AI BriefingSuno hack leaks training data proving copyright infringement
Key point
A hacking incident at generative music AI startup Suno leaked a list of training data unauthorizedly scraped from YouTube, Deezer, and other sources.
Details
While stealing Suno's source code, the hacker 'ellie.191' exposed a massive list of audio training data collected from YouTube Music, Deezer, Genius, Pond5, and more.
The specific scale of the leaked data is as follows:
- YouTube Music: approximately 2.01 million clips (113,879 hours)
- Deezer: 12,287 hours
- Genius: 17,615 hours
- Pond5: 62,117 hours
- Podcasts: approximately 420,000 files collected via RSS feeds (approximately 1 million hours)
This leak is likely to become more than just a security incident—it could develop into a major legal issue. The RIAA (Recording Industry Association of America) is currently pursuing a lawsuit against Suno on charges of 'stream ripping,' and this leaked data list could serve as decisive evidence supporting the claim that Suno used copyrighted music for training.
According to reports, Suno used Bright Data's commercial scraping infrastructure to extract data from YouTube, and it was revealed that this notably included routines for finding Acapella versions of songs.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
NSA Spending Billions to Test Frontier AI Models for Security Vulnerabilities
Ed Newton-Rex Highlights Creator Harm and Regulatory Need for 'Derived Data' in AI Training
Meta AI Agent 'Muse' Internal Structure Revealed: Analysis of 6.8GB FilesystemAnalysis of the 'Inference Gap' Between Claude's Reduced Reasoning Tokens and Benchmarks