The American startup Suno, which developed an AI-based music generator, admitted that its product was trained on “all music files of reasonable quality accessible on the open Internet,” amounting to “tens of millions of recordings.” Founded in 2022, Suno has since raised $125 million and responded to a lawsuit filed by the three largest music production companies in the world by claiming it is “fighting the monopolies” of the record companies.
“It is no secret that the tens of millions of recordings on which Suno’s generator was trained likely included recordings whose rights are owned by the plaintiffs in this case,” Suno wrote in a court filing. Regarding UMG, one of the plaintiffs and one of the largest companies in the world, Suno pointed out that “the enormous size of UMG’s catalog was created through decades of mergers and acquisitions,” positioning itself as a modern-day Robin Hood while accusing the major labels of “trying to leverage their exclusive rights under copyright law to prevent music users from producing music using artificial intelligence tools.” In May, Suno revealed that its music generator had been used 12 million times since its launch in December 2023.
In addition to UMG, Warner Records and Sony Music are also suing Suno, with the assistance of the Recording Industry Association of America (RIAA). Another startup, Udio, which operates similarly by allowing users to create songs from text in various styles, is also being sued.
These companies’ algorithms were built by training on datasets of all the recorded music available on the Internet. Initially, the companies did not disclose how their models were trained or what data was used, but as their tools became more widespread, similarities between the generators’ outputs and original works became apparent. Udio, for example, is accused of producing outputs that sound like Bruce Springsteen and Michael Jackson. Later, the startups admitted to using the catalogs of the music giants. In March, an investor in Suno, Antonio Rodriguez of Matrix Partners, told Rolling Stone, “Honestly, if we had deals with the labels when the company started, I probably wouldn’t have invested in it. I think they should have produced this product without constraints.”
The three music companies claim that the startups engaged in “intentional copyright infringement on an almost unimaginable scale,” demanding that they admit their models were trained on their catalogs, that the court impose injunctions on this activity, and that they be awarded damages of up to $150,000 for each song used to train the model. The three music companies own the copyrights of almost every major artist—living or dead.
Suno and Udio’s response to the court was an admission of guilt, which immediately turned into an attack. Citing the “fair use” exclusion in the law, which many AI companies use to defend against lawsuits, the startups argued that it is permissible to make copies of songs for “technological processes” like training an AI model, as long as consumers do not interact with the actual song copies. They then described their work as musical emancipation. “The outputs created by Suno are new sounds inspired by the ‘styles, arrangements, and sounds’ of their predecessors,” Suno claimed. A post published by the startup last Thursday reiterated their court claims, asserting that the big companies’ fight stems from the threat AI poses “to their business.”
Udio echoed this sentiment, stating, “Helping people create new artistic expressions is what copyright law is meant to encourage, not prohibit.” Udio argued that using existing sound recordings for data mining and analysis to identify patterns in musical styles, enabling people to create their own new compositions, constitutes fair use under copyright law.
Suno and Udio are not the first to build technology that disrupts the music industry. Last June marked the 25th anniversary of the launch of the first file-sharing service—Napster. Launched before the dot-com bubble burst, the company created the world’s largest free music service by allowing millions of users to search for and download MP3 files on their personal computers. A few months after its launch, the first copyright infringement lawsuit was filed against it by the RIAA, demanding $100,000 for each copyrighted song shared on the network, totaling about $100 million. The band Metallica also got involved after discovering that hundreds of thousands of Napster users had downloaded their songs before they were even available for purchase. “It was never about money and greed,” the band’s drummer Lars Ulrich said at the time. “I have no problem giving [music], but I want to decide what to give.”
The public still loved Napster, driven by anti-establishment sentiment and a perception of the industry as exploitative of artists, even though they missed the fact that it also stole from the musicians. Napster eventually paid the production companies $26 million in settlements before declaring bankruptcy in 2002. But neither the Internet nor the music industry was the same after that. “Napster was a cultural revolution,” company co-founder Sean Parker told The New Yorker in 2014, “much more so than any legitimate company ever was.”
Suno and Udio may be laying the groundwork for the idea that anyone can make synthetic music without any skills, but they highlight the same original sin as Napster: talented artists are left behind. After Napster was shut down, alternative pirated services quickly emerged, but legal alternatives like Apple Music and Spotify also arose, inspired by Napster’s foundation. These companies also disrupted the music industry, but were less destructive to the creators.
“There is nothing fair about stealing an artist’s life’s work, stripping away its core value, and repackaging it to directly compete with the original,” the RIAA said in a press release. “Their vision of the ‘future of music’ is likely one where fans will no longer enjoy music by their favorite artists because those artists can no longer make a living.”
