Another day another Suno lawsuit, but the latest legal filing against the music AI company is very different to everything that has gone before, because all the legal claims in it relate to the publicity rights of musicians under US law rather than copyright. 

Filed by a group of musicians led by Jason Isbell, the new class action lawsuit claims that Suno built and trained its AI model to “index musicians by name”, so that it could generate new music “in the style” of existing musicians, exploiting each musician’s “identifying attributes”. 

As a result, Isbel’s lawsuit claims, Suno has “encoded musicians’ identities into its AI model to create its commercial product without consent”, and now “exploits those identities by publicly using musicians’ names to Suno’s considerable commercial benefit”. 

The various previous lawsuits filed against Suno by record labels, music publishers and musicians primarily make copyright infringement claims. They accuse the AI company of making unlicensed copies of recordings unlawfully stream-ripped off platforms like YouTube in order to build a training dataset. 

In response, Suno argues AI training constitutes a ‘fair use’ of existing music under US copyright law, meaning it didn’t need licences from the music industry to make those copies. This new lawsuit is very different and will require a very different response from Suno’s lawyers. 

First, they will need to respond to the allegation that Suno’s training processes take metadata associated with any one recording in order to “pair a musician’s name with recordings bearing that musician’s identifying vocal and instrumental characteristics”, so that Suno’s model can “associate a specific name with a specific set of identifying attributes”. 

And second they will have to address the claims that those training processes violate the publicity rights of musicians that sit within state law in the US. Publicity rights generally give people legal control over the commercial use of their name, image, likeness and voice. 

The lawsuit accuses Suno of violating relevant laws in California, Florida, Georgia, Hawaii, Illinois, Massachusetts, Michigan, New Jersey, New York, Ohio, Pennsylvania, Puerto Rico, Tennessee, Texas and Washington. 

The lawsuit is also interesting in the context of the licensing deals that some music companies – like Warner Music and BMG – have now agreed with Suno. 

Musicians often don’t own the copyright in their recordings. That has prompted much debate as to whether the likes of Warner and BMG can opt tracks they control into their AI licensing deals without getting the specific consent of each artist. 

But publicity rights belong to the musicians, and if those publicity rights have been exploited in the way this lawsuit describes, through the use of each artist’s name and “identifying attributes”, then explicit consent from artists will be required – directly or via their labels – in addition to any permissions the labels can unilaterally provide in relation to copyright. 

Suno will almost certainly begin by arguing that the new lawsuit misrepresents how its AI is trained, that artist names are not retained as part of its training processes, and that its platform has restrictions in place to stop users from requesting tracks be generated “in the style” of specific artists. 

Indeed, a Suno spokesperson has already put out a statement insisting that “Suno exists to help people create new, original music, not to trade on anyone’s name”, and that  “we stand by the many protections we have put into place across the platform, including blocking prompts for specific artists’ names or copyrighted songs”. 

Isbell’s lawsuit acknowledges the “protections” that Suno’s spokesperson wanted to highlight, but basically says they aren’t fit for purpose. And that the fact you can circumvent the restrictions and get Suno to basically output tracks in the style of existing musicians, including those involved in this lawsuit, proves the AI company is exploiting the names and identifying attributes of those musicians. 

According to the lawsuit, “Suno claims users are blocked from using a specific name as a prompt, but that claim is false. In reality, Suno not only accepts names but also responds to the name with an output based on that musician’s identity. When ‘Jason Isbell’ was entered, Suno produced an Americana song imitating Isbell’s characteristic clear male vocals and country twang”. 

In theory the Suno platform does try to stop users from prompting with an artist’s name, but – the lawsuit claims – the company’s “purported filter is easily defeated by inserting a space between each letter of a name – an alteration a human reader and Suno’s own underlying model still recognises as referring to the same person”. 

And not only is it relatively easy to circumvent Suno’s supposed filters, the lawsuit claims, the AI company has also allegedly partnered with content creators who explain how to get round the restrictions in YouTube tutorials. 

Plus, the lawsuit goes on, even when the Suno filter does work – removing a user’s artist name prompt – the platform then suggests “similar styles” that can be used to prompt the AI instead. Which, the lawsuit reckons, shows that Suno did in fact “identify the musician” from the name prompt, and used information connected to that name to offer alternative style-based prompts. 

We await to see how Suno responds to each of the specific allegations regarding the limitations of its filters. And to the assumptions made in the lawsuit that are based on outputs that have been generated with artist name prompts, which were accepted by the Suno platform in one form or another. 

Whether or not the publicity right claims made here are functionally or legally sound remains to be seen, but these are novel arguments making this a particularly interesting music AI legal battle to follow. 

​