Plaintiffs in the landmark Kadrey v. Meta case have sex between married couples videosalready submitted the U.S. Copyright Office's controversial AI report as evidence in their copyright infringement suit against the tech giant.

Last Friday, the Copyright Office quietly released a "pre-publication version" of its views on the use of copyrighted works to train generative AI models. The consequential report contained bad news for AI companies hoping to claim the fair use legal doctrine as a defense in court.

Less than a day after the report was published, Shira Perlmutter, the head of the Copyright Office, was fired by President Donald Trump. It's still unclear exactly why Perlmutter was fired, but the move alarmed some copyright lawyers, as Mashable previously reported.

And on May 12, the plaintiffs in Kadrey v Meta, which includes artists and authors such as Junot Diaz, Sarah Silverman, and Ta-Nehisi Coates, submitted the report as an exhibit in their class action lawsuit.

What's in the U.S. Copyright Office's AI report?

The Office's report was the conclusion of a three-part investigation into copyright law and artificial intelligence, which it calls uncharted legal territory. The "Copyright and Artificial Intelligence Part 3: Generative AI Training" report examined exactly the type of legal issues at stake in Kadrey v Meta.

While some copyright lawyers and Democratic politicians have speculated the report led to Perlmutter's firing, there are other possible explanations. In a blog post, copyright lawyer Aaron Moss said "it’s more likely that the Office raced to release the report before a wave of leadership changes could delay — or derail — its conclusions."

The report addressed in detail the four factors of the fair use doctrine. Meta and other AI companies are being sued for using copyrighted works to train their AI models, and Meta in particular has claimed this activity should be protected under fair use.

The lengthy 113-page report spends around 50 pages delving into the nuances of fair use, citing historic legal cases that ruled for and against fair use. It doesn't goes as far as making any blanket conclusions, but its analysis generally favors copyright holders over AI companies and their unprecedented stockpiling of data for model training.

The Copyright Office's stance on the white hot issue doesn't line up with the wishes of Big Tech titans, who have cozied up to the Trump Administration. In general, President Trump has taken a pro-tech approach to AI regulation.

The plaintiffs in the Kadrey v. Meta case are clearly hoping the report could tip the scale in their favor. The lawyers who submitted the report as evidence on Monday didn't explain in detail why it was submitted as a "Statement of Supplemental Authority." The brief simply said, "the Report addresses several key issues discussed in the parties’ respective motions regarding the use of copyrighted works in the development of generative AI systems and application of the fair use doctrine."

AI models can harm creative markets, Copyright Office finds

The part of the report that's potentially the most damning for Meta is the Copyright Office's assessment of the fourth factor of fair use, which considers the effects on current or future markets.

• Meta forges ahead with facial recognition for its AI glasses
• Mark Zuckerberg named in lawsuit over Meta’s use of pirated books for AI training
• Meta lobbies Apple and Google to take over age verification for young people online

"The use of pirated collections of copyrighted works to build a training library, or the distribution of such a library to the public, would harm the market for access to those works," said the pre-publication version of the report.

The analysis also considers possible market dilution for authors. "If thousands of AI-generated romance novels are put on the market, fewer of the human-authored romance novels that the AI was trained on are likely to be sold. Royalty pools can also be diluted," the report stated. In addition, the plaintiffs have argued that Meta's use of piracy to access the authors' books deprived them of licensing opportunities.

For its part, Meta argues that its AI model Llama doesn't compete with the authors' market, and that the model's transformative output makes the fair use argument irrelevant.

While the report favors the plaintiffs' argument, we don't know if the judge in the case will agree. And because this is a pre-publication version, it could be edited or even rescinded by a future leader at the Copyright Office.

Disclosure: Ziff Davis, Mashable’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.