Musicians including SZA and producer Kenneth Blume have voiced strong objections after discovering their songs in datasets used to train AI music generators. The reactions followed the launch of an AI detection tool by The Atlantic last week.
The tool, created by researcher Alex Reisner, lets artists search whether their work appears in four datasets containing over 21 million songs. These collections draw from catalogs of major and independent artists alike.
SZA posted on Instagram that 238 of her songs, some unreleased, were used for AI training. She called the practice "degenerate shit" and questioned why Black artists appear disproportionately targeted.
Producer Kenneth Blume, known as FKA Kenny Beats, wrote that he could not imagine earning a paycheck by "obliterating the work and dreams of artists." DJ Sabrina the Teenage DJ noted that 22 of her tracks appeared in a Suno dataset.
Google and Stability have admitted using such datasets. Suno and Udio face lawsuits from major labels, while the American Federation of Musicians recently sued Universal Music Group and Warner Music Group over similar issues.
