Country musician Tift Merritt’s most popular song on Spotify, “Traveling Alone,” is a song whose lyrics are reminiscent of solitude and the open road. When prompted by Reuters to create “an American song in the style of Tift Merritt,” artificial intelligence music website Udio quickly produced a song called “Holy Grounds,” whose lyrics are about “driving down old backroads” while “watching the fields and skies move.”
Merritt, a Grammy-nominated singer and songwriter, told Reuters that the “imitations” created by Udio “are not appropriate for any of my albums.” “It’s a great example of how this technology is not transformative,” Merritt said.
“This is theft.” Merritt, a longtime advocate for artists’ rights, isn’t the only musician sounding the alarm. In April, she joined Billie Eilish, Nicki Minaj, Stevie Wonder and dozens of other artists in penning an open letter warning that AI-generated music trained on their recordings could “harm creativity” and sideline human artists.
Big record labels are also concerned. Sony Music, Universal Music Group and Warner Music sued Udio and another music AI company Suno in June, marking the music industry’s entry into a high-stakes copyright battle over AI-generated content that’s just beginning to make its way through the courts. “It’s not creative to use a huge amount of creative labor to copy it,” said Merritt, an independent musician whose first record label is now owned by UMG but who said he’s not financially involved with the company. “It’s stealing to compete and replace us.”
When asked for comment for this story, Suno and Udio pointed to past public statements defending their technology. They filed their initial responses in court on Thursday, denying any copyright infringement and arguing that the lawsuits were attempts to stifle smaller competitors. They compared the labels’ opposition to the industry’s past concerns about synthesizers, drum machines and other innovations replacing human musicians.
Uncharted territory
Both companies, which have attracted venture capital funding, have said they prevent users from creating songs that blatantly mimic top artists. But the new lawsuits say Listen and Udio can be induced to reproduce elements of songs by Mariah Carey, James Brown and others, and to mimic the voices of artists such as ABBA and Bruce Springsteen, suggesting they misused the labels’ catalogs of copyrighted recordings to train their systems.
Mitch Glazier, CEO of the music industry trade group Recording Industry Association of America (RIAA), said the lawsuits “document the shameless copying of a plethora of recordings to flood the marketplace with cheap imitations and siphon off listening and revenue from real human artists and songwriters.” “AI has great potential — but only if it’s built on a solid, responsible, licensed foundation,” Glazier said.
Warner Music referred Reuters to the RIAA when asked for comment on the matter. Sony and UMG did not respond.
The labels’ claims echo allegations leveled by novelists, news outlets, music publishers and others in high-profile copyright lawsuits over chatbots such as OpenAI’s ChatGPT and Anthropic’s Cloud that use generative AI to create text. Those lawsuits are still pending and in their early stages.
Both cases have raised new questions for courts, including whether the law should make exceptions for AI’s use of copyrighted material to create something new. The record labels’ cases, which can take years to settle, also raise unique questions related to their subject matter — music. The interplay of melody, harmony, rhythm and other elements can make it hard to determine whether parts of a copyrighted song have been infringed, said Brian McBrearty, a musicologist specializing in copyright analysis.
“There are so many more factors in music than just a stream of words,” McBrearty said. “It has melody, it has rhythm and it has harmonic context. It’s a rich mix of different elements that makes it a little less straightforward.” Some claims in AI copyright cases can rely on comparisons between the output of an AI system and the material allegedly misused to train it, requiring the kind of analysis that has challenged judges and juries in music cases. In a 2018 decision that a dissenting judge called “a dangerous precedent,” Robin Thicke and Pharrell Williams lost a case brought by Marvin Gaye’s estate over the similarity of their hit “Blurred Lines” to Gaye’s “Got to Give It Up.” But artists including Katy Perry and Ed Sheeran have since had similar complaints over their own songs dismissed.
Suno and Udio argued in very similar court filings that their outputs don’t infringe copyright and said U.S. copyright law protects sound recordings that “mimic or emulate” other recorded music. “Music copyright has always been a messy world,” said Julie Albert, an intellectual property partner at law firm Baker Botts in New York who is tracking the new cases. And even without that complexity, Albert said rapidly evolving AI technology is creating new uncertainty at every level of copyright law.
Appropriate use of what?
The musical intricacies may ultimately matter less if, as many expect, AI cases become limited to a “fair use” defense against infringement claims — another area of U.S. copyright law that is fraught with open questions. Fair use promotes freedom of expression by allowing unauthorized use of copyright-protected works under certain circumstances, with courts often focusing on whether the new use alters the original works. Defendants in AI copyright cases have argued that their products constitute fair use of human compositions, and any court ruling to the contrary would be potentially devastating to the multi-trillion dollar AI industry.
Suno and Oudio said in response to the labels’ lawsuits on Thursday that using existing recordings to help people create new songs is “a quintessential ‘fair use.'”
“Fair use can make or break cases,” legal experts said, but no court has yet ruled on the issue in the context of AI. Albert said AI companies that create music may have more difficulty proving fair use than chatbot makers, who can summarize and synthesize text in a way that courts are more likely to consider transformative. Imagine a student using an AI to prepare a report about the American Civil War that includes text from a novel on the subject, as opposed to someone asking an AI to create new music based on existing music.
Albert said the student’s example “definitely sounds like a different purpose than logging onto a music creation tool and saying, ‘Hey, I want to make a song that sounds like a top 10 artist.’” “The purpose is pretty much the same as what the artist would have set out to do in the first place.”
Last year’s Supreme Court ruling on fair use could have a huge impact on music cases because it primarily focused on whether a new use has the same commercial purpose as the original work. This argument is a key part of the Suno and Udio complaints, which state that the companies use the labels’ music “for the ultimate purpose of luring listeners, fans, and potential licensees of the sound recordings (they) have copied.”
Merritt said he worries that technology companies might try to use AI to replace artists like him. He said that if musicians’ songs can be extracted for free and used to copy them, the economics are straightforward. “Robots and AI don’t get royalties,” he said.
(Except for the headline, this story has not been edited by NDTV staff and is published from a syndicated feed.)
