FOLLOW US

In a seismic ruling for the future of artificial intelligence and copyright law, a California federal judge on Wednesday declared that Meta Platforms Inc. lawfully used copyrighted material to train its large language model (LLM), LLaMA, rejecting a high-profile lawsuit filed by bestselling authors including Sarah Silverman, Ta-Nehisi Coates, and Junot Díaz.

The landmark fair use ruling delivered by U.S. District Judge Vince Chhabria represents a significant victory for the tech industry, reinforcing legal protections for AI developers that rely on copyrighted text as training material.

A Legal Win Echoing Through Silicon Valley

The decision came just 48 hours after another California judge found in favor of Anthropic PBC in a similar copyright dispute — signaling a growing judicial trend in support of AI training under fair use doctrine.

In his summary judgment order, Judge Chhabria stated bluntly that the plaintiffs failed to show any meaningful harm to the market for their books, a key factor in assessing whether an alleged infringement falls outside the bounds of fair use.

“Meta has defeated the plaintiffs' half-hearted argument that its copying causes or threatens significant market harm,” Judge Chhabria wrote.

No Market Harm, No Case

The authors had alleged that Meta trained LLaMA on their copyrighted works without permission, in some instances allegedly acquiring the material via pirated websites. But the court was unmoved, stating that the plaintiffs didn’t prove their book sales or licensing opportunities were negatively affected by LLaMA’s development.

The judge had previously hinted that Meta’s use might be deemed “transformative,” another major pillar supporting a fair use defense. However, his final ruling rested more squarely on the lack of demonstrable economic impact.

Legal Firepower on Both Sides

The case featured a who’s who of legal heavyweights, with the authors represented by powerhouse firms such as Boies Schiller Flexner LLP, Joseph Saveri Law Firm LLP, and Lieff Cabraser Heimann & Bernstein LLP. Meta, meanwhile, was backed by Cooley LLP, Cleary Gottlieb, and Paul Weiss, among others.

None of the parties involved issued an immediate comment following the ruling.

A Precedent-Setting Moment for Generative AI

This decision may embolden tech companies developing AI to continue mining copyrighted material for training purposes, as long as their use remains transformative and doesn’t directly undercut the source material’s market. The implications stretch far beyond LLaMA, affecting every developer training generative AI on vast corpora of internet-based text.

The ruling also adds to the growing body of case law in favor of fair use in the context of machine learning, a battleground that is rapidly becoming one of the most legally consequential arenas in tech.