Photo Credit: Adarsh Chauhan

Do YouTube’s terms of service grant Google “a broad license” to train AI on uploaded music? The tech giant believes so, and it’s doubling down on the position as it aggressively pushes to dismiss an artist-led copyright lawsuit.

That class action suit set sail in March and counts as plaintiffs the same individuals (including members of Chicago-based bands Attack the Sound and Directrix) who are suing Udio as well as Suno for alleged infringement. (One plaintiff, singer-songwriter Sam Kogon, is a party to the Google litigation but not the older cases.)

Clocking in at a healthy 118 pages, the complaint also encompasses non-copyright claims (and corresponding proposed sub-classes); more on this in a moment. But at the top level, the plaintiffs maintain that Google used their releases to train AI models like Lyria without authorization.

Unsurprisingly, Google is vehemently denying the allegations, including in a newly submitted dismissal motion that itself spans 41 pages.

Overall, the filing’s most interesting component concerns YouTube’s terms – especially given the straightforward argument’s implications for separate-but-similar actions.

The way the defendant sees things, the fine print that the plaintiffs accepted when uploading their music “grants Google a broad license to use” the relevant “content in connection with Google’s business.”

And according to Google, said license covers, among much else, AI training. Of course, if the court rules in favor of the YouTube parent here, it’d be a major setback for the litigating artists.

Nevertheless, the dismissal document aims to refute distinct arguments as well. First, some of the plaintiffs are allegedly seeking relief for works that were unregistered with the Copyright Office when the suit kicked off. And the Supreme Court’s far-reaching Cox v. Sony decision is a key focus in Google’s attempt to toss a contributory infringement claim.

The action “contains no allegations that Google intentionally induced users to generate infringing songs, as necessary to establish contributory copyright infringement under recent Supreme Court authority,” a related line reads.

On the vicarious side, the plaintiffs’ claim “fails out of the gate for the same reasons as their contributory infringement claim,” referring in part to the purported absence of “any specific direct infringements,” per Google.

What about Google/YouTube’s alleged violation of the Digital Millennium Copyright Act?

The DMCA is front and center in multiple AI lawsuits now. In this case, the plaintiffs are taking issue with the alleged removal of copyright management information from training data and the alleged circumvention of third-party platforms’ access controls.

At present, the debate about precisely what constitutes an access versus a copy control is ongoing. In any event, the motion calls out the complaint’s alleged lack of CMI specifics and criticizes the access control allegation as making “no sense.”

“Google would not need to circumvent any access measure, because Plaintiffs voluntarily gave Google access to the works when they chose to upload their content to YouTube,” the text reads.

Finally, Google is also pushing back against several state-law claims. But as mentioned, the terms argument is particularly important – including because DIY distributors are compelling artists to grant expanded derivative-work permissions when making their music available via platforms like Spotify.