Judge allows authors' copyright lawsuit against Meta to move forward

Meta tried to throw out a massive copyright lawsuit from authors over its Llama models. A federal judge just said: not so fast.

U.S. District Judge Vince Chhabria ruled on Friday that the copyright infringement lawsuit filed by authors Richard Kadrey, Sarah Silverman, and Ta-Nehisi Coates can move forward, rejecting Meta’s push to dismiss the core of the case.

Hiding the paper trail

The authors claim Meta vacuumed up their copyrighted books without permission to train its Llama models. But the punchier accusation is that Meta deliberately stripped out copyright management information, known as CMI, before feeding the text to its models.

Judge Chhabria agreed that this argument holds water for now. In his ruling, he wrote that the allegations support a reasonable inference that Meta removed the data specifically to stop Llama from outputting copyright notices and revealing where its training data actually came from.

Meta did score one minor win. The judge tossed out claims brought under the California Comprehensive Computer Data Access and Fraud Act, pointing out that the authors never accused Meta of actually hacking their personal computers or servers, only scraping their books.

Zuck's green light in the spotlight

This lawsuit has already turned up embarrassing internal laundry for the tech giant.

Earlier court filings from the plaintiffs claim that Mark Zuckerberg personally gave the Llama team permission to train on copyrighted material. Internal messages cited in the case also show staff debating the legal risks of scraping questionable sources.

Meta has leaned heavily on the fair use defense, arguing that scraping text for machine learning is legally protected and that the authors lacked standing to sue. While the judge previously warned the authors' legal team about over-the-top rhetoric, he clearly found their central injury claim solid enough to survive.

This ruling lands amid a broader wave of legal pressure on big tech, joining high-profile battles like The New York Times taking on OpenAI.

Why it matters

We think this decision chips away at the tech industry's favorite legal shield. For the last two years, AI giants have treated fair use as an all-access pass to the open web, assuming plaintiffs could never prove concrete harm or bad faith.

By letting the CMI stripping claim proceed, the court is targeting the intention behind data cleaning pipelines. Stripping attribution metadata is standard operating procedure for many web scrapers, but framing it as an active attempt to conceal infringement puts Meta in serious legal jeopardy.

If courts eventually conclude that removing attribution to prevent AI leakages is intentional concealment, the bill for training frontier models will skyrocket. For creators, this is tangible leverage. Big AI labs may soon have no choice but to stop scraping first and apologizing later, forcing them to sign licensing deals before a single page touches a GPU cluster.