Google cites YouTube terms in AI music copyright lawsuit

MUMBAI: The legal battle may be about AI-generated music, but Google’s defence is playing from an older score, the fine print. Google is leaning on YouTube’s terms of service in its fight against a copyright lawsuit that accuses the company of using artists’ music without permission to train Lyria 3, its artificial intelligence music model.

The case was brought by a group of independent artists, songwriters and producers who claim Google used music uploaded to YouTube to train AI systems without securing consent or offering compensation.

In a recent court filing, Google argued that creators who upload content to YouTube grant the platform a worldwide, non-exclusive, royalty-free, transferable and sublicensable licence to use that material. According to the company, those rights extend to activities such as reproducing, distributing, displaying and creating derivative works connected to YouTube’s services and broader business operations.

Google’s central argument is straightforward: even if the allegations are assumed to be true, the claims should fail because creators had already authorised YouTube and its parent company to use the content under the platform’s terms.

Notably, the company stopped short of confirming whether the artists’ music was actually used to train Lyria 3. Instead, Google argued that the question is largely beside the point because the licence granted through YouTube would have permitted such use if it occurred.

The filing describes the lawsuit as resting on an unproven assumption that Google’s AI systems were trained on the plaintiffs’ specific works. At the same time, the company maintained that the contractual rights provided by users would cover the conduct being challenged.

Lyria 3, introduced through Google’s Gemini ecosystem earlier this year, allows users to generate songs and music clips using text prompts. Google has previously said the model was trained using content it was entitled to access through platform agreements, partner arrangements and applicable laws.

The defence sets Google apart from several other AI companies facing similar copyright disputes. Firms including Suno, Udio and Anthropic have largely relied on the argument that AI training qualifies as “fair use” because the resulting systems are transformative.

Google, however, is taking a different route by focusing on contractual permissions rather than copyright exceptions. Legal observers note that if courts accept this interpretation, the case could reshape how digital platforms use user-generated content in the development of future AI models.

The dispute centres primarily on music uploaded directly by creators. Major record labels and publishers often operate under separate licensing agreements with YouTube, many of which contain their own provisions governing AI-related usage.

The artists have also argued that Google’s ownership of YouTube and access to its Content ID infrastructure gave the company a unique advantage in accessing and analysing music at scale, making the case distinct from other AI copyright battles currently working their way through courts.

As generative AI continues to blur the lines between creativity, ownership and technology, this case could help determine whether the next chapter of AI development is written by copyright law, fair use or the terms and conditions most users never read.