By all accounts, the massive popularity of Josh Fawaz's remix of a classic Madonna single should be an Australian success story. Instead, it's become wrapped up in allegations of AI-generation and inspired broader questions surrounding authenticity and the rights of artists in the modern age.
Originally released on April 14th, Fawaz shared his version of Madonna's 1989 single Like A Prayer, having offered a reworking of the track which pairs thumping beats and bass drops with passionate female vocals.
All wrapped up in just over two minutes, this blend of a remix and a cover felt custom-built for the TikTok generation, resulting in success for the song which was both immediate and noticeable – currently boasting more than 37 million streams on Spotify alone.
Over on social media, Fawaz himself couldn't believe the attention that the track had received. "Believe me when I say I was about to give this music stuff up 2 months ago," Fawaz wrote in a video shared in early June. "Posting my songs online for years, and nothing really happened."
Don't miss a beat with our FREE daily newsletter
"Told myself I'll give it one last try, then I'll quit," he added. "On the verge of giving up I posted Like A Prayer and went to sleep that night. But I woke up and my life was changed forever."
In May, the track impacted Australian radio in a big way, debuting on the Radio Monitor Hot 100 at No. 26 – just one week after the original version of the song had itself re-entered the chart at No. 98. In June, the song topped the chart, becoming the most-played song on Australian radio for the preceding week.
Most recently, the song was sitting at No. 2 on the Radio Monitor charts last week with 2,745 spins across metro and regional Australian radio. From Monday to Friday last week, between 6am and 6pm, that single track accounted for 36% of all Australian spins on KIIS Sydney, 25% on 2Day FM, and 22% on Nova Sydney.
On paper, Fawaz's side of things feels like a triumph in which a musician, disillusioned with the industry, decided to have one last shot, only to receive massive success.
However, suspicions and allegations soon arose surrounding the notion that the track is AI-generated. Australian producer Mitch Thomas otherwise known as Needs No Sleep, took to social media to share a video in which he labelled Fawaz "the biggest issue in music right now" and alleging that his music "is generated entirely with AI".
Thomas continued by claiming that Fawaz "types a prompt, and then the track is made itself", citing a previous song released by Fawaz as ostensible proof that it's "not the same producer at all, not even remotely close".
Fawaz responded in the comments of Thomas' video, telling him to "stop having a little sook" and admitting that he uses "AI as a tool" in pursuit of "providing my listeners with good music", but did not specifically address claims that his rework of Like A Prayer is AI-generated.
Viewed on streaming services like Spotify, Fawaz's rework of Like A Prayer credits songwriters Madonna and Patrick Leonard as lyricists on the track, while only Fawaz himself is listed as a performer (with his role specifically labelled as "synethesiser"), and no AI disclosure is included.
Fawaz also released a full album in June titled Dance Like Everybody's Watching. The 14-track record is entirely composed of similarly-styled reworkings of songs by TLC, Oasis, Marvin Gaye, and more. Again, only original writers are credited, while Fawaz is listed as the sole performer with no AI disclosure.
Additionally, streaming services also list Hallwood Media as the copyright holder for Fawaz's music. In July 2025, Hallwood made headlines after signing a record deal with Imoliver, an artist described by the company as "the top-streaming music designer on AI-powered platform Suno".
The Music has since reached out to Fawaz's management for comment but have not received any response to these requests.
This situation may feel somewhat familiar, given that in November of 2025, the track I Run by English outfit HAVEN. made headlines after it was removed from streaming services following allegations of being AI-generated.
The track found itself facing accusations of utilising vocals created by generative artificial intelligence music creation platform Suno. A re-recorded version of the track, utilising real, human vocals from singer Kaitlin Aragon was later released on streaming platforms, though HAVEN. had been forced to come clean about the situation and the role of AI usage in the track by this point.
"It shouldn’t be any secret that I used AI-assisted vocal processing to transform solely my voice for I Run." HAVEN. said in a statement given to Billboard. "As a songwriter and producer I enjoy using new tools, techniques and staying on the cutting edge of what’s happening.
"To set the record straight, the artists behind HAVEN. are real and human, and all we want to do is make great music for other humans."
Notably, the original version of I Run was also removed from the UK charts as well, with a statement from the Official Charts Company noting it is against their policy to "include within our charts any repertoire which is believed to potentially be infringing".
In the case of the ARIA charts, no such removal was reported. On November 17th, the track was No. 8 on the Singles chart, and by November 24th, it was absent, with the ARIA Report including it in its list of songs which "were in the Top 100 last week".
The following week, on December 1st, the reworked version of the song – now crediting Aragon's vocals – debuted at No. 32.
Again, ARIA did not remove the track from their weekly charts. Rather, the removal of the song from streaming services left no data for ARIA to analyse, meaning the song literally fell off the charts until the new version was issued.
If ARIA were in a position to remove the song, however, it would be a precedent-setting incident. The most recent version of the ARIA Charts Code of Practice (dated December 2025) does not make any mention of music made by, or assisted by, AI.
Fawaz himself released his own deep house remix of I Run the same week that the controversy broke, albeit using the original, AI-assisted version of the original song.
As it stands, Fawaz' version of Like A Prayer currently sits at No. 68 on the Singles chart, one position lower than the peak of its nine weeks in the chart. Additionally, it's No. 4 on the Australian Artists chart after a peak of No. 2, it's No. 4 on the Dance Single charts, and currently No. 1 on the Australian Artist Dance Single chart.
It's also worth noting that, while AI-generated music has become a hot-button topic in the world of electronic music, it's not wholly exclusive to this genre.
Last year, The Velvet Sundown – who were described as a blend of “’70s psychedelic alt-rock and folk rock” with flourishes of “modern alt-pop and indie structures" – generated controversy for acquiring massive streams despite being a band which does not actually exist.
Likewise, fans of Melbourne's King Gizzard & The Lizard Wizard thought the group had made a return to Spotify when a track titled Rattlesnake by a group called King Lizard Wizard appeared on the platform.
"We are truly doomed," frontman Stu Mackenzie succinctly said in response to the AI-generated song.
On the opposite side of things, however, US podcast Hyperfixed this year investigated a band called Sjowgren whose music had attracted allegations of AI generation due to the group's mysterious, anonymous existence and their lack of social media presence.
In this case, the music was confirmed to be real, albeit made by an artist whose reticence to participate in the industry in a traditional way resulted in unfortunate accusations.
While the latter example did not feature the same soulless hallmarks of AI generation as the likes of The Velvet Sundown and King Lizard Wizard, this is where the likes of Fawaz's song has also drawn criticism, with many labelling it either lazy or, as one Instagram user said in a comment on Thomas' video, the "cheesiest garbage".
However, the large issue at play here is not whether Fawaz's music is AI-generated or created organically but mainly that – for the most part – the general public don't seem to care about the larger issues at play regarding artist copyright.
While Fawaz himself has admitted to using AI as "a tool", it appears as though this is the direction in which the winds of the music industry are blowing.
In December, Harvey Mason Jr. – the boss of The Recording Academy and the Grammys – explained that AI usage was not a dealbreaker when it came to being considered for the industry's top honour. He also pointed out that AI usage had become so rampant, that almost every songwriter or producer he spoke to had used AI in the studio in some way.
“It runs the gamut of people texting lyrics or ideas or how they feel when they wake up and generating an entire track, lyrics and melody from it,” Mason explained.
So if AI-generated music is becoming more and more considered the norm in the industry, what then is the issue?
As Mitch Thomas pointed out in a follow-up video to his allegations of Fawaz's track being AI, programs like Suno are "training itself on copyrighted music made by real artists and then using that information to generate competing songs."
Indeed, only a few weeks ago, it was revealed that thousands of songs by Australian artists were among the millions stolen for use in AI datasets.
The claims came from The Atlantic reporter Alex Reisner, whose ongoing investigations into the world of AI training data had uncovered massive datasets of songs which are reportedly being shared within the community focused around AI development.
According to Reisner, one of the four datasets uncovered include 12 million songs, another has nine million, while two others include more than 100,000 songs. While tracks by local artists such as AC/DC, Midnight Oil, Crowded House, Sia, Yothu Yindi, and others are included, so too are songs by international acts like The Beatles, Nirvana, Taylor Swift, and Madonna – including her song Like A Prayer.
The issue here is that all of these artists have organically created these songs, only for them to be stolen for training in AI datasets – without proper compensation – which will ostensibly use the artistic and creative choices within these songs to inform AI-generated songs.
Put simply, if an artist wants to use an AI prompt to craft a song that sounds similar to, say, Lorde, then the real Lorde's music will be used to train a dataset that will in turn craft an imitation. If you squint your ears, it can be a very passable imitation at that. It's essentially musical plagiarism on a grand scale.
This is the reason as to why the Australian music industry is pushing for strengthened copyright law in the face of AI. Earlier this month, numerous Australian musicians and industry figures visited Parliament House in Canberra to protest any potential future text-and-data-mining exceptions in the Copyright Act.
Though the Federal Government ruled out such an exception last October, new reports that the federal government has considered new plans to allow AI companies to train on Australian copyright material without permission inspired this pilgrimage to Canberra.
“ The work that we've written, and that I've written and recorded, is essentially owned by me,” said Hunters And Collectors' Mark Seymour. “On that basis, the law states that it can't be used without us granting our permission first.”
“For my work to be used without my permission, it's theft,” echoed Something For Kate's Paul Dempsey. “That's basically what it amounts to.”
“ You want to use our work? Ask us,” added producer KLP. “You want to use our work? Credit us, and also pay us.”
Just last week, alongside the allegations of Fawaz's song being AI-generated reaching public prominence, the IFPI announced a new global music industry standard for AI labelling in musical recordings.
Welcomed by ARIA, and set to be rolled out in the "near future", the forthcoming standard is designed to give listeners clearer information about what they’re hearing and whether songs they listen to contain generative AI.
Labelling recordings as AI will take place under two umbrellas. The first is “AI-Generated,” which will indicate that generative AI was behind the entire production or primary creation of a recording, whether it’s the lead vocal or an instrumental.
The second is “AI-Assisted,” which will apply to recordings that are substantially recorded and created by humans but utilise generative AI for expressive elements.
This AI labelling has been something that has been called for greatly in recent months. In fact, Thomas said as such to Fawaz in a comment shared on Instagram, explaining that transparency is far more respectable than allegedly hiding behind a façade of organic creation.
"At least label it as AI," he wrote. "If you were transparent, I genuinely wouldn’t have an issue. You’re taking credit for the work of the millions of artists the AI was trained on."
Spotify last year introduced fresh measures to strengthen AI protections for artists and producers, including a new impersonation policy, and the investing of more resources into their content mismatch process.
Additionally, they also spoke of a new industry standard disclosure for music which features AI, which – like the recent announcement by the IFPI – means that both artists and rightholders are able to disclose whether AI was used in terms of vocal generation, instrumentation, or post-production.
At the time of the announcement in September, the disclosure process had not been implemented, and as of April, the platform had launched a beta feature which allowed artists to share how they used AI in their music.
However, while credits for "specific contributions like vocals, lyrics, or production in Song Credits on mobile" were offered to the consumer, this was only made available "where artists have chosen to disclose through their label or distributor".
Effectively, if an artist did not want to disclose the fact that they had utilised AI in the creation of their work, it is not yet mandatory for them to do so.
"Because we depend on artist disclosure, the absence of a credit doesn’t mean AI wasn’t used," Spotify wrote. "Not all distributors enable artists to disclose yet, but we intend to expand this more broadly over time."
Currently, many distributors accept music that is AI-generated, with many requiring discretionary disclosure of the use of AI tools.
Tunecore, for example, note that they "will not distribute any works that are 100% AI-generated", but add they are in support of "the use of AI technology that enhances human creation, assisting with artists' productivity and increasing fan engagement and artist revenue."
In a comment on Reddit earlier this year, another distributor, Tunearo, stated that their policy is "anything below 50% AI score will be distributed, however anything over 50% won't be distributed," adding that "we focus on original content".
Speaking to The Music while Australia this week, Sam Duboff, Spotify's Global Head of Artists, Marketing & Policy, responded to the recent AI music standard introduced by the IFPI, referring to AI transparency as both "really important and really complicated," and hinted towards penalties for those who did not disclose AI credits in their work.
"When we launched AI credits in partnership with distributors, we weren't really sure if artists were willing to share [this information]," Duboff explained. "We're seeing really significant uptake. We're getting tens of thousands of AI credits every day already.
"The real goal is to have an easy way for artists in good faith to share how they're using AI, and then it might mean something like where if an artist isn't sharing that they're using AI in a significant way, maybe that's when you can add different incentives and penalties."
While Duboff did not outline the sort of incentives and penalties that could be introduced, perhaps this is a vital first step that all outlets – not just streaming platforms, but radio, charts, and more – need to be introducing.
These new global music industry standards from the IFPI will ostensibly be ensuring that music fans are aware the recordings they're listening to contain AI music if they are labelled as such, though no mention of penalties for failure to disclose have been announced either.
Last month, streaming platform Tidal announced that while they "will accept AI-generated music" and will identify and tag "content we identify as 100% AI-generated", their new approach means they will "not knowingly attribute royalties to music we identify as wholly AI-generated".
Though it's obvious how this may be rolled out and noted on the likes of streaming platforms and the charts, the question does become how this will be extended to platforms like radio. Will this mean stations will be forced to announce whether a song is AI before playing? Or will digital stations include an AI disclosure on the data presented to listeners on screens?
Despite questions that arise, this may also be a positive step in stigmatising AI music in general.
Presenting listeners with options to filter out AI music feels like an obvious next step, it also provides radio stations with the opportunity to position themselves as being anti-AI. Perhaps marketing and branding for these stations can be centred around the fact that only music created by humans makes it onto their playlists.
However, the spectrum of AI-generated music does present issues. At which point does AI become an issue? 100% AI-generated music is undoubtedly problematic, though many naysayers may agree that a songwriter utilising AI to find the perfect rhyme for their chorus may be less of a deal-breaker in comparison.
Should, therefore, the latter be tarred with the same brush as the former? Should thresholds of acceptable AI-generated content be introduced? How does one determine if a song is 50% AI? How does 75% AI compare with 74% AI?
These are definitely questions the music industry is being forced to reckon with, and the decisions need to be made sooner rather than later.
However, the real issue isn't that artists like Josh Fawaz are creating songs alleged to be AI-generated, it's the fact that – when the livelihoods of artists are at stake – the response from the general public is so muted.
But let's play devil's advocate for a second. Why would the general public be angry about such an incident? After all, the song impacted radio, it crept into the charts, and the general consensus is that it's a popular song. Why wouldn't people have a dismissive attitude? It's not their fault that Fawaz crafted an allegedly AI-generated song, and it's not their fault that they enjoy it.
But that's exactly the point – they're supposed to like it. They liked it 37 years ago when Madonna released her original, and now that AI software has leaned upon the original to create a modern recreation, it's just history repeating.
As CD Media Consultancy Founder and Principal, and former ARN Chief Content Officer, Duncan Campbell said in a comment shared on a LinkedIn post recently, "Radio's job is to play popular hits"
"These days radio is not the only platform to hear hits and generally are not hit makers [like] they once were," he added. "If the listeners like it and want it, which research would validate, then radio would be silly not to play it."
However, the general public are forgetting the fact that if AI-generated music is taking over the industry, what room is left for human artists to thrive?
Their music is already getting lifted illegally to make poor imitations, their efforts are being overlooked for music that can be created in seconds, and their share of the royalties and airplay are getting progressively smaller.
Not only do Australian artists have to contend with the disappointing ACMA ruling earlier this year which entrenched the tiny amount of Australian music that commercial radio has to play during peak hours, but they now have to compete with AI-generated music within that quota.
It's no wonder that local artists are angry.
The general public, in contrast, aren't thinking about this. They're listening to what is being served to them and responding based on their instincts of whether they like the music or not.
Notwithstanding the opinions as to whether songs like the Like A Prayer rework are subjectively 'good' or 'bad', it's this servicing of AI-generated music which needs to be fixed.
As Thomas pointed out, if AI-generated music is dominating the airwaves or streaming platforms, it diminishes the amount of royalties that can be paid towards real musicians.
"Spotify doesn't have an unlimited resource pool of money to give to artists," he explained. "It has a finite amount of money that it allocates for distribution.
"So by these AI songs collecting millions of streams worth of royalties on their platform, it's taking away money from other artists who have done the right thing."
The presence of AI-generated music has also begun to infiltrate live performances as well. Earlier this month, musician Aidan Sammut performed at an event put together by music services agency Good Intent at Bootleggers Newtown, taking to social media the following day to claim "the artist that played before us was AI-generated".
"We only found out because a member of another band had happened to go through their Instagram account," Sammut noted. "The account contains AI-generated visuals over what I'm fairly certain is AI-generated music, lyrics, and vocals.
"Worse still, there was no disclosure on any of the event's promotional material that it contained an artist that was AI-generated."
The artist in question, Afro Charles, describes themselves as a "Sydney based pop music band" which features its namesake as the "male, virtual avatar" of "lyricist, vocals (live performances) and producer" Damian Amamoo, while "Mei Ling is a female virtual avatar who sings alongside Afro Charles".
"Damian will sing and perform the band's songs live on stage adding his vocals to Afro's and Mei Ling's vocals," the outfit's bio said of their approach to live performances. "In time, the band will have other performers live on stage, playing orchestral instruments that appear in our songs like the saxophone, cello, violin and others."
Good Intent’s founder and director, Rob Carroll, took to social media to "take full responsibility" for the artist's booking, admitting that while he "didn’t look at this particular artist’s socials", he added that their "representatives' communication used wording consistent with genuine artists".
"I hadn’t seen them live, but now looking at their socials - their content is clearly and blatantly AI."
Again, if these artists are being served to the general public, and no mandatory disclosure of their artificial nature is forthcoming, it can be easy for consumers to miss the fact that these acts are AI.
While there is undoubtedly less at stake if a radio listener passively listens to an AI-generated artist than if a promoter books one for a live show, this infiltration of AI music is providing us all with more reason to research whether the artists being served to us are human or not.
Frankly, the onus to verify human music should not be on the job of the general public. This is what needs to be left up to streaming services, to distributors, to broadcasters, and to everyone who has the ability to let music consumers know whether or not the music has been created with actual humans.
So where do we go from here? Radio airplay being dominated by a track alleged to be AI is one thing, but if nothing is done, then where will it stop?
If mandatory disclosures for AI content are not introduced, then the general public will continue to have no idea that the music they're consuming is not created by humans. If broadcasters are not telegraphing this information, then the proliferation of AI music on the airwaves will only continue.
If this information is not shared with consumers, they will have no idea that the music they're enjoying is not organic, and as a result, the real musicians out there who create without relying on AI will continue to suffer – creatively, financially, and almost any other way you can think of.
Though it looks like, for better or worse, AI tools being used in the music creation process may become the new normal, the line has to be drawn somewhere.
There is a massive gulf between an artist topping the charts with a track crafted by writing a prompt on Suno and a budding songwriter finalising a chord progression with the help of ChatGPT. This feels obvious. However, to say that both are worthy of the same amount of airplay, royalties, attention, and acclaim is an argument which feels fundamentally flawed.
It seems as though the presence of AI within music is here to stay, and whether you think this is good or bad, this feels like a new normal we have to become acclimatised to. What we can push back on, however, is AI-generated music being presented, and accepted, as the standard.
It's easy to share opinions on whether AI music is 'good' or 'bad', but it's harder to get the general public to actually care about the impact of what its ubiquity can mean.
Music isn't just a throwaway commodity which should be listened to passively, for many creators, it's not just a passion project, but it's their income and their livelihood.
Without humans, AI music won't exist, and if AI music dominates, there's no room left for humans.
