Jason Moran, the jazz composer and pianist who served as artistic director for jazz at the Kennedy Center, doesn't use Spotify. He puts his music on Bandcamp. So when bassist Burniss Earl Travis called him last month about a new EP titled For You on Moran's Spotify profile — moody anime cover art, indie pop sound, not a piano in sight — the problem was clear. Someone, or something, had attached AI-generated music to his name. "There's not even a piano player on this whole damn record," Moran said. Moran is one case in a systemic pattern. In jazz alone, AI impersonation has hit pianist Benny Green, saxophonist Antonio Hart, drummer Nate Smith, the Australian band Hiatus Kaiyote, and singers Dee Dee Bridgewater, Jazzmeia Horn, and Freddy Cole. Beyond jazz: indie rock musicians Luke Temple and Uncle Tupelo, the late electro-pop artist Sophie, country singer Blaze Foley, and the rapper Drake have all been targeted. When King Gizzard and the Lizard Wizard removed their music from Spotify in December, an AI impersonator called King Lizard Wizard filled the void with identical song titles and imitated artwork. The economics make the incentive structure obvious. Morgan Hayduk, co-CEO of fraud detection firm Beatdapp, estimates 5% to 10% of all streams across the industry are fraudulent — worth $1 billion to $2 billion per year. That's money diverted from legitimate artists into the pockets of bad actors. Last month, Michael Smith pleaded guilty to defrauding streaming platforms by flooding them with thousands of AI-generated songs, using bots to inflate listens into the billions, and collecting more than $10 million in royalties over seven years. AI didn't create streaming fraud, but as Hayduk put it: "AI has become an accelerant." The marginal cost of producing a fake song has collapsed to near zero, and any removed content is instantly replaceable. Spotify's response has been real but structurally insufficient. The company removed more than 75 million "spammy tracks" in the 12 months before last September. It announced a new tool last month to let artists review and approve or decline releases before they go live — a feature the company says no other streaming service offers. A spokesperson said Spotify "employs a range of safeguards" including detection systems, human review, and takedown processes. For deceased artists, Spotify says estate or rights holders can opt into the new tool if they have an account; for those without accounts, the company relies on internal detection. The structural problem is who bears the cost. Moran's takedown process required navigating a chatbot, then a human conversation, then a 72-hour wait. "They allow it to just kind of sit there unless the artist finds it and checks it," he said. Days after For You was removed, another album he hadn't made — by the real Belgian band Schntzl — appeared on his profile due to what Spotify called a metadata mix-up. Three weeks ago, For You reappeared on YouTube with the same anime artwork and tracklist. The burden of detection, reporting, and verification falls on artists — many of whom, like Moran, don't even use the platform. And as Moran pointed out: "How does John Coltrane verify or Billie Holiday verify that this new record is not some fake, you know, '1952 just-found concert from Paris'?" Adam Berkowitz, a PhD candidate at the University of Alabama studying AI and copyright law in music, identified the deeper structural tension: streaming platforms are being pushed into a law enforcement role they're not designed for. "It gets a little complicated because all of a sudden, the private sector is enforcing law. And that's just not how it's supposed to be," he said. But the courts can't keep pace with the volume. The onus, Berkowitz concluded, will probably remain on artists. The extraction pattern is clear: generative AI has collapsed the production cost of fraudulent music to near zero while the detection and enforcement costs remain high and are externalized onto individual artists. Spotify's new pre-release approval tool is a step, but it only works for artists who have accounts and actively monitor them. The dead, the disengaged, and the under-resourced remain unprotected. Moran uses Bandcamp precisely because it gives him control over his profile and pricing. "One thing that [people] can never get charged for is the power of the songs," he said. The question is whether the power of the songs matters when the platform doesn't know which songs are real.
