Is the most popular song played on Australian radio stations the product of generative AI?

Josh Fawaz’s song, a cover of Like a Prayer, has raised questions about how generative AI is being used in music and whether it should be declared

An Australian producer has gone from little-known artist to viral sensation in a matter of months, with his hit song catapulting on to global charts and receiving thousands of radio spins.

There’s just one problem: music experts and other musicians are questioning whether he produced it. They claim Josh Fawaz’s most popular song, a cover of Madonna’s Like a Prayer which reached the No 1 spot on the National Radio Airplay chart, could have been made using AI.

While producers often use software including Ableton Live or pitch-shifting tools in their work, or use artificial intelligence to mix, master or otherwise enhance tracks, in an AI-created song, generative AI is the creator, requiring nothing from a human but a text prompt.

The song credits on Like a Prayer – and all his tracks – list Fawaz as the “performer”, meaning the vocalist; and his uncle Fadi Fawaz (best known as George Michael’s former partner) on synths and production.

But Sam Whiting, a senior research fellow at RMIT’s school of media and communication, and others, say Like a Prayer has hallmarks of AI music generators like Suno, such as being “heavily compressed”.

“This is a very ... impressive vocal performance if it was delivered by a human but if it’s not, that brings in really worrying questions around what we value any more in terms of human expression,” Whiting said.

Breakthrough hits

“I use AI as a tool,” Fawaz said on Instagram under a post criticising his work. “What I care about [is] providing my listeners with good music.”

Fawaz began releasing music in the 2010s. But it wasn’t until he pivoted to covers with vocals this year that he gained commercial success, first with Like a Prayer in April and months later with his debut 18-track album, Dance Like Nobody’s Watching – filled with heady, club-friendly reimaginings of hits including Oasis’s Wonderwall and Cyndi Lauper’s Girls Wanna Have Fun.

Since its release, Fawaz’s version of Like a Prayer has had 35m streams on Spotify and topped the iTunes Electronic chart worldwide. His album climbed to No 18 on the Aria Australian artist albums chart.

On 1 July, a new commercial radio code of practice came into effect requiring programs to be transparent about using AI-generated voices on air. But it doesn’t apply to music.

The success of Like a Prayer comes as big tech companies are asking for Australian copyright laws to be watered down, to allow them to scrape Australian output to improve their AI models, prompting alarm from musicians.

The large commercial radio networks ARN, Nova Entertainment and Southern Cross Austereo did not respond to requests for comment or provide their policies on AI use.

Late last month a producer and DJ called Needs No Sleep took to Instagram to describe AI-generated content as the “biggest issue in music right now”. He said he was speaking up because music produced by AI receives royalties each time it is streamed or played on the radio. Plus the work of artists, including Needs No Sleep, is used to train AI models .

“If this is the future of music production … then we really are cooked,” he said.

Neither Fawaz nor his management company, Hallwood, responded to questions from Guardian Australia. But he hit back at Needs No Sleep on Instagram, claiming that he used AI “as a tool”.

“It’s not that deep,” he wrote. “[I’ve] been releasing music way before AI was invented. What I care about [is] providing my listeners with good music. Maybe you should do the same.”

Needs No Sleep told Guardian Australia he became aware of Fawaz’s music when it landed in his inbox from a DJ promoter service.

“I’ve been doing it for 15 years and you can tell the way that AI music is produced – it has distinguishable features,” he said, pointing to “sloppy drums”, vocals with “artefacts” and a low-quality streaming file.

“Those royalties that are collected are diverted away from other artists making real music. Artists count on these royalties, and they’re now going to compete with AI music for a smaller and smaller slice of the pie.”

A spokesperson for the Australasian Performing Right Association and Australasian Mechanical Copyright Owners Society, who provide royalties to musicians, told the Guardian that Fawaz has been a member of both bodies since 2021, and how Like a Prayer was recorded wouldn’t impact the payment of royalties “to the original human owners” of the copyright.

“The song Like a Prayer is a remix/cover of a musical work written by Madonna L Ciccone and Patrick R Leonard,” they said. “As the original human rights holders of that musical work, they will be entitled to be paid all performance royalties in the usual manner.”

If Fawaz is using prompts to make his music, he’s not the first to do so successfully since AI music tools went mainstream. Last year three AI songs reached the highest spots on Spotify and Billboard charts – and some major labels are even striking deals with AI music companies.

A spokesperson for the Media, Entertainment and Arts Alliance said this was a “growing trend”.

The MEAA is advocating for equitable remuneration (ER) to be introduced into Australian copyright law so musicians and creatives are compensated when their work is used by AI companies, and for mandatory watermarking of AI-generated content.

“Massive datasets of music, films, journalism, books and voices are being scraped to train commercial AI systems,” the spokesperson said. “This is driving job losses, undercutting incomes, and hollowing out Australia’s creative and media industries.”

Whiting said the past decade of streaming culture and “TikTok-ification” of music had “primed” listeners for AI-produced sounds.

“The reason why AI music is not being more heavily interrogated is because streaming has conditioned us all not to engage with music in a critical, proactive way,” he said.

“And commercial radio in Australia has decided to heavily promote this track without any scrutiny of where it’s come from, who’s involved, or where those vocals were generated.”