Google says YouTube’s terms let it train AI on uploaded music

Google Moves to Dismiss Artists’ Copyright Suit, Arguing YouTube Terms Authorize AI Training

Google filed a motion on Monday to dismiss a copyright lawsuit brought by independent musicians, arguing that YouTube’s terms of service grant the company a “broad license” to use uploaded music for training its Lyria 3 AI model — a legal strategy that sidesteps the fair use debate dominating other AI copyright cases.

The License Argument

In the filing submitted to the US District Court for the Northern District of Illinois, Google’s legal team from Quinn Emanuel argued that the plaintiffs “granted YouTube and Google a broad licence” when they uploaded their music to the platform. The motion urged the court to throw out the complaint “with prejudice.”musicbusinessworldwide

Rather than invoking fair use, Google pointed to a specific provision in YouTube’s terms of service that grants the platform and its affiliates a “worldwide, non-exclusive, royalty-free, sublicensable and transferable licence” to “reproduce, distribute and prepare derivative works” from uploaded content. Google’s lawyers contend that “affiliates” encompasses any division of parent company Alphabet, including Google DeepMind, which developed Lyria 3.billboard

“Their lawsuit rests on the unfounded assumption that Google trained its model on their specific works,” the motion states. “Even if we accept their unverified claims as true, the complaint fails.”billboard

Origins of the Lawsuit

The case dates to March, when a coalition of independent artists, songwriters, and producers filed a 118-page complaint alleging Google copied “at least 44 million clips and 280,000 hours of music” from YouTube to train Lyria 3. The model, launched in February through Google’s Gemini app, generates audio clips with vocals and lyrics for more than 750 million monthly active users.musicbusinessworldwide

The artists argue Google exploited its dual role as a distribution platform and AI developer, transitioning “from distributor to competitor” by siphoning copyrighted audio into training data without compensation.billboard

Broader Implications

If the court accepts Google’s interpretation, the ruling could establish that any music uploaded directly to YouTube is available for AI training — a precedent with far-reaching consequences for millions of creators on the platform. The case is one of several AI training disputes now moving through federal courts, including lawsuits against Meta Platforms over its Llama model and against Snap for allegedly scraping YouTube videos. In June 2025, two judges in the Northern District of California found that using copyrighted works to train AI models that do not substantially reproduce their training data constituted transformative fair use — but Google’s terms-of-service strategy charts a different legal path entirely.techbuzz