• | 8:00 am

Sinatra sings Nirvana: Here’s why 200 music artists just signed an open letter critical of AI

Just as creators of many other major art forms have come out against AI, a group of heavy hitters in the music world, including Billie Eilish, are now sounding off. Here are just some of the AI creations proving their point.

Sinatra sings Nirvana: Here’s why 200 music artists just signed an open letter critical of AI
[Source photo: Sarah Morris/WireImage, Frank Micelotta/Getty Images, Votava/Imagno/Getty Images]

It was only a matter of time before musicians joined writers, actors, and filmmakers in banding together against the threat AI poses on their art and livelihood. On Tuesday, nonprofit advocacy group Artist Rights Alliance released an open letter demanding more responsible use of AI, and that artists receive fair compensation from it. The letter was signed by a who’s who of heavy hitters—from Billie Eilish to Billy Porter and Jon Batiste to Jon Bon Jovi.

Crucially, these artists are not calling for an end to the use of AI in music creation—only the kind that is used “to infringe upon and devalue the rights of human artists.” While companies like Universal Music Group have been working to find artist-centric solutions to the threat of AI—lobbying Congress for stronger regulation and suing an AI startup for training its model on copyrighted lyrics—it may be a while before such regulations become widely adopted. In the meantime, as a trip down the AI corridor of YouTube confirms, the genie is already out of the bottle. Song-generating tech is now readily available to anyone with a computer, some spare time, and a desire to make their music mash-up fantasies a reality. (Kind of.)The rise of AI’s copyright threat can be traced through the evolution of one particular fan’s remix of “Baby Shark.” In late 2020, a YouTuber who goes by the name, Mr. Grande, released a clip of himself rapping in the style of Nicki Minaj over a beat that samples the famous children’s tune. It was the kind of relatively harmless infringement the music industry has been combating in court since the late ‘80s and the war on sampling. Cut to last fall, however, and Mr. Grande could now make an AI-generated Nicki Minaj sing his song. Clearly, we’ve entered a new era.

That song remains up on YouTube seven months and 1.3 million views later, which may be part of the reason Minaj signed the open letter. She is joined by the estates of several artists who are no longer around to defend themselves, including Frank Sinatra and Bob Marley. While the remaining Beatles may have briefly resurrected John Lennon for a final “new” Beatles song last fall, Marley and Sinatra are regularly revived online without their estates having any say in the matter. In fact, thanks to a YouTuber named Ihor Palamarchuk, the two even collaborated recently—on an AI Marley cover of Sinatra’s “Fly Me to the Moon” with near-passable accuracy.

Ol’ Blue Eyes himself has been made to cover everyone from Michael Jackson to Nirvana lately. Perhaps it’s because the crooner has such a massive back catalog for AI models to study, but the Sinatra faux-vocals in particular are harder to clock as digital forgery. Helping matters not one bit, the arrangements backing his Nirvana cover sound completely legit—conjuring images of session musicians in the future losing out on work to simple AI programming.

Armchair music producers are doing more than just making The Beatles (badly) cover Nirvana, though; they’re also generating brand new songs.

Rather than settle for an AI Nicki Minaj karaoking his lyrics, someone who goes by the handle, Good Luck Chuck, created an entire “new” Nirvana song based on the band’s output. Is it any good? Absolutely not. The guitar riff sounds like it was based on someone who half-remembered the song “Polly” attempting to hum it, and the singing seems assembled, ransom note-style, from Kurt Cobain outtakes. But, could the song fool an only-somewhat-gullible fan into thinking it was newly unearthed from a studio vault? Almost certainly.

While the trend of making public figures like Angela Merkel cover Eminem’s “Lose Yourself” or President Biden and Donald Trump cover Kanye and Jay-Z’s “Ni**’s in Paris” seems like just-for-laughs infringement, these “new” songs appear intended for more than mere novelty. They are either supposed to fool people or give them something to jam out to in earnest.

Last spring, a TikTok user, Ghostwriter977, released an unauthorized, AI-generated collab between Drake and the Weeknd called, “Heart on My Sleeve.” It quickly collected more than 10 million views on TikTok and a quarter-million Spotify streams before the companies scrubbed it in response to copyright claims from the UMG label, Republic Records, which reps both artists. The only reason this song was booted, though, is because the level of attention it managed to capture on these platforms was unusual. The steady supply of less popular but still widely consumed copyright violators is extensive enough that labels are going to have to be on constant whack-a-mole patrol to keep up with it.

Perhaps worst of all, judging from the YouTube comments, a lot of people are ready to fully embrace similar synthetic songs.

“I needed this vibe in my life,” one commenter wrote about AI Bob Marley’s cover of Nirvana’s “Come As You Are.”

In response to an “unreleased” AI Billie Eilish track called “Emotion,” the top comment says: “Billie definitely needs to hear that song and record it for real.”

The commenter is right that Eilish definitely needs to hear it, but only so she can pass it along to her lawyer. Perhaps by signing the open letter, the newly minted two-time Oscar-winner can help ensure she won’t even need to do that in the future to prevent anyone from degrading her work.

  Be in the Know. Subscribe to our Newsletters.

ABOUT THE AUTHOR

Joe Berkowitz is an opinion columnist at Fast Company. His latest book, American Cheese: An Indulgent Odyssey Through the Artisan Cheese World, is available from Harper Perennial. More

More Top Stories:

FROM OUR PARTNERS

Brands That Matter
Brands That Matter