An article published by The Atlantic on June 14 detailed how AI companies use massive datasets of copyrighted songs to train music-generating models, creating AI-generated tracks that strongly resemble the existing intellectual property of other artists. Alex Reisner discovered "four giant datasets" of songs that are being shared within the AI development community. The datasets included more than 900 entries for Willie Nelson songs that could be used to train AI music-generating systems. Reisner wrote that Suno and its competitor, Udio, can also function as listening platforms similar to Spotify or YouTube. The Atlantic’s four datasets comprising 12 million tracks would take 91 years to listen to. AI companies implement detection software to prevent their products from generating songs that explicitly duplicate existing music, but platforms like Suno or Udio don’t stop users from generating songs imitating the style of real artists. The article states that musicians and labels have filed at least 12 lawsuits against AI companies for training models on copyrighted music.
entertainment
Over 900 Willie Nelson Songs Found in AI Training Music Datasets
An investigation found more than 900 Willie Nelson songs in four datasets with 12 million tracks are being used to train AI music models.
Published June 15, 2026 at 6:46pm by Mars Salazar
