AI music startup Suno says copyrighted music training model is ‘fair use’

Following a recent lawsuit filed by the Recording Industry Association of America (RIAA) against music-generation startups Udio and Suno, Suno admitted in court on Thursday that it did indeed train its AI model using copyrighted works, but argued that doing so was consistent with the fair use doctrine.

The RIAA filed a lawsuit against Udio and Suno on June 24, alleging that the companies trained their models using copyrighted music. While Suno’s investors have previously suggested that the startup didn’t have permission from record labels to use copyrighted material, that wasn’t as directly stated as it is in today’s lawsuit.

“It is no secret that the tens of millions of recordings on which the Suno model was trained likely included recordings owned by Plaintiffs in this case,” the filing reads.

Suno CEO and co-founder Mikey Shulman continued in a blog post published the same day as the legal filing, saying, “We train our models on mid- to high-quality music that we can find on the open internet… Much of the open internet does contain copyrighted material, some of which is owned by major record labels.”

Shulman also argued that training his AI model with data from the “open internet” is no different than “a kid writing his own rock songs after hearing the genre.”

“Science is not a violation. It never was, and now it is,” Shulman added.

The RIAA responded with this: “This is a serious admission of facts they spent months trying to hide and only admitted when forced to do so by a lawsuit. Their industrial-scale infringement does not qualify as ‘fair use.’ There is nothing fair about stealing an artist’s life’s work, extracting its essential value, and repackaging it to compete directly with the originals… Their vision for the ‘future of music’ is clearly that fans will no longer enjoy the music of their favorite artists because those artists will no longer be able to earn a living.”

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