Suno AI's Training Data Exposed: YouTube, Deezer, and Genius Scraped for Music Models

A security breach at AI music generator Suno has exposed internal documents from 2023-2024, revealing that the company scraped millions of songs and lyrics from YouTube Music, Deezer, and Genius to train its AI models. This leak provides the first concrete evidence of Suno's training data sources, substantiating the RIAA's allegations of "stream ripping" from YouTube. For broader context, explore our AI News.

The Scope of the Data Scraping Operation

The leaked source code, reportedly shared by a hacker, outlines specific instructions and dataset details that shed light on the scale of Suno's data acquisition. According to the breach, Suno's training datasets included a staggering 2,013,545 clips from YouTube Music, totaling 113,879 hours of audio content. Additionally, tens of thousands of hours of content were reportedly scraped from other platforms, underscoring a broad approach to data collection.

Beyond YouTube Music, the leaked code specifically points to scraping activities from Deezer, Genius, Pond5, Jamendo, Freesound, IMSLP, and PodcastIndex. While Suno had previously acknowledged in court filings that it trained on "tens of millions of recordings" and "essentially all music files of reasonable quality" available on the open internet, it had never publicly disclosed these specific sources until now.

Security Breach Reveals Customer Data and Unnotified Users

The security incident not only exposed Suno's training data practices but also compromised sensitive customer information. The breach reportedly accessed customer emails, phone numbers, and Stripe payment details. Despite the severity of this exposure, Suno claims the incident was contained in November 2025 and that no sensitive personal information was ultimately compromised.

However, a significant concern arising from the breach is that affected customers were never formally notified of the data compromise. This lack of transparency raises questions about data privacy protocols and user communication following security incidents, particularly when financial details like Stripe payment information are involved.

Implications for AI Music and Copyright Law

This leak provides crucial evidence that directly supports allegations made by the RIAA (Recording Industry Association of America) concerning "stream ripping" from YouTube. The RIAA has been vocal about AI companies allegedly using copyrighted material without proper licensing to train their models. The detailed scraping instructions and dataset specifics from Suno's breach offer tangible proof that could significantly impact ongoing legal battles and future copyright discussions in the AI music space.

The revelation highlights the complex legal and ethical landscape surrounding AI model training, especially when it involves vast quantities of existing creative works. As AI music generation tools become more sophisticated, the debate over fair use, intellectual property rights, and compensation for creators is intensifying. This incident serves as a stark reminder of the need for greater transparency from AI developers regarding their data sourcing practices.

What This Means for the Future of AI Music Tools

The exposure of Suno's data acquisition methods could set a precedent for increased scrutiny on other AI music generation platforms. Developers may face mounting pressure to disclose their training datasets and ensure compliance with copyright laws. For users and creators, this news underscores the importance of understanding how the AI tools they use are built and the potential legal ramifications involved.

As the industry evolves, we can expect more robust discussions and potentially new regulations around data provenance for AI models. This incident reinforces the critical need for ethical AI development that respects intellectual property while fostering innovation.

Sources

• HAIM: Human-AI Music Datasets for AI Music Production Tracking Benchmark
• https://www.theverge.com/ai-artificial-intelligence/966072/suno-ai-music-training-scraping-youtube-hack
• Hack suggests AI music generator Suno scraped YouTube for training data - TechCrunch
• Record labels claim AI generator Suno ripped their songs from YouTube | The Verge
• https://www.404media.co/hack-reveals-suno-ai-music-generator-scraped-youtube-deezer-and-genius/