SZA criticized AI music tools after an AI music database search appeared to show 238 of her songs had been used in training datasets.
SZA has sharply criticized AI music tools after sharing an Instagram Stories post that appeared to show 238 of her songs listed in AI training datasets, escalating her public objections to the technology’s use in music.
The Grammy-winning American singer-songwriter, 36, said she searched her name on an AI music database and found more than 200 songs connected to training data, including what she believed may have been unreleased material. The claim, reported by Euronews, adds to a growing dispute between artists and AI companies over whether creative work is being used without permission to build tools that can generate music.
“Checked and music AI has trained off 238 of my songs. I’m certain some unreleased,” SZA wrote, according to the report. “If you’re a musician and you support this degenerate shit? You’re disgusting and there’s NOTHING YOU COULD EVER SAY TO ME TO MAKE THIS OKAY.”
SZA, known for songs including “Kill Bill” and “Luther” with Kendrick Lamar, also singled out AI music company Suno and producer Diplo in a later post. She alleged that Diplo has equity in Suno and accused the company of seeking to train its tools on Black writers and producers. The Euronews report did not include a response from Diplo.
Suno has publicly defended its approach to AI music generation. In a LinkedIn post last week cited by Euronews, Suno chief product officer Jack Brody wrote that “preserving artistry and human creativity is just as important as product innovation itself,” and said the company has rules against users uploading or distributing content they do not own or have rights to use.
Brody also said Suno works with providers including Audible Magic, Musixmatch and ACRCloud to help identify and prevent misuse. Addressing concerns that AI models can reproduce training material, he wrote that Suno’s position is that “no, that should not happen,” and said the company does not use artist names as a training metadata category because its goal is to help users make new songs rather than music that sounds like existing artists.
SZA’s latest comments build on earlier criticism of AI. In March, she told i-D Magazine she felt “at war because of AI,” arguing that AI-generated music was affecting Black artists disproportionately. Last year, she also criticized AI’s environmental costs and warned that Black and brown communities would bear the brunt of its consequences.
The exchange underscores a central unresolved question for the music industry: how AI companies obtain and use training data, and what protections artists have when their work is used to develop systems that can create new songs.
Comments (0)