Allegedly leaked Suno code sharpens the fight over how AI music training data was obtained

TechCrunch and 404 Media report that a hacker used compromised employee credentials to obtain Suno source code that appears to document large-scale collection of audio and lyrics from YouTube Music, Deezer, Genius, Pond5, and other sources. The reported material is potentially significant because Suno’s method of acquiring training files is central to record-industry litigation and future licensing leverage. The alleged code has not been independently authenticated here, although prior court filings and Suno’s earlier admissions support the broader proposition that its models used large quantities of publicly accessible copyrighted music.