Home / Arts and Entertainment / AI Music Training Data Sparks Artist Uproar
AI Music Training Data Sparks Artist Uproar
21 Jun
Summary
- A new tool reveals if artists' music is in AI training data sets.
- Musicians express anger, claiming AI exploits artists, especially black artists.
- Some AI companies used data sets that bypassed creator earnings mechanisms.
A recently launched AI detection tool has ignited controversy within the music industry. Developed by researcher Alex Reisner, this tool enables artists to search for their music within extensive data sets used for training AI music generators. These sets reportedly contain millions of songs, including those from major stars and independent artists.
Several musicians have expressed significant anger and concern upon finding their work within these training data sets. SZA, for example, stated that 238 of her songs were used, some potentially unreleased, and asserted that AI companies exploit black artists disproportionately. Producer Kenneth Blume (formerly Kenny Beats) directly criticized AI music company Suno for allegedly stealing from musicians and "obliterating the work and dreams of artists."
Questions have arisen about the methods AI developers use to access these data sets. Some platforms reportedly utilize automated methods that bypass logins and advertisements on services like YouTube and Spotify, potentially violating their terms of service and depriving creators of earnings. While companies like Google and Stability have acknowledged using these sets, the extent of other developers' usage remains unclear.
However, not all artists share the same outrage. Producer Hudson Mohawke suggested that the music industry has never been fair and that artists have long faced issues with illegal releases and sampling. Meanwhile, prominent AI music companies like Suno and Udio have faced lawsuits from major labels, and the American Federation of Musicians has also sued major music groups over the use of their music by AI companies.
