Hackers expose how music AI app Suno scraped over 2 million clips from

A recent hack has exposed how the AI music generation app Suno scraped millions of songs, likely including copyrighted material, from YouTube Music, Genius, and Deezer to train its models, 404 Media reports.

Source code obtained by a hacker under the pseudonym ellie.191 revealed a directory titled 'youtube_music' containing 2,013,545 clips. The dataset totals 113,879 hours of audio files, demonstrating the massive scale of Suno's unauthorized training practices.

Suno defended its actions in public filings, stating that its AI models were trained on publicly available music files accessible on the open internet, aiming to help users create new music rather than replicate existing songs.

The startup currently faces multiple copyright lawsuits from major record labels seeking damages for copyright infringement in 2026.