Needless to say, the revelation that singer Xania Monet is an AI creation that prompted a label bidding war and a multimillion-dollar advance is not Kehlani‘s favorite music-business development. “Nothing and no one on Earth will ever be able to justify AI to me,” the R&B star declared on TikTok last week. “Especially not AI in the creative arts, in which people have worked hard for, trained for, slept on the floor for, f–king got injuries for, worked for their entire lives. I’m sorry, I don’t respect it.”
Monet’s success — her five songs have racked up 17 million U.S. streams and generated an estimated $52,000 over two months — has led to ethical conflicts in the business. Some reps from indie labels who spoke with Billboard say they would never sign an AI artist because they’re committed to human creators. Like Kehlani, they resent the idea that a poet in her bedroom pushing a few buttons can compete with human artists. But others in the business say Xania Monets are acceptable if the music companies behind them behave responsibly, respect international copyright law and follow the policies of streaming services like Spotify (which recently updated its policies to avoid “AI slop” and removed 75 million “spammy tracks”) and Deezer (which announced in June that it would flag AI-generated content with a prominent message to users).
“It just depends. If you cook a meal for somebody, but you’re doing so with stolen food, that’s different from going to the supermarket, buying food and cooking the meal,” says Meng Ru Kuok, founder and CEO of Caldecott Music Group, which operates BandLab Technologies, a music-creation service that enables AI tools. “We don’t want to punish people for doing things the right way.”
Monet is the creation of 31–year-old poet Telisha Jones from Olive Branch, Miss., whose manager says she writes her own lyrics but uses Suno, an AI music service, for other elements of her tracks. That is concerning for major labels, which sued Suno and another AI-music firm, Udio, in June 2024, alleging copyright infringement on what they called “an almost unimaginable scale.” (A rep for the Recording Industry Association of America, which represents the three majors and led the lawsuits, declined to comment.)
“There’s an ethical component here,” says Ryan Schmidt, a Savannah, Ga.-based music lawyer. “If you created your label because you want to promote young, up-and-coming talent, and you want to advance music, then AI art might not be it for you. If you are a label who wants to be in business to have the biggest catalog possible, and maximize earnings, AI’s certainly one avenue to do that.”
On Thursday (Sept. 25), Spotify attempted to draw ethical boundaries by beefing up protections against what it calls “bad actors” that flood the music streaming service with often AI-enhanced tracks that “dilute the royalty pool,” according to a Spotify announcement.
Spotify’s royalty payout formula takes the total number of streams on the service per month and divides that by the market share of individual rightsholders on the platform. So if so-called “AI slop” artificially boosts the overall track number, or accrues a significant stream count, the proportion paid to each rightsholder shrinks. According to Spotify, overall payouts to music rightsholders increased from $1 billion in 2014 to $10 billion last year.
In the case of Monet, according to Alex Bestall, founder of production-music company Rightsify, it’s unclear how much of her creation involved AI systems and how much was human input from Jones. Under U.S. copyright law, music created solely by machines cannot be copyrighted. “If you just say, ‘Make me a pop song,’ that’s generic,” Bestall says. “But if you gave it the exact chord progressions, the structure and add in your own human vocal, then yes, that should probably be copyright protected.” Regarding Monet — and AI artists in general — disclosure of what came from robots and what came from humans, he adds, is “the great first step.”
In their Suno and Udio lawsuits, the major labels criticized the tech companies for “unlawfully” copying labels’ recordings to “train their AI models to generate music that could saturate the market with machine-generated content.” Still, one major label made an offer to sign Monet, even though her creator used Suno — which would have meant the label was profiting off the use of Suno, potentially undermining its lawsuit against the company. In the end, the winning label was the indie Hallwood Media, led by Neil Jacobson, a former executive with Universal Music Group-owned Interscope Records.
Terry McBride, co-founder and longtime CEO of Nettwerk Music Group, the Canadian indie label that broke Sarah McLachlan, Paris Paloma and many others, says he would never have considered entering such a bidding war. Regarding Monet, he says, “I believe the author is writing the lyrics and the AI is taking that and putting an AI voice and music to it — that’s not going to be a touring entity as we know it,” he says. “We would not sign that. Even if it did hundreds of millions of streams, we have no interest in that.”
Ben Swanson, COO of Secretly Group, the Indianapolis indie label home of Mitski, Phoebe Bridgers, Angel Olsen and others, has a more nuanced position on Monet. He acknowledges AI is a revolutionary tool for music-making, comparing it to the way ProTools, synthesizers and drum machines once changed music. But he advocates for a compensation system when AI trains its source material on copyrighted music created by others.
“There will be some AI artists that become culturally significant — and maybe even financially significant — but I think those will be the outliers,” says Swanson, adding that he has never been involved in “any sort of negotiation” with an AI artist. Does he fear competition from labels that sign AI artists, because they could potentially dilute the royalty pool for Secretly Group — or simply compete for exposure with his label’s artists on streaming services, radio stations and so on? “If the majors want to get into a bidding war and spend $3 million on an AI artist,” he responds, “have at it.”