Google has filed a motion to dismiss a copyright lawsuit brought by independent musicians over the training of its Lyria 3 AI music generation model. The company's central argument: artists who uploaded their work to YouTube agreed to let Google use it for this purpose by accepting the platform's terms of service. The motion, reported on Music Business Worldwide in June 2026, turns what has been a pending legal dispute into a direct confrontation over what platform terms of service actually allow.

The Core Dispute

Independent musicians filed suit against Google earlier this year, arguing that the company used songs from YouTube's catalog to train Lyria 3 — Google's AI music generator — without asking permission or paying licenses. The musicians contend that uploading a song to YouTube to reach listeners is fundamentally different from agreeing to have that song fed into a competing AI product. As Billboard reported in March 2026, the lawsuit centers on a straightforward question: did Google's use of those recordings fall within whatever license the artists actually gave, or did Google cross the line into unauthorized copying and unauthorized derivation.

The complaint goes beyond just the training question. It challenges Google's broader AI music practices, suggesting the plaintiffs are not focusing narrowly on Lyria 3 but are contesting the entire pipeline that feeds YouTube content into Google's AI systems.

Google's Argument: You Agreed to This

Google's defense is that YouTube's terms of service — the rules users accept when they upload — already cover AI model training. When artists accepted those terms, Google says, they granted a broad license that includes letting Google use their content as training material. Billboard's June coverage frames this as Google arguing there was no copyright violation because there was no unauthorized use in the first place.

Google has been consistent on this point since Lyria 3 launched in late March 2026. When the model first appeared, The Decoder reported that Google stated the model was trained on material that both YouTube and Google had the right to use under their terms of service and existing agreements. The dismissal motion translates that public statement into legal language: no infringement occurred because nothing was used without permission.

The strength of this argument depends entirely on what YouTube's terms of service actually say, and crucially, how a court decides to interpret those words. The relevant clause in YouTube's standard upload agreement gives Google a broad, worldwide license to use, reproduce, distribute, and create derivative works from uploaded content in order to operate and improve Google's services. The legal question now is whether "improving Google's services" stretches far enough to include training a commercial AI product that can itself generate music — and potentially replace the very artists whose songs trained it.

How We've Seen This Pattern Before

This playbook is familiar. We watched it unfold in the early days of the internet, when platforms built major businesses on user-uploaded content while their terms of service gradually expanded to cover almost everything except outright ownership. The music industry contested these expansions repeatedly, from the original Napster fights through the long battles over copyright safe harbors and what became known as the "value gap" — the sense that platforms profited from music without fairly compensating artists. Each time, technology moved faster than the law. Courts and lawmakers always arrived late and always made imperfect decisions.

AI training litigation is running that same dynamic much faster and under higher stakes. The source material is copyrighted creative work. The end product is a commercial generative system. The contractual language Google is relying on — a platform terms of service — was almost certainly written before anyone contemplated AI training as a use case. Whether courts will decide that a broadly written license for one purpose automatically extends to something genuinely new is still an open question, and whatever the court decides here will matter far beyond this one case.

What Really Hangs in the Balance

The Lyria 3 lawsuit sits at the intersection of three separate legal questions that AI music cases are starting to sort out: whether Google committed direct copyright infringement by copying protected works to train the model, whether the model's outputs violate the original creative character of the training material, and — the newest and potentially most important — whether platform terms of service can count as blanket permission for AI training.

The first two questions are being litigated in courts around the world and apply to all kinds of AI systems. The third is newer and arguably more significant for creators across the board. If Google's terms-of-service argument wins at this early stage, it creates precedent that every major platform — YouTube, SoundCloud, Spotify's podcast service, any site hosting user content under similar language — could use to shield its AI training from copyright claims by the creators whose work powers those systems.

For independent musicians, this matters sharply. Major record labels have enough negotiating power to strike separate licensing deals with AI companies, and several have already done so. Independent artists typically have no bargaining power beyond accepting or refusing the take-it-or-leave-it terms a platform offers. If a court rules that those terms authorize AI training, independent creators will have contributed to commercial generative models without payment, without meaningful consent, and with no legal recourse.

The broader context here is worth considering: even if Google ultimately wins this argument in court, the political and legislative fallout will likely not follow the same path as the legal ruling. Lawmakers in Europe and, more slowly, in the United States have been watching AI training and copyright closely. A court decision that terms written in the mid-2010s implicitly cover 2026-era generative AI would almost certainly push legislatures to write new, explicit rules — as has happened in previous cycles when courts interpreted older laws in ways that left creators unprotected.

What Comes Next

Google's dismissal motion is a threshold move: the company is asking the court to throw out the case without examining the facts, based solely on the theory that the terms of service provide legal cover. If the court rejects this motion, the case moves forward into discovery — the phase where Google's lawyers would be forced to provide detailed records of how Lyria 3 was actually trained. Which audio files were used, in what volumes, and under what internal authorization process would become central to the dispute. That factual record, if it emerges, will inform not just this case but the broader landscape of AI copyright litigation.

For the independent musicians who brought this suit, surviving the dismissal motion is an early and significant hurdle. Getting past it keeps the case alive and preserves the possibility of investigating how Google actually developed the model. Losing on terms-of-service grounds would establish a powerful precedent and close off that line of inquiry.

The next meaningful development will come when the plaintiffs file their response to Google's motion and when the court rules. Both deserve attention.