Federal Judge Vince Chhabria ruled in favor of Meta over the authors who sued the company for training its large language model on their copyrighted works without obtaining consent. Chhabria ruled that Meta didn't violate copyright law after the plaintiffs failed to show sufficient evidence that its use of their work hurt them financially.
Does there have to be financial damages for copyright infringement to occur? Historically, no. That's where statutory damages have typically come into play.
Chhabria admitted that it is illegal to feed copyright-protected materials into LLMs without getting permission or paying the copyright owners for the work, but ultimately decided that Meta's training was considered fair use.
He wrote that by “training generative AI models with copyrighted works, companies are creating something that often will dramatically undermine the market for those works, and thus dramatically undermine the incentive for human beings to create things the old-fashioned way.” However, Llama isn't capable of generating enough text straight from books to violate fair use laws, and so the authors aren't entitled to the “market for licensing their works as AI training data.”
Chhabria noted that the authors should have focused on the argument that Meta copied their books to create a product that has the capability to flood the market iwth similar works, thereby causing market dilution, and that would've given them the win, but they barely touched the argument and presented no evidence of it.
While the authors may have lost the first argument, Chhabria confirmed that they would meet with Meta on July 11th to “discuss how to proceed on the plaintiffs' separate claim that Meta unlawfully distributed their protected works during the torrenting process.” This is where the statutory damages may come into play — because Meta did in fact torrent their books instead of buying them — which is illegal. (Or at least it has been in the past.)
The decision could set an early precedent in a growing number of copyright lawsuits involving generative AI and comes just a few days after a federal judge sided with Anthropic in a similar lawsuit. There are several other active lawsuits against AI companies for training their LLMs on copyrighted works including The New York Times suing OpenAI and Microsoft for training their models on its news articles, and Disney and Universal suing Midjourney for training its models on their films and TV shows.
The decisions also open the door to other questions including: What else is it fair use for AI companies to train their models with? My website content, including if I block their crawlers? My product and sales data? The lawsuits being decided now will have a lasting impact on just how much free reign access AI has to the web and digital content.