Hacked source code from AI music generator Suno reveals the company scraped YouTube Music, Deezer, Genius and other platforms to train its models, according to a report by 404 Media. The leak, obtained by a hacker who breached Suno's systems, corroborates allegations made by Universal Music Group and Sony Music Entertainment in their ongoing copyright infringement lawsuit against the company, coordinated by the RIAA, potentially strengthening the labels' case as it heads through the courts.

The code logged huge volumes of scraped material, including over 2 million clips totalling 113,879 hours from YouTube Music plus a further 152,162 hours tagged separately, alongside 62,117 hours from stock library Pond5, 19,514 hours from IMSLP, 17,615 hours from Genius and 12,287 hours from Deezer. The same hacker also accessed data on hundreds of thousands of Suno customers, including emails, phone numbers and Stripe payment details. The code reportedly shows Suno using Bright Data proxies to scrape YouTube and searching for a capella tracks to isolate vocals, backing the RIAA's September 2025 claim that Suno "stream ripped" songs from YouTube by circumventing its anti-piracy encryption, a potential breach of the Copyright Act's anti-circumvention rules. Suno has previously admitted in court filings to training on "essentially all music files of reasonable quality" available on the open internet, arguing this is protected as fair use.