Leak of Suno code shows AI scraped 2M YouTube Music tracks
It turns out AI music generation isn't pure digital magic born from synthetic souls. A massive leak just pulled back the curtain on how these algorithms really learned to sing, and copyright lawyers are already pouring champagne.
A hacker breached internal repositories of Suno back in November 2025, revealing internal metadata and scripts that the company never publicly disclosed. The leaked codebase contained specific instructions for harvesting over two million music videos from YouTube Music using proxy infrastructure from Bright Data.
The repository listed scrapers for platforms like Deezer, Genius, Pond5, Jamendo, and Freesound, along with scripts aimed at isolating clean acapella vocals. To round out the digital buffet, the system was configured to ingest roughly one million hours of podcasts via Podcast Index.
Representatives at Suno confirmed to 404 Media that their models were trained on publicly accessible web files, while declining to address specific scraped services. Because user payment processing was handled externally by Stripe, management decided against notifying customers about the breach.
When fair use arguments collide with automated wholesale harvesting of global culture, legal definitions face their ultimate trial by fire. Music labels and independent creators will now decide whether the future of synthetic art belongs to technological pioneers or corporate plunderers.
Source: 404 Media
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.