On Nov. 7, the U.S. District Court for the Southern District of New York dismissed a copyright-related lawsuit brought against an artificial intelligence (AI) company for a lack of standing.

Unlike other copyright litigation involving generative AI (Gen AI), the plaintiffs, which were news organizations, chose not to allege actual infringement of their copyrighted articles. Instead, the plaintiffs argued the Digital Millennium Copyright Act (DMCA) applied and claimed allegedly removing copyright management information (CMI) from the articles before they were allegedly used to train a Gen AI tool entitled the plaintiffs to damages and an injunction.

The plot twist did not convince the district court. The court rejected the plaintiffs’ attempt to use DMCA claims as a way to redress their true grievance: the alleged use of their articles to develop a Gen AI model without compensation.

The court found that neither of the plaintiffs’ arguments – 1) that “unlawful removal of CMI from a copyrighted work is a concrete injury” and 2) that there was allegedly a “substantial risk that Defendants’ program will ‘provide responses to users that incorporate[] material from Plaintiffs’ copyright-protected works or regurgitate[] copyright-protected works verbatim or nearly verbatim” – constituted the “concrete injury-in-fact sufficient to establish standing” to maintain their lawsuit.

In reaching its decision, the court first found the plaintiffs failed to establish standing to seek damages for alleged retrospective harms. Seizing on the plaintiffs’ choice to not allege infringement, i.e., that “a copy of their work from which CMI has been removed has been disseminated by [the Gen AI model] to anyone in response to any specific query,” the court rejected the plaintiffs’ argument that merely the CMI’s removal caused an injury sufficiently analogous to the tort of copyright infringement.

The court held the DMCA provision under which the plaintiffs sued “protects copyright owners from specified interferences with the integrity of a work’s CMI” – and does not, as the plaintiffs claimed, give the plaintiffs “the sole prerogative to decide how future iterations of the work may differ from the version the owners published” (emphasis original).

The court also found the plaintiffs “have not alleged any actual adverse effects stemming from this alleged DMCA violation,” i.e., “their copyrighted works (absent CMI) were used to train an AI-software program and remain in [the Gen AI model’s] repository of text.” Instead, the court held the plaintiffs’ argument was akin to one previously rejected by the U.S. Supreme Court.

The New York court also found the plaintiffs failed to establish standing for an injunction requiring the AI company to “remove all copies of Plaintiffs’ copyrighted works” from which CMI was allegedly removed from the Gen AI model’s “training sets and any other repositories.” The plaintiffs argued that, whether or not the Gen AI model had already reproduced their works without CMI, there was a “substantial risk” it would do so in the future.

The court disagreed, holding the likelihood the Gen AI model would “output plagiarized content from one of Plaintiffs’ articles” seemed “remote,” given “the quantity of information contained in the repository,” and that the Gen AI model “synthesizes the relevant information in its repository into an answer” to a user’s prompt.

Takeaways

This decision is notable because the plaintiffs’ theory was a twist on the traditional copyright-infringement claims being litigated against AI companies. But because the court found the plaintiffs lacked standing to assert the CMI removal theory, this approach to AI-related copyright litigation may not be viable in the future.