The chart now asks who sang
Australia's official charts just became the first to write an AI eligibility test into their code of practice, with powers to strip a number one after the fact. The trigger was a Madonna cover that fooled a whole winter.
By Katie Delaney · 2026-08-30 · 8 min read
The cover that forced the question#
Every rule has a story it is embarrassed by, and for the new rules on ai music artists it is a Madonna cover. An AI-assisted version of Like a Prayer by Australian producer Josh Fawaz became, per ABC News, the most-played song on Australian radio and reached number four on two ARIA charts in July, before Fawaz updated the track listing on Spotify to say the vocals and drums were AI-generated. The chart had celebrated a voice nobody owned, and the ai music artists debate stopped being theoretical overnight.
So the answer to the question people now type into search, have any ai songs charted?, is yes, emphatically, and at the top of the radio pile. That is what turned a policy discussion into a code change. On 25 August the Australian Recording Industry Association announced new eligibility rules: wholly AI-generated recordings are out, effective with the chart dated Monday 31 August, published Friday 28 August.
The line ARIA drew is careful. A lead vocal generated by AI makes a track ineligible, as does a key instrumental performance, or music generated entirely from a prompt. A human lead with AI backing textures underneath stays in. The test, in the code's own words, is whether the recording is substantially human made and raises no stream or chart manipulation concerns, per ARIA's chart changes FAQ.
ARIA chief executive Annabelle Herd put the stakes plainly: a chart that rewards unlicensed AI output would undercut the very basis of the recorded music the organisation exists to represent. And the teeth are real. The updated code lets ARIA exclude or remove recordings prospectively or retrospectively, adjust positions, withdraw accreditations, and revoke or request the return of number one awards.
A declaration, not a detector#
Here is the detail most coverage skipped, and it is the one that decides whether this works: enforcement runs on paperwork, not on listening. Every release submitted to the ARIA survey now requires a declaration about the use of generative AI, and that declaration is what ARIA works from, per its own FAQ. The chart is trusting the den to describe its own tracks.

That is not naivety, it is necessity. Reliable detection does not exist yet at broadcast quality. A study submitted to arXiv this month, Assessing AI-generated music detection in real-world broadcast monitoring, built a 40-hour dataset of real television recordings and found that detection models achieving near-perfect scores in clean lab conditions degrade substantially under synthetic broadcast conditions. Its conclusion: current approaches remain insufficient for reliable detection in broadcast monitoring. The paperwork exists because the science is not ready.
The definitions themselves come from upstream. ARIA applies the labelling standard the global music community announced on 10 July, per IFPI's labelling programme, which distinguishes AI-Generated from AI-Assisted recordings, and follows the global chart eligibility principles IFPI published on 30 July: the AI service used is properly authorised and lawful, the track is substantially human made, and it raises no manipulation concerns. IFPI chief executive Victoria Oakley's framing travels well: official music charts do more than just track sales, they celebrate human artistry and endeavour.
Watch the rollout clause. IFPI says the principles are expanding across more than twenty official chart programmes worldwide, per the same announcement. Australia moved first, but this is a template on tour, and Music Ally's read is that other chart bodies will follow within months rather than years.
The flood the chart is standing against#
Why write rules for ai music artists now? Because the volume stopped being a curiosity. Deezer, the one platform publishing hard numbers, reported in July that fully AI-generated tracks passed half of all daily uploads at their June peak, roughly 90,000 tracks a day, and that it detected and tagged 13.4 million AI tracks in 2025. The undergrowth is filling faster than anyone can walk it.
The counts behind that line: 10,000 a day in January 2025, 30,000 by September, 50,000 by November, 60,000 in January 2026, 75,000 and 44 percent of all uploads by April, per TechCrunch's April report, then past half by June, per its July follow-up. Almost none of it is listened to. Deezer puts AI music at between 1 and 3 percent of total streams.
That second chart is the quiet confession in the whole affair. The ai generated songs flooding the platforms are mostly not art seeking an audience, they are volume seeking a payout: up to 85 percent of their streams read as fraudulent to Deezer's own systems, against 8 percent across the catalogue, per Deezer's July figures. The chart rule and the fraud fight are the same fight wearing two coats.
What ai music artists' marketers do next#
Australia is drawing a line in the AI music debate. Fully AI-generated songs will be excluded from the official ARIA charts and ARIA Awards, while music created with AI assistance can still qualify if humans wrote the song and performed the core vocals and instruments. That distinction may become a template for the rest of the industry: AI as a tool is acceptable. AI as the artist is not.
The template reading is the right one, and it hands working acts, and the ai music artists working with hybrid tools, three practical jobs. First, document the process. The declaration system makes provenance a marketing asset: an act that can show its human hands has a claim its rivals cannot fake, and that claim now has chart consequences. Provenance is brand work, the kind the folkfox brand strategy practice builds spines for.
Second, treat AI assistance as a disclosure question before it becomes a scandal question. Fawaz's cover did not fall foul of a rule when it charted, but the retrospective powers in the new code mean the next quiet deployment of an AI vocal can cost an accreditation after the fact. The music industry teams we work with are writing AI-use lines into release checklists this week, alongside the content plans that carry a release.
Third, stop treating the charts as a vanity metric. A chart position that survives an AI eligibility audit is now a verified scarcity, which makes it worth more in a pitch deck, a sync negotiation or a festival booking than it was a month ago. Scarcity is what the whole outfoxing comes down to: when anyone can generate a plausible song, the provable human one carries the premium. The moonlit shortcut got fenced, and the long trail just got more valuable.
Frequently asked questions#
Have any ai songs charted?
Yes. An AI-assisted cover of Madonna's Like a Prayer by producer Josh Fawaz became the most-played song on Australian radio and reached number four on two ARIA charts in July 2026, before its AI vocals and drums were disclosed. That episode directly preceded the new rules.
Are ai music artists banned from the ARIA charts?
Wholly AI-generated recordings are ineligible from the chart dated 31 August 2026. Artists who use generative AI in a supporting role remain eligible, provided the recording is substantially human made and raises no stream or chart manipulation concerns.
How does ARIA know whether a song used AI?
Through declarations. Every release submitted to the ARIA survey must now declare its use of generative AI, applying the global labelling standard that distinguishes AI-Generated from AI-Assisted recordings. Reliable automated detection at broadcast quality does not yet exist, so the system runs on disclosure.
What happens if an AI track is discovered after it charts?
The updated code of practice gives ARIA retrospective powers: it can remove recordings from charts after the fact, adjust positions, withdraw accreditations, and revoke or request the return of ARIA number one awards. Ineligible recordings are also excluded from the ARIA Awards.
Do ai generated songs actually get listened to?
Barely. Deezer reports AI music at between 1 and 3 percent of total streams despite being over half of daily uploads at the June 2026 peak, and up to 85 percent of streams on fully AI tracks were flagged as fraudulent in 2025.
Will other charts adopt the same AI rules?
Very likely, and ai music artists everywhere should assume it. ARIA's rules implement IFPI's global principles, published 30 July 2026, which IFPI says are expanding across more than twenty official chart programmes worldwide. Australia is the first mover, not the exception.
Read more on this topic#
A Munich court just put a price on the training set
The other front in the same war: what courts say about the data AI music is built from.
Read the pieceChatGPT Paid Ads for Music & Events
Where answer-engine advertising fits for acts, venues, promoters and ai music artists working with hybrid tools.
Read the pieceThe regulator followed the footnote home
From this morning's edition: another industry learning that claims get audited after publication.
Read the pieceBuilding an act the algorithm cannot fake?
folkfox markets music on provenance: human process, honest claims, and campaigns that turn verified artistry into the premium it now is.