• 2 min read
Hack exposes Suno’s use of 2 million YouTube tracks
Leaked Suno code points to music scraped from YouTube Music, Deezer, Genius, and more as lawsuits over copyright training continue.
Image: TechRadar
A hack of Suno appears to have revealed new details about how the AI music company built its training data, including references to songs pulled from YouTube Music, Deezer, Genius, and the International Music Score Library Project.
According to 404 Media, a hacker using the name ellie.191 accessed Suno source code and training libraries. The material, dating from 2023 and 2024, includes references to 2,013,545 tracks ripped from YouTube Music and 12,287 hours of music ingested from Deezer.
The leaked code also appears to show some of the company’s collection methods. Suno’s software reportedly searched YouTube for acapella versions of songs to train on vocals, and it also targeted large numbers of podcasts.
Copyright lawsuits against Suno
Suno is already facing multiple lawsuits over training its models on copyrighted music without permission, including a case filed with the participation of the Recording Industry Association of America (RIAA).
The central dispute is not whether Suno used copyrighted music. The company has already said it trained on “essentially all music files of reasonable quality that are accessible on the open internet.” The legal fight is over whether that use qualifies as fair use.
Recommended reading
Hassabis says STEM makes you 10x better at AI
The same conflict is playing out across other creative industries, including writing, photography, and filmmaking, where rights holders argue that AI companies rely on human-made work without compensation.
Online reaction to the latest leak has been blunt. One Reddit commenter wrote, “this is literally what every LLM in existence has done,” while another described the practice as “staggering theft.”
AI Editor
Ava covers the rapidly evolving world of artificial intelligence, from foundational models and research labs to the real-world economics of intelligence. With a background in computational linguistics, she cuts through the hype to find out what actually works. She firmly believes that benchmarks are just marketing until reproduced in the wild.
via TechRadar
