Hacked Files Show Suno Scraped 2M YouTube Clips
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Suno scraped at least 2,013,545 YouTube Music clips and hundreds of thousands of hours of audio from Deezer, Genius, IMSLP, Jamendo, Pond5, Freesound, and MuseScore lyrics to train its AI models, according to hacked data shared with 404 Media.
- The hacker "ellie.191" shared Suno source code from 2023 and 2024 containing scraping instructions, with additional code suggesting Suno used third-party firm Bright Data to rip music from YouTube and searched for a cappella-only vocal tracks.
- Suno also sought to download roughly one million hours of podcasts via a tool called PodcastIndex, according to the leaked code.
- The RIAA has sued Suno alleging copyright infringement, with an amended complaint claiming Suno unlawfully "stream ripped" tracks from YouTube — allegations the leaked data now appears to substantiate.
- Suno defended its methods to 404 Media, stating its models were trained on "publicly available music files and related metadata accessible on third-party websites on the open Internet."
- Suno customer information — including email addresses, phone numbers, and Stripe payment details — was also accessed in the hack, and customers contacted by 404 Media said the company never notified them of a breach.
- Suno said it learned of the incident in November 2025, claimed the exposed source code was outdated and no sensitive personal data was compromised, and concluded individual breach notifications weren't warranted.
Why it matters: The leaked data gives the RIAA concrete evidence — over 2 million YouTube clips and a scraper apparently designed to bypass protections — to back its allegations that Suno trained on copyrighted music, potentially undermining the fair use defense Suno has staked its legal position on. Suno's simultaneous failure to notify affected customers adds a privacy compliance dimension on top of the copyright fight.
