Following the recent lawsuit filed by the Recording Industry Association of America (RIAA) against music startups Udio and Suno, Suno issued a Filing with court on Thursday that it had indeed trained its AI model using copyrighted songs. But it claimed that doing so was legal under the fair use doctrine.
The RIAA filed suit against Udio and Suno on June 24, claiming the businesses trained their models with copyrighted music. While Suno's investors already indicated that the startup didn’t have permission from the music labels to make use of the copyrighted material was not expressed as directly as in today's filing.
“It is not any secret that among the many tens of tens of millions of images used to coach Suno's model, there have been likely images whose rights are owned by the plaintiffs on this case,” the filing states.
Suno CEO and co-founder Mikey Shulman continued in a blog entry was released the identical day because the lawsuit was filed and states: “We train our models on medium and top quality music that we will find on the open Internet… Much of the open Internet actually accommodates copyrighted material and a few of it’s owned by major record labels.”
Shulman also argued that training his AI model using data from the “open web” was no different from “a child writing his own rock songs after listening to the genre.”
“Learning isn’t an intervention in society. It never was and it still is,” Shulman added.
The RIAA responded with this response: “This is a significant admission of facts they tried to cover for months and only admitted when forced to achieve this by a lawsuit. Their industrial-scale copyright infringement doesn’t meet the necessities of 'fair use.' It isn’t fair in any respect to steal an artist's life's work, extract its core value and repackage it to compete directly with the originals… Their vision of the 'way forward for music' is seemingly one during which fans can not benefit from the music of their favorite artists because those artists can not make a living.”
The query of fair use has never been easy, but in the case of training AI models, even established doctrines may not apply. The end result of this early-stage case is more likely to set an influential precedent that would determine the long run of greater than just the 2 startups named in it.
