Fraudsters Are Releasing AI Music Under Real Musicians’ Names

Photo Credit: Emily Portman by Harald Krichel / CC by 3.0

Artists like Emily Portman are perplexed by fraudsters releasing AI-generated music in their names and to their official catalogs.

Folk singer-songwriter Emily Portman would normally have been elated to receive a message from a fan about how much they loved her latest album. But the artist hadn’t released a new album, so what was the new release that appeared on Spotify, iTunes, and more platforms, and where had it come from?

Portman said she followed a link from her fan to what appeared to be her latest release, an album called Orca, which she had never seen nor heard before. “I clicked through and discovered an album online everywhere—on Spotify and iTunes and all the online platforms,” she told the BBC. “It was called Orca, and it was music that was evidently AI-generated, but it had been cleverly trained, I think, on me.”

The tracks were “uncannily close” to titles the artist says she might have chosen. And the voice, supposedly hers, was eerily similar but sang in “a folk style probably closest to mine that AI could produce.” The whole thing, she said, was “really creepy.”

Usually when AI-generated music rears its head, it usually appears under fake names, or tries to imitate major artists but without actually appearing on their official channels. However, there’s a steadily growing trend of established (but notably not “superstar”) artists being targeted by fake music appearing on their official streaming pages. Even artists who have long since passed away are suddenly having “new” material, generated by AI, appearing on their catalogs.

Another album has since appeared on Portman’s streaming pages, but this time, less effort had gone into creating something that seemed like a believable release. Instead, it was “20 tracks of instrumental drivel,” she said. “Just AI slop.”

Portman has filed copyright complaints to get the fraudulent albums taken down, but it’s been an ongoing process. She says that some platforms were quick to take action, but Spotify took three weeks—and Portman says she still hasn’t regained control of her own Spotify artist profile.

According to Spotify: “These albums were incorrectly added to the wrong profile of a different artist by the same name and were removed once flagged.” But Portman questions the legitimacy of that claim.

Meanwhile, other artists have been fighting similar battles, like New York-based producer and songwriter Josh Kaufman, who is known for playing on Taylor Swift’s Folklore album. “I just started getting messages from fans and friends about some new music I just released, and how much of a shift it was [stylistically],” says Kaufman.

“I think most people were hip to the fact that it was somebody else just using my artist profile as a way to release some strange music that clearly was computer generated,” he said. “It was embarrassing and then just kind of confusing. […] [Music] is the thing that we do, right? This is the signature of our soul, and that someone else can walk in there and just have access like that…”

Other folk musicians who have had fake tracks posted under their names in recent weeks include Wilco frontman Jeff Tweedy, Father John Misty, Sam Beam (Iron & Wine), Teddy Thompson, and Jakob Dylan.

Interestingly, all seem to be from the same source, with releases using the same style of AI artwork and credited to three record labels—two of which have apparent Indonesian names. Many list the same songwriter: Zyan Maliq Mahardika, a name that has also been credited on other songs mimicking real artists in other genres, including metalcore and Christian music.

According to Tatiana Cirisano of media and technology analysis company Midia Research, AI is “making it easier for fraudsters” to fool listeners, who are notably “more passive” in the age of streaming and algorithms. Cirisano says bad actors posing as real artists are hoping to “rack up enough streams” to earn a nice check.

“I would think that the AI fakes are targeting lesser-known artists in the hopes that their schemes fly under the radar, compared to if they were to target a superstar who could immediately get Spotify on the line,” she says.

However, streaming services and distributors are “working hard” to get better at spotting AI fakes, but “ironically, also by using AI and machine learning.”

“I think it’s clear to everyone that every stakeholder must do their part,” says Cirisano. “But it’s complicated.”

And it will probably get a lot more complicated before it gets better. While distributors and streaming services are using AI and machine learning to better spot the fakes, bad actors are also using AI and machine learning to get better at skirting detection. The issue is reminiscent of the rise in piracy during the dawn of online file sharing, and undoubtedly will cause more headaches for artists and rights holders before the law can catch up.