Generative AI concerns focus on Copyright, but critical legal questions linger

AI, whether used to create songs with limited human intervention or as an aid to songwriting, is a cause of great concern for many in the music industry. Beyond the often-discussed copyright issues are some vexing legal questions

A version of this essay first appeared in Bill Werde’s free, weekly Full Rate No Cap email. Werde is a former Billboard Editorial Director and Director of The Bandier Program for Music and the Entertainment Industries of the Newhouse School of Public Communications at Syracuse University.

by Lewis D. Sorokin, Esq., Entertainment Attorney, Associate at Law Office of Simon Rosen

Gamer Presidents. AI cover songs. “New” music from beloved, deceased artists. Emin-AI-em.

The speed of development in AI is staggering, but existing law did not – could not  – anticipate these new forms and approaches to music. So while fans are having a blast creating and consuming this new content, legal frameworks and potentially critical business models are about to be tested like never before. Much of this discussion has understandably centered on copyright law, But I would argue that another area of intellectual property – rights of publicity – is just as critical.

For the uninitiated, rights of publicity, sometimes called personality rights, are the rights of an individual to control the commercial use of their identity such as their name, image, likeness, and other unequivocal identifiers. One of the least-discussed forms of intellectual property (compared to its cousins copyright, trademark, and patent), rights of publicity sit at the heart of some of the most prominent uses of AI in music today. Case in point? The music industry is keeping close tabs on a class action that was filed by a group of artists against the developers of Stability AI, DeviantArt, and Midjourney. While the focus of the suit is copyright infringement, the plaintiffs also allege violation of the California statutory right of publicity, among other claims. The artist plaintiffs claim that, by advertising that these generative AI programs can create “art ‘in the style’ of Plaintiffs’ work,” the defendant software companies “generated valuable business” off of the artists’ names, without compensation.

Consider a few recent or possible music scenarios:

There’s a lot of talk in the music industry about the forthcoming sale of Michael Jackson’s catalog. What if that sale were to include the exclusive legal right to use his AI-generated voice in future works? How would that affect the valuation of the catalog, already expected to approach $1 Billion? 

David Guetta has provided probably the highest-profile single example to date of a scenario that implicates both AI and rights of publicity (albeit not with the voice of a deceased artist). In February, Guetta told a generative AI website to write lyrics in the style of Eminem talking about “future rave music,” and then another AI tool to generate the lyrics in Eminem’s voice. Guetta used this vocal to create a track suitable for a live DJ set which he recorded to share his “Emin-AI-em” concept with the world on social media — of course, not without the (potentially lawyer-appeasing) disclaimer “obviously I won’t release this commercially.”

Let me introduce you to… Emin-AI-em 👀 pic.twitter.com/48prbMIBtv

— David Guetta (@davidguetta) February 3, 2023

To that end, what happens when some company or brand inevitably follows Guetta’s example and uses AI to recreate a famous voice for use in a commercial without the consent of that voice’s “owner”? One court answered a similar question involving a soundalike of Tom Waits, so there is some amount of precedent in certain jurisdictions. In that case, Waits, famous for his voice, filed suit against companies using soundalikes in marketing campaigns. He prevailed in his lawsuit against Frito-Lay, demonstrating that a strong right of publicity is effective at defending individuals’ rights.

Tom Waits for No Doritos. Click through for audio from the offending commercial.

A similar issue is being litigated right now, wherein Rick Astley is suing Yung Gravy over the soundalike voice used in Gravy’s 2022 smash hit “Betty (Get Money)” to replicate Astley’s timeless classic meme hit “Never Gonna Give You Up.” But even here, there was some other human voice that replicated Astley’s. What if that voice were a synthetic recreation made by an AI trained on Astley’s voice? Would the law treat that differently from a human voice? Food for thought.

The courts may decide if Gravy rocked Rick inappropriately.

At the heart of all of these examples and hypotheticals are rights of publicity, which are, as a legal right, recognized inconsistently from state to state, if recognized at all. It is difficult to say that the current state of these rights is sufficient to address Web 3.0 issues, when the rights aren’t even clear in yesterday’s Web 2.0 world. For example, the laws of the two foremost entertainment industry centers in the United States, New York and California, have fundamentally different approaches to their rights of publicity, such that somebody living and working in New York has much weaker protection over their rights of publicity than someone in California. Not only that, but New York does not recognize post-mortem rights of publicity and limits the right to the lifetime of the person, whereas Oklahoma, for example, allows a 100 year post-mortem right. Nonetheless, both a New Yorker and a Californian have stronger rights than someone in Delaware, for example, where no personality right is recognized at all. 

As this technology becomes increasingly accessible to the everyday person, it seems abundantly clear that Congress should enact some federal statutory right of publicity akin to the federal statutes recognizing every other form of intellectual property. Calls for a federal statutory right of publicity date back as early as 1998 when the American Bar Association drafted a proposal to Congress on this subject at the request of the International Trademark Association. The time has come to revive this proposal, which seeks to harmonize the “patchwork” of inconsistencies across the states by inventing a uniform federal right of publicity. Central tenets of the proposal include allowing for the rights to be transferable and descendible for a fixed term after death, allowing a “grandfather clause” for rights recognized prior to the enactment of the federal right, and certain other legal considerations. This would not only unify this right nationwide, but would also lay a foundation for the issues that lie ahead where rights of publicity and AI intersect. Rather than a new set of regulations specific to AI, this federal right of publicity would extend across all media and technologies in order to protect individuals’ rights.

Let’s reconsider our earlier examples under this new framework. If Michael Jackson’s catalog sale were to include the right to use his voice in future AI-generated works (notwithstanding the almost certain pushback from some number of fans), what happens if some third-party uses his voice to make unauthorized AI-generated songs in a state where no rights of publicity are recognized? Can the new owner of the lawful right file suit, or is the infringer free because his state does not recognize the right imbued by the catalog sale? The answer is unclear and unknown under the existing legal frameworks, but a federal right of publicity would give direction.

Or, consider that Delaware is the corporate home to countless businesses big and small. Suppose one of those businesses were to commercially exploit “Emin-AI-em” or an AI soundalike of another famous voice for marketing purposes (like in the Tom Waits lawsuit). It is unlikely that the true owner of that voice would prevail in a lawsuit under Delaware’s existing state laws. But, a federal statutory right of publicity that protects an individual’s voice as a clear marker of their identity would almost invariably lead to a more just conclusion in such a dispute. 

To that end, it is up for debate specifically which state or states’ laws an ideal federal statute should be modeled after. How that issue would shake out in Congress remains to be seen. But the need for that conversation is imminent.