Suing Suno and Anthropic together signals the real fight: licensing

Why would a music publisher sue a chatbot maker and an AI song generator on the same day, in the same court? Because the publisher sees little difference between them. In our view, that is the most interesting thing about Reservoir's new lawsuits, and it should change how the music business plays this fight.

Two defendants, one theory: the training data is the product

According to Billboard, Reservoir filed a pair of federal cases on October 21, accusing Suno and Anthropic of infringing songs by Madonna, Johnny Cash and Joni Mitchell to train their models. Music Business Worldwide reports that the Suno complaint covers 1,050 compositions and puts statutory damages at $157.5 million for those works, just over $186 million once the DMCA counts are added.

Reservoir isn't alone. Round Hill filed similar suits against both companies in August. The pattern is clear: independent publishers, represented by the same lawyer, are treating "AI that makes songs" and "AI that does everything" as the same legal problem.

The song generator was the easy villain; the chatbot is the real test

Suno was always the tidy target. It sells a product that outputs songs, and Reservoir argues it was built to substitute for the artists it copied. That story is intuitive for a jury.

Anthropic is harder, and more consequential. At a hearing on October 21, Reuters reported, Anthropic argued that training on millions of lyrics is transformative fair use, while the judge pressed both sides and also questioned whether Claude's lyric writing threatens the publishers' market. Two earlier Northern California rulings split on the broader question. We think a general-purpose lab has a much better "this isn't a music product" defense than Suno does, and that is exactly why publishers want to pursue both: a loss against the chatbot would hurt more than a win against the startup would help.

Still, Reservoir's complaint contains the strongest argument for the industry's side, and it isn't about outrage. It points to Suno's licensing deals with Warner Music Group, BMG and Believe as evidence that a functioning market for training data already exists. Fair use leans heavily on market harm. Every deal signed is an Exhibit A for the next plaintiff.

Litigation is leverage, not a destination

Here is where we part ways with the maximalists. Round Hill's CEO says he intends to take these cases to trial and won't accept a resolution that leaves songwriters short. Principled, and probably useful as a negotiating posture. But as a strategy, years of case-by-case rulings across judges who have already disagreed is a gamble with the whole catalog as the stake. One adverse fair use ruling could undercut every publisher at once, including those that never sued.

The Warner settlement and the BMG and Believe deals show the alternative: disputes that can end in licensing agreements. Even then the picture is messy, since UMG and Sony are still suing Suno and say its new model remains illegal. And note the irony Anthropic's lawyer reached for in court, saying UMG itself has called AI "overwhelmingly a net positive." The industry's own mixed messages weaken its hand.

What we'd push for

The winners from a prolonged stalemate are the labs, which keep building while courts deliberate. Suno's strategy, as Reservoir alleges it, is "infringe now" and "settle on the cheap later." The only answer to that is to make settling cheap a bad bet: keep the suits alive, keep the damages claims credible, and demand licensing terms that cover training and outputs, with real money reaching songwriters rather than just the largest rightsholders.

Our verdict: sue both, but sue to get to the table. Publishers who confuse a courtroom win with a business model risk getting a ruling and no royalties. The smart play is a licensing market for general AI labs as well as music startups, and Reservoir just made that harder for Anthropic to ignore.