Google says YouTube uploaders licensed their music for AI training, and asks the court to toss the Lyria lawsuit
Google asked a court in June 2026 to dismiss the copyright lawsuit a group of independent artists filed over Lyria, its music AI model. The argument skips fair use entirely: per the filing, the artists “each granted YouTube and Google… a broad licence” to their music the moment they uploaded it, and the licence “authorised the conduct alleged in the complaint,” as Complete Music Update reports.
In plain terms: you agreed to this when you clicked upload.
What Google’s motion to dismiss argues
The artists’ complaint says Google took music they uploaded so fans could stream it and used it as training data for Lyria, which now generates music competing with those same artists. Their original framing was blunt: Google had every opportunity to train its AI legally and chose the cheap route instead.
Google’s response doesn’t defend the training as fair use. It says no permission question exists, because the permission was already granted in the terms of service, per Music Business Worldwide.
The YouTube terms clause doing the heavy lifting
Google’s lawyers point at the licence every creator accepts at upload.
A worldwide, non-exclusive, royalty-free, sub-licensable and transferable licence... to reproduce, distribute and prepare derivative works.
The filing adds the licence extends to YouTube’s “affiliates,” meaning every division of Alphabet. So music uploaded for streaming on YouTube is, in Google’s reading, licensed to the teams building Lyria, the model behind the song generation inside the Gemini app for 750 million users.
What the filing means for musicians who upload to YouTube
If the court accepts the argument, the playbook spreads. Every platform with a broad upload licence in its terms gains a ready-made AI training defense, and the question shifts from “was this fair use” to “what did you click agree to in 2015.” Independent artists have already accused Google of using ContentID to launder AI training at scale; this filing turns the suspicion into open legal strategy.
Skipping YouTube is not a real option for a working musician. The platform is discovery, fan service, and revenue at once. The leverage sits in legislation instead, like France’s Darcos Bill, which would force AI companies to prove they didn’t train on copyrighted music rather than letting buried clauses settle it.
Frequently asked questions
What did Google argue in the Lyria AI training lawsuit?
In a June 2026 court filing, Google asked for the independent artists' lawsuit to be dismissed, arguing each plaintiff granted YouTube and Google a broad licence to their uploaded music by accepting YouTube's terms of service. Google says the licence authorised the conduct alleged in the complaint, meaning AI training required no extra permission.
Does uploading music to YouTube let Google train AI on it?
That is Google's position. YouTube's terms grant a worldwide, non-exclusive, royalty-free, sub-licensable and transferable licence to reproduce uploads and prepare derivative works, extending to Alphabet affiliates. Whether a court reads this hosting clause to cover AI training is exactly what the Lyria case will decide.
Who sued Google over the Lyria music AI model?
A group of independent artists. Their complaint says Google used music they uploaded to YouTube for fans to stream as training data for Lyria, which now generates music competing with them. They argue Google had every opportunity to license AI training properly and chose not to.

