Following the Recording Industry Association of America's (RIAA) recent lawsuit against music generation startups Udio and Suno, Suno admitted in a court filing on Thursday that the company does use copyrighted songs to train its artificial intelligence models. But it claims that doing so is legal under the fair use doctrine.

The Recording Industry Association of America filed a lawsuit against Udio and Suno on June 24, accusing the two companies of using copyrighted music to train models. While Suno's investors have previously hinted that the startup doesn't have permission from music companies to use copyrighted material, they haven't said so outright as in today's filing.

"It is no secret that the tens of millions of recordings on which Suno's models were trained may include recordings to which the plaintiffs in this case have copyrights," the filing states.

Suno CEO and co-founder Mikey Shulman published a blog post on the same day as the legal filing, which continued: "We train our model on medium to high-quality music that can be found on the open internet...Much content on the open internet does contain copyrighted material, some of which is owned by major record labels."

Schulman also believes that using data on the "open Internet" to train artificial intelligence models is no different from "children composing their own rock songs after listening to rock music."

"Learning is not a violation. It wasn't then and it isn't now," Schulman added.

The RIAA responded: "This is a significant concession from facts they spent months trying to hide, only to be forced to admit by lawsuits. Their industrial-scale infringement does not qualify as 'fair use.' There is nothing fair about stealing an artist's life-long work, extracting their core value, and repackaging it in direct competition with the original work... Their vision for the 'future of music' is clearly one in which fans will no longer enjoy the music of their favorite artists because these artists can no longer make a living."

The issue of fair use is never simple, but when it comes to training AI models, even established theory may not apply. The case is still in its early stages, and its outcome is likely to set an influential precedent that could determine the future of more than just the two startups named in the case.