Udio Admits Using YouTube Audio For AI Training Amid Sony Music Lawsuit
Major labels like Sony Music Group, Universal Music Group, and Warner Music Group allege AI platforms used copyrighted music without permission
Major labels like Sony Music Group, Universal Music Group, and Warner Music Group allege AI platforms used copyrighted music without permission
AI music platform Udio has acknowledged that it sourced audio data from YouTube to train its models, as part of its legal response in an ongoing lawsuit filed by Sony Music Entertainment.
The admission comes as the case intensifies, with Sony alleging that Udio used “stream-ripping” techniques, tools that extract audio from YouTube videos, potentially bypassing the platform’s protections to access copyrighted material.
In its filing, Udio confirmed that it “obtained audio data from YouTube” and used a tool known as YT-DLP to gather training material. However, the company continues to defend its actions, arguing that its use of copyrighted recordings falls under the “fair use” doctrine.
The lawsuit is part of a broader legal battle launched in 2024 by major record labels including Sony Music Group, Universal Music Group, and Warner Music Group, accusing AI music platforms like Udio and Suno of using copyrighted works without permission to train generative models.
While Universal and Warner have since settled their disputes with Udio and moved toward licensing agreements, Sony has continued to pursue the case, making it a key test for how copyright law will apply to AI training practices.
A US federal judge has already allowed Sony’s claims to proceed, noting that the label had plausibly argued that Udio may have circumvented YouTube’s technical safeguards, an issue central to potential violations under the Digital Millennium Copyright Act (DMCA).
The case also reflects a wider industry concern over how AI companies source training data. Earlier complaints and amended filings have alleged that platforms like Udio “illegally scraped” content from YouTube and other sources to build their datasets.
As litigation continues, the outcome could have far-reaching implications for the AI music ecosystem, particularly around whether unlicensed use of copyrighted material for training purposes can be legally justified, or will require formal licensing agreements with rights holders.