Leaked data from a recent hacking incident has revealed new details about how AI music company Suno built its training datasets, suggesting the platform collected millions of songs and lyrics from popular online music services, including YouTube Music, Deezer, and Genius.
The leaked information, first reported by 404 Media, offers one of the clearest looks yet at Suno’s training process. The company has largely avoided disclosing what data it used to train its AI models, even as it faces multiple copyright lawsuits over the issue.
According to the leaked materials, the hacker, identified as “ellie.191,” obtained Suno source code dating from 2023 and 2024. The files reportedly include instructions for scraping audio and lyrics from several online platforms, including YouTube Music, Deezer, Genius, Pond5, Jamendo, Freesound, and the International Music Score Library Project (IMSLP).
The leaked code also appears to show that Suno relied on web scraping services from Bright Data to collect music from YouTube. In addition, some of the code reportedly searched specifically for a cappella versions of songs, allowing the company to gather vocal-only recordings for AI training.
One file referenced in the leak indicates that Suno had downloaded more than 2 million YouTube Music clips, with the dataset containing 2,013,545 audio clips at the time the file was last updated.
The revelations add to the legal challenges already facing Suno. The company is being sued by the Recording Industry Association of America (RIAA), which alleges that Suno trained its AI models using copyrighted music without permission. Suno has acknowledged using copyrighted material during training but argues that the practice is protected under the legal doctrine of fair use.
The RIAA has also accused Suno of deliberately bypassing YouTube’s copyright protections through a process commonly known as “stream ripping” to obtain music from the platform. The newly leaked files reportedly support those allegations by showing automated tools and scraping instructions used to collect audio from various online sources.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
The leaked documents provide rare insight into the datasets behind one of the most widely used AI music generators and are likely to play an important role as ongoing copyright lawsuits against the company continue through the courts.
