Frank Sinatra, Kurt Cobain also getting screwed by AI, new study finds


Chart-topping music artists who have long been dead and buried – including Frank Sinatra, Freddie Mercury, and Kurt Cobain – are also being taken advantage of by AI, a new study reveals. This is as Universal announces plans for the world’s first AI-generative large music model.

Using unlicensed works to train AI models without an artist's explicit permission or compensation has garnered the ire of musicians, producers, and record labels worldwide since generative AI has taken over.

But it’s not just living creatives that are losing out on music royalties – the use of deceased artists' voices has recently become another bone of contention.

ADVERTISEMENT

According to a new study released Friday by musicMagpie, a British-based online retailer of refurbished electronics and physical media, the issue only highlights the "growing impact AI is having on the music industry."

MusicMagpie says there are an estimated 1.63 million AI covers currently on YouTube alone, raising concerns about the “authenticity of music, whether human creativity is important, and the financial implications on artists.”

The "Bop or Bot?” study shows that legacy voices – meaning those of artists no longer living – make up a whopping 40% of the top 15 streamed artists plagiarized by AI.

The top three most AI-plagiarized legacy voices are, at present, Frank Sinatra, Freddie Mercury, and Kurt Cobain, with Michael Jackson, Elvis Presley, Hank Williams, Johnny Cash, Prince, The Beatles, and the popular Indian musician Kishore Kumar, all making the top 10 list of the most streamed AI legacy artists.

To crunch the numbers, the study ironically used AI to scrape YouTube, analyze the results, and compare them with ‘cost to stream’ plays on Spotify.

Top 10 streamed AI legacy artists
Top 10 streamed AI legacy artists. Image by musicMagpie.

Furthermore, the study also found that nearly 50% of Brits and more than 65% of American fans were unable to distinguish between songs created by human musicians from those created by their artificially intelligent counterparts.

The findings

ADVERTISEMENT

For the 1.63 million AI-generated covers on YouTube, each has a median average of about 2,100 views each, musicMagpie said.

Additionally, out of the 1,500 AI-generated songs and covers analyzed in the musicMagpie study, the music of legacy artists alone counted for “a mind-blowing 19 million views.”

From a legal standpoint, the duration of a copyrighted work, known as an Artist's Resale Right (ARR), lasts for 70 years from first publication, or 50 years for a recorded live performance, according to Julia Lowe, Head of Dispute Resolution at UK-based law firm Higgs LLP.

Lowe further stated that because non-living artists "are unable to consent to use... their legacy and reputation may be damaged in the eyes of the listening public, which is problematic.”

After converting the number of views to Spotify plays ( $0.004 per stream), the eCommerce site calculated the total revenue loss for legacy artists at more than $13.5 million.

Of the top 15 most-streamed AI-recreated artists, more than a third were deceased artists, the study showed including Jackson, Presley, Queen's Mercury, and Nirvana's Cobain – which also makes them the top artists losing out on revenue due to AI.

The staggering amount of lost revenue from unauthorized AI covers per deceased artist is costing their estates tens to hundreds of thousands of dollars, which could have a major financial impact on estates that rely on royalty payments.

AI legacy artists revenue lost
Potential Lost Revenue to AI-created songs and covers of both living and legacy artists. Image by musicMagpie.

"On one hand, AI-generated music of those who have passed could be viewed as continuing an artist’s legacy, allowing fans to reimagine their voices in new and imaginative ways," Jon Miller, musicMagpie’s Chief Commercial Officer pointed out.

“However, this approach is controversial. The use of AI to replicate deceased artists raises ethical concerns about consent and authenticity, “ Miller said.

ADVERTISEMENT

AI’s first Large Music Model in development

The study happens to coincide with a new “strategic collaboration” between Universal Music Group and the Los Angeles-based AI music company KLAY Vision, announced on Monday, aimed at ushering in a new era of innovation and protecting the rights of artists.

The two companies are working on the development of a state-of-the-art Large Music Model (KLayMM) for “AI-generated music that works in collaboration with the music industry and its creators,” they said.

Michael Nash, executive vice president and chief digital officer at Universal Music, said the project will “explore new opportunities and ethical solutions for artists and the wider music ecosystem.”

The goal: to ethically build a foundational AI music model respectful of copyright, name, and likeness rights, dramatically lessening the threat to human creators by providing new opportunities for copyright monetization, while still having “the potential to transform human creativity” Nash said.

Top 15 most streamed AI artists
Top 15 most streamed AI music artists, living and legacy. Image by magicMagpie.

Meantime, Universal is currently in a legal battle with Anthropic AI, Suno, and Udio over the use of the label's recordings to train music-generating AI systems.

Last week, Universal, along with Sony, and Warner Music Groups, signed an open letter with a growing list of over 28,500 musicians, composers, songwriters, and other creatives, protesting the “unlicensed use of creative works” to train generative AI.

The latest effort – and the second open letter of its kind – was spearheaded by Abba’s Björn Ulvaeus, The Cure’s Robert Smith, and Radiohead’s Thom Yorke.

The first Artists Rights Alliance petition was crafted in April, and signed by Robert Smith, Stevie Wonder, Billie Eilish, Nicki Minaj, R.E.M., Peter Frampton, Jon Batiste, Katy Perry, Sheryl Crow, Smokey Robinson, as well as the estates of Bob Marley and Frank Sinatra.

ADVERTISEMENT

The ethical commercial AI model is expected to be launched in the coming months.