A Queensland DJ called Josh Fawaz released a cover of Madonna’s “Like a Prayer” this year. It did what every artist dreams of. It became the most-played song on Australian radio, a fixture on commercial playlists, north of 48 million streams on Spotify alone.
On Friday it stopped being eligible for Australia’s official charts.
Not because listeners went off it. Because of how it was made, and because generative AI credits were only added to it after people started asking questions.
The rule that landed on Friday
The Australian Recording Industry Association updated its charts code of practice with effect from 28 August. Music made wholly or mostly with generative AI can no longer chart. To qualify, a recording has to be substantially human-made and raise no concerns about stream or chart manipulation. AI used as a tool in a supporting role is still fine.
ARIA gave itself teeth as well. It can remove ineligible recordings, alter chart positions, revoke accreditations and ask for physical awards back. Wholly AI-generated tracks are also barred from the ARIA Awards.
Their chief executive put the reasoning plainly: a chart that rewards unlicensed AI output would undercut the basis of the industry it exists to represent. Their other line is the one that should interest anyone selling anything, which is that artists are competing for attention in the most crowded market in history.
The song won and it didn’t matter
Sit with that for a second, because it’s the actual news.
By every metric the industry has used for fifty years, that track was a hit. Radio played it more than anything else. Millions of people streamed it voluntarily. It performed. And the performance was ruled irrelevant, retroactively, because of a question about who made it.
That’s a change in what counts. For twenty years the only test that mattered online was whether something worked. Did it rank, did it convert, did people click. Where it came from was nobody’s business. That test just got a second half bolted onto it, and the second half is applied after the fact, by people who weren’t asking at the time.
The disclosure is the part that bites
Note the sequence. The AI credits went on after the backlash, not before release.
That’s the bit worth being nervous about, and it’s got nothing to do with music. Every business quietly using generative tools for the work customers actually see is carrying the same exposure. Not because anyone is checking today, but because the rules are being written now and applied backwards, and the sin that gets punished is never really the tool. It’s the not saying.
We wrote earlier this year about consumers punishing brands that lean on AI, with distrust doubling in twelve months. That was people voting with their attention. This is different and more serious. This is an institution writing it into a rulebook with enforcement attached.
It won’t stop at music
The mechanism is easy to see once you’ve seen it once. Anywhere there’s a gate, someone eventually has to decide what qualifies to pass through it.
Industry awards will ask. Publishers and platforms will ask. Corporate procurement will ask, because their legal departments are already nervous about training data and copyright. Anyone running a directory, a chart, a listing or a shortlist has the same problem ARIA had, which is that their thing is only worth something if it measures what it claims to measure.
South African businesses will meet this through the back door, probably in a tender document or an agency contract, phrased as a warranty that the work delivered is original and cleared. Most people will sign it without knowing whether it’s true, because they never asked their supplier.
What to actually do about it
Nothing dramatic. Just stop being uncertain about your own material.
Know what’s in your marketing. If an agency or freelancer hands you copy, photography, video or music, ask what was generated and get the answer in writing. Your name goes on it, not theirs, and the liability follows the name. Keep the boring records: who wrote it, who shot it, what was licensed, what was made in-house.
Be straight where it counts. Nobody needs a disclosure notice on an internal spreadsheet. But if a customer would feel misled to learn how something was made, that’s your answer, and finding out later is always worse than saying so upfront.
And take the opportunity, because there is one. ARIA has effectively created a certified category called substantially human-made, and other institutions will copy the language. Being able to prove your work is yours is becoming an asset, at exactly the moment everyone else is finding out they can’t.
The song is still on Spotify. It just can’t win anything anymore.
If you’d rather know exactly who made your work, that’s how we operate.