SZA and Musicians Oppose AI Music Generators Using Songs Without Consent
Musicians including SZA discovered their songs were used to train AI music generators without consent after The Atlantic released a detection tool. SZA and other artists protested the practice, with SZA alleging that Black musicians are disproportionately targeted for data collection.
What happened
The Atlantic released an AI detection tool allowing artists to check if their music appears in datasets used to train AI music generators. The tool encompasses over 21 million songs including work by artists such as Bad Bunny, Taylor Swift, Beyoncé and independent musicians.
SZA discovered that 238 of her songs had been used to train AI music models, some reportedly unreleased. She posted on Instagram that AI companies disproportionately exploit Black artists and called on musicians not to support what she termed 'degenerate' AI music use.
Reported impact
SZA stated that Black artists lack protection in legislature, medicine or creative domains and are 'the easiest to steal from.' Producer Kenneth Blume expressed concern about 'stealing from countless struggling musicians' and questioned how developers can 'earn a paycheck obliterating the work and dreams of artists.'
Index analysis
Suno's chief product officer stated the company's training metadata does not include artist names, cannot replicate material it was trained on and that Suno is working to improve impersonation detection. Google and Stability have admitted to using the datasets, though it remains unclear which other developers have done so.
Governance controls
The Atlantic's tool was created by researcher Alex Reisner to allow artists to identify their music in training datasets. The tool reveals that datasets feature songs accessed via YouTube and Spotify links, which developers can search using automated methods that may bypass creator protections.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.