Suno AI Music Generator Accused of Stealing Songs From YouTube, Deezer, Genius, and More

Suno, one of the internet’s most popular AI music generators, is in hot water after a hack revealed that it was trained on a large collection of songs, lyrics, and podcasts taken from other platforms. Those platforms include YouTube Music, Deezer, Genius, Pond5, Jamendo, Freesound, IMSLP, and podcast RSS feeds, as reported by 404 Media.

The hacked files, obtained by someone who goes by “ellie.191” online, contain millions of recordings and over a hundred thousand hours of audio. One internal dataset tied to YouTube Music alone has over two million clips. Other datasets include stock music libraries, classical scores, lyrics archives, and spoken content from podcasts.

In the past, Suno has openly admitted that it trains its models on “essentially all music files of reasonable quality that are accessible on the open internet” (allegedly subject to paywalls and password protections), as well as related metadata for the audio and lyrics. Labels have accused the company of ripping songs directly from musical platforms in ways that break copyright law, platform terms, anti‑circumvention rules, and more. But Suno claims that using this material is fair use and has set limits to prevent the close mimicking of specific songs.

Notably, Warner Music Group (WMG) settled a massive legal dispute with Suno last year, ultimately agreeing to grant Suno the rights to train on its music catalog. But artists on other platforms arguably have not consented to their music being used as training fodder for AI, putting Suno at risk of fresh legal issues and a massive loss of consumer trust.