-generated works could indirectly substitute for human-created works by competing in the same markets.

Simply put, if a consumer purchases an AI-generated work instead of a human-created one, the absence of direct copying in the final output does not necessarily mean there has been no economic harm.

The brief also cites recent research examining the self-published book market between 2023 and 2026. According to the study cited by the amici, the volume of available titles expanded far more rapidly than overall sales, while AI-generated books captured an increasing share of the market. The researchers concluded that the resulting expansion in supply contributed to declining revenue per title and greater competition for visibility.

Broader Implications for Creative Markets

Although the Mosaic litigation concerns literary works, the issues raised by the brief extend across creative industries.

Photography, illustration, video, music and other forms of media increasingly exist in markets where generative systems can produce large volumes of competing content at extremely low marginal cost.

For the visual media community, the underlying question is significant:

Must an AI-generated image closely reproduce a particular photograph or illustration before the economic impact on creators becomes relevant, or can large-scale substitution and market dilution themselves constitute meaningful market harm?

That question is likely to become increasingly important as courts continue to develop the law surrounding generative AI and fair use.

DMLA’s Interest

DMLA joined the brief alongside the Authors Guild, National Press Photographers Association, American Society of Media Photographers, Romance Writers of America, Association of American Literary Agents, American Society for Collective Rights Licensing, Society of Composers & Lyricists, Dramatists Guild of America, Graphic Artists Guild and American Photographic Artists.

The brief describes DMLA as representing 48 companies operating throughout the visual media ecosystem, including major licensors and distributors as well as technology companies developing tools for rights management, provenance, authentication and responsible artificial intelligence.

Collectively, DMLA member companies represent, license, distribute or provide market access for the works of millions of photographers, illustrators, videographers and other creators worldwide.

As AI technology continues to reshape creative markets, DMLA will continue to engage in discussions around copyright, licensing, provenance, responsible AI development and the economic sustainability of human-created content.

DMLA members may access legal documents including Amicus Briefs in the Member Documents Library.