Internal Data Breach Reveals Scale of AI Training

According to reports from 404 Media, the AI-powered music generation platform Suno recently suffered a security breach. The compromised database provides a rare look into the training data utilized by the company, confirming long-standing suspicions regarding the platforms from which Suno harvests its audio material.

The leaked files reveal an extensive collection of data scraped from major sources, including YouTube Music, Genius, and Deezer. Notable statistics from the breach include:

  • Over 2 million music clips associated with a file labeled "youtube_music."
  • Approximately 113,879 hours of content from YouTube Music.
  • Over 150,000 hours of tagged YouTube content.
  • Significant data volumes from Genius and Deezer.

The total duration of audio ingested suggests that the platform has processed over a decade's worth of music. Furthermore, the internal code indicates the usage of third-party proxies and resources like PodcastIndex to identify and scrape hundreds of thousands of podcast files.


Suno's Official Response

In a statement regarding the incident, a spokesperson for Suno clarified the company's stance:

«As we have stated in public filings and disclosures, Suno's AI models have been trained on publicly available music files and related metadata accessible on third-party websites on the open internet.»

The company also addressed the security incident itself, claiming that in November 2025, they contained a breach that primarily involved obsolete source code. Suno emphasized that no sensitive user information was compromised during the event.


Legal Context and Industry Skepticism

The debate surrounding AI training and copyright remains complex. While the Recording Industry Association of America (RIAA) has criticized the unauthorized scraping of tracks, Suno maintains that its activities fall under fair use protections. Previous court rulings involving major tech firms like Anthropic and Meta have occasionally favored the "fair use" argument, though the legal landscape remains volatile.

Suno asserts that its AI outputs are distinct from the original copyrighted works. The company claims to be implementing safeguards to prevent artist impersonation and misuse.

Despite these assurances, the music industry remains deeply skeptical. Many creators express strong opposition to their work being used as fuel for generative AI models. Catherine Anne Davies, a musician and board member for the Featured Artists Coalition, highlighted the sentiment among many artists: «Most people don't even want their work to be used for training AI.»