Artists Say Lyria 3 Model Stole YouTube Songs

A group of independent artists, songwriters and producers has brought a new copyright infringement lawsuit accusing Google of training its Lyria 3 artificial intelligence model on unlicensed music pulled from YouTube.

The complaint, filed on Friday (March 6) and reviewed by Billboard, follows a series of copyright cases brought by music rightsholders against AI song generators Suno and Udio over the past two years. This is the first such lawsuit to go after Google, which launched Lyria 3 through the Gemini app last month, allowing users to generate up to 30-second AI songs based on text and image prompts.

Google’s entry into the AI music market differs from existing services because it already owns YouTube, home to millions of songs uploaded by artists and labels. Friday’s lawsuit alleges Google used this position “to pivot from distributor to competitor,” supposedly extracting audio elements from YouTube videos and feeding them into the Lyria training set without paying artists.

“Google had every opportunity to develop this product legally,” reads Friday’s lawsuit. “It owns YouTube and runs Content ID. It has long-standing relationships with major labels and distributors. It has the technical infrastructure, financial resources and industry connections to clear rights before training. Google chose not to do so, not because licensing was impossible, but because copying was faster and cheaper.”

Reps for Google did not immediately return a request for comment on the lawsuit on Monday (March 9). In a statement to Billboard last month, the company said Lyria 3 is mindful of copyright concerns and only trains on music that YouTube and Google have “a right to use under our terms of service, partner agreements and applicable law.”

The lawsuit was filed by an all-indie group of music creators: singer/songwriter Sam Kogon, composer/producer Magnus Fiennes, songwriter/producer Michael Mell, R&B group Attack the Sound, father-and-son folk rock duo Stan Burjek and James Burjek, and the Chicago-based band Directrix.

All of those artists publish music on YouTube and therefore allege their work was likely included in the Lyria 3 training dataset. They’re bringing copyright infringement claims against Google as to both publishing and recorded music rights, plus various other intellectual property, privacy and consumer protection claims.

The case is a proposed class action, meaning the plaintiffs want relief for all indie artists who have allegedly been harmed by Google’s conduct. They’re seeking financial damages and noting that music produced by Lyria 3 is taking real work, such as sync licensing opportunities, away from human beings.

“Google markets these outputs for the same uses, and to the same buyers, as the music plaintiffs create and license,” reads the lawsuit. “The result is not incidental competition from a tool with a different purpose. It is direct, designed-in market substitution at a scale that individual human creators cannot match.”

The same group of artists previously brought similar copyright lawsuits against Suno, Udio and AI music generator Mureka. Another indie artist, Tony Justice, also has pending class action lawsuits against both Suno and Udio.

The more closely watched front in the AI music copyright war, however, is litigation brought by the three major label groups. Back in 2024, Universal Music Group (UMG), Warner Music Group (WMG) and Sony Music teamed up to sue both Suno and Udio for allegedly training on stolen work “at an almost unimaginable scale.”

Udio settled with both UMG and WMG near the end of 2025, striking landmark licensing deals that promise to compensate rightsholders and keep songs generated on the service within a so-called “walled garden.” WMG then signed a settlement with Suno, which allowed licensed AI music to be downloaded and distributed on streaming.

UMG is continuing to hold out in its fight against Suno, and Sony has not settled either case. For the claims that remain, Suno and Udio are mounting a defense that their AI training was permissible under the principle of “fair use” — a tenet of copyright law that allows unlicensed work to be used in “transformative” fashion.