SZA, Kenny Beats speak out against their catalogs being used to train AI
A number of artists made similar statements after using a tool that reveals songs used to train AI music generators.
Photo by Jamie McCarthy/Getty Images
Two weeks ago, the American Federation of Musicians sued UMG and Warner Music over their alleged licensing of proprietary titles to two “AI music generator” companies Suno and Udio. Last week, The Atlantic released a tool that allows anyone to search a comprehensive dataset of over 21 million tracks being used to train AI, often without the express consent of their progenitors. In testing the tool, the magazine found songs that strongly resembled hits by Michael Jackson, Ed Sheeran, Chuck Berry, and more.
Musicians, having gotten their hands on the tool, are now speaking out against it. But none as forcefully as SZA, who wrote in an Instagram story yesterday: “Jus checked and music AI has trained off 238 of my songs. I’m certain some unreleased. If your a musician and you support this degenerate shit? Your disgusting and there’s NOTHING YOU COULD EVER SAY TO ME TO MAKE THIS OKAY.” The musician continued the rant on a spam account, noting, “I AINT HEARD A WHITE AI SONG YET.. why so disproportionate? We have no protection in legislature medical or creative. The easiest to steal from.” In the same statement, she alleged that the DJ Diplo had equity in Suno “and is actively attempting to train it on the best and brightest black minds.”
Producer and DJ Kenny Beats, born Kenneth Blume, also refused to mince words against Suno. “I can’t imagine going into work daily knowing you are stealing from countless struggling musicians,” he wrote. “I can’t imagine being proud to earn a paycheck obliterating the work and dreams of artists.” As reported in The Atlantic, Suno wrote in 2024 that it trained its models using “essentially all music files of reasonable quantity” that were downloadable from the internet. The company, alongside Udio, has argued that such processes are legal due to fair use laws. Users of the platform are subject to detection software meant to prevent them from generating songs meant to imitate extant ones, but, as The Atlantic’s investigation makes clear, loopholes abound.