Millions of pop songs, hip hop, reggae, dancehall and songs from other genres, recorded by some of the biggest stars in music, are being made available to artificial intelligence companies, potentially to train the next generation of AI music models.
According to a report in the Atlantic titled, “The Millions of Songs Mashed Into AI-Generated Music,” AI developers have access to massive music datasets containing millions of songs.
A search of the music datasets holding millions of songs by World Music Views found that dancehall and reggae music is heavily represented in these datasets. Artist like Drake, Beyoncé and Vybz Kartel appear with over 500 results each, Popcaan with 232, Shaggy with 283, Major Lazer with 200 titles, Masicka with 296, Shenseea with 139, Stefflon Don with 148, Spice with 141, Shabba Ranks with 147 and Rvssian with 58. Even Yellowman’s Where Is Santa Claus is being used to train audio generating models. And those are only a few names.
As the AI continues to train and request more information, just about every prominent artists whose music is available online including producers, featured artists, riddims, collaborations, older catalogs, and Jamaican-linked songs — could push the number into the hundreds of millions of tracks being available to AI developers.
The AI being trained such as Suno, uses existing recordings by breaking songs into tiny pieces of sound, study the patterns, and learn how to generate new tracks that can sound like real human music.
Alex Reisner’s Atlantic report says he found four large music datasets circulating in the AI-development community. One contains 12 million tracks. Another contains 9 million. Two smaller datasets contain more than 100,000 songs each. The 12-million-song dataset alone would take 91 years to listen to from beginning to end.
Reggae and dancehall is just a drop in the bucket being used by these datasets as it affects major artists across Hip Hop, pop, rock, jazz, classical, and other genres.
The report explains that some datasets are shared as lists of links to songs on YouTube and Spotify. Developers can then use automated tools to download the audio. Some of those tools bypass logins, ads, and systems that help artists and platforms earn money. That raises serious questions about consent and compensation.
Grammy-Award winning producer Alex Antaeus took to Instagram on Saturday to sound the alarm stating, “That has happened without our knowledge, without our permission, and without any compensation to the creators.”
Alexx said he is not against AI but says, “I believe AI has the potential to be a powerful force for innovation and creativity.”
AI companies often argue that training on music is “fair use.” But artists and record labels disagree. According to the report, musicians and labels have filed at least 12 lawsuits against AI companies over the use of copyrighted music to train models. Major labels have sued Suno and Udio, two popular AI-music companies.
The concern is not just that AI may copy a song directly. The concern is that AI can absorb the sound, style, rhythm, and vocal patterns of real artists, then generate new music that competes with them.
A platform could generate a “dancehall-style” song in seconds. It could imitate the bounce of a riddim, the vocal delivery of a deejay, or the sound of modern Jamaican production. That could reduce opportunities for real Jamaican artists, producers, engineers, and songwriters.
The Atlantic report also points out that AI music is spreading quickly. Last September Spotify removed millions of spammy AI-generated tracks, while Deezer has reported that a large portion of the music it receives daily is AI-generated. This means AI music is not a future problem. It is already here.
If hundreds of thousands of dancehall songs are helping train AI, then the creators behind those songs deserve answers. Which songs were used? Who downloaded them? Were rights holders asked? Were artists paid? Can they opt out?
Alexx Antaeus argued that, “if an artist’s work is being used to train these systems, there must be transparency, consent, and fair compensation. Innovation and creators’ rights should go hand in hand.”
