There’s a particular kind of corporate whiplash that happens when a company’s PR arm and its legal team appear to be reading from different scripts. Google just supplied a textbook case, and independent musicians and their copyrights are the ones caught in the middle.
On June 18, as the Senate Judiciary Committee prepared to vote on the NO FAKES Act, YouTube’s global head of music, Lyor Cohen, and Recording Academy CEO Harvey Mason Jr. published a joint op-ed making the case for legislation that would let individuals authorise or block AI-generated digital replicas of their voice and likeness.
It was framed as an urgent, artist-first stand: AI is moving fast, and the law needs to catch up before creators lose control of their own identities.
Ten days earlier, Google’s own lawyers had filed something that reads like the opposite argument.
The case centres on Lyria 3, the AI music generator Google folded into Gemini in February 2026, capable of producing full songs, vocals, lyrics and all, from a text or image prompt.
A group of independent artists, including singer-songwriter Sam Kogon, producer Magnus Fiennes and the Chicago band Directrix, filed a proposed class-action lawsuit in March alleging their music was ripped from YouTube and used to train Lyria 3 without compensation.
Google’s response, filed on June 8 by the litigation firm Quinn Emanuel, skips the argument nearly every other AI company has used. Suno, Udio and even Anthropic have leaned on fair use, the idea that training a model is a “transformative” act protected under copyright law.
Google didn’t bother. Instead, its lawyers argued that the plaintiffs had already granted YouTube, and Google by extension, a broad licence to use their uploaded content the moment they hit “upload”. No fair-use debate required, just a contract, allegedly signed the moment anyone agreed to YouTube’s terms of service.
It’s a clever pivot. As music litigator Ken Anderson told Billboard, turning to a contract argument rather than fair use is a shrewd move, since contracts carry weight that fair-use claims often don’t in American courts. But it’s also a much colder one; it doesn’t ask whether using an artist’s song to train a commercial AI model is fair.
It simply asserts that the artist already agreed to it, buried in terms of service that almost nobody reads in full.
Two messages, one week, zero coincidence?
The optics are hard to ignore.
Cohen’s op-ed argues that artists deserve the right to authorise or block the use of their voice and likeness in the AI era. Meanwhile, his own employer’s legal team is arguing in federal court that artists already handed over that authority passively and automatically, without ever being asked a direct question about AI.
Artist advocates have noticed. Ian Harrison, CEO of the American Association of Independent Music, told Billboard that the use of anyone’s creative work should require clear consent, transparent terms and shared value and that Google’s position runs opposite to that.
It's a fair point: there's a meaningful difference between “we support guardrails for AI” as a policy stance and “we already had the rights we needed” as a litigation defence.
Why the copyright and contract argument matters more than it looks
This isn’t just a music-industry squabble but a preview of how Big Tech may defend AI training claims going forward.
Fair use arguments invite courts to weigh transformation, market harm and creative intent, a messy, unpredictable terrain that’s produced mixed rulings across the AI copyright landscape. A terms-of-service argument sidesteps all of that. If a platform can point to a clause buried in its user agreement and say, “This covers it”, the entire fair-use battle becomes irrelevant for that platform, at least.
The catch, according to Anderson, is that adhesion contracts, the kind users have no real ability to negotiate, can theoretically be voided by courts if judged unconscionable, though he notes that this legal outcome succeeds rarely, even when the facts appear to support it.
For now, the case sits in the U.S. District Court for the Northern District of Illinois, with opposition papers still to come.
But the underlying tension isn’t going away: a company that owns both the world’s biggest video platform and one of its most ambitious AI labs is discovering that it’s very hard to advocate for creator consent on paper while arguing in court that consent was already baked into a checkbox nobody read.
Also read: YouTube paid out $8 billion to the global music industry in 2025, the best since 2022
Technext Newsletter
Get the best of Africa’s daily tech to your inbox – first thing every morning.
Join the community now!
