The AI-Assisted Madonna Cover That Forced Australia to Rewrite Its Chart Rules

A cover of Madonna’s “Like a Prayer” has become one of the clearest examples yet of how quickly generative AI can collide with the rules of the traditional music business. Australian producer Josh Fawaz used AI-generated elements in his version of the song, including vocals and drums, and the recording went on to become a major commercial success. Its rise helped intensify a debate that has now pushed the Australian Recording Industry Association to draw a much sharper line between AI-assisted music and recordings considered substantially generated by AI.

Madonna performing during The Celebration Tour

Photo: Ronald Woan / Wikimedia Commons. Licensed image.

Key facts

The Josh Fawaz version of “Like a Prayer” reached the upper end of several Australian charts.
Reuters reports that the track also became the most-played song on Australian radio.
Its credits disclosed AI-generated vocal and drum elements.
ARIA has now clarified that eligible chart recordings must be substantially human made.
AI-assisted production can remain eligible, but primarily AI-generated performances can be excluded.

Large concert crowd representing the commercial success of an AI-assisted hit

An AI-assisted record became a mainstream hit

The significance of the Fawaz case is not simply that generative AI was used in production. AI tools are already appearing throughout modern workflows, from stem separation and restoration to sound design and arrangement assistance. What made this recording different was the prominence of generated performance elements inside a record competing directly in mainstream charts and radio.

That forced the industry to answer a much harder question: when does AI stop being a production tool and start becoming the performer?

ARIA’s new line is about authorship, not software

ARIA has avoided a blanket ban on all music made with AI tools. The association instead distinguishes between music that remains substantially human made and recordings in which generative systems create the lead vocal, a principal instrument or a major share of the performance itself.

That distinction matters because a prohibition on any AI use would be almost impossible to enforce. Modern producers already use machine-learning features inside mixing, mastering, restoration and composition software. The chart rule therefore focuses less on whether AI touched a project and more on whether humans remain responsible for the essential creative performance.

Why it matters: the Fawaz case shows that chart regulators are moving toward a contribution test. The important question is becoming “what did the human actually create or perform?” rather than simply “was AI used?”

Radio success made the debate impossible to ignore

The commercial reach of the recording is what turned a technical discussion into an industry issue. A niche experiment can remain outside the charts conversation. A track receiving heavy radio rotation and millions of streams cannot.

The recording demonstrated that listeners may accept or simply not notice generated performance elements when the song itself is familiar and professionally produced. That creates an uncomfortable challenge for labels, broadcasters and chart organizations that have traditionally assumed the credited performer corresponds to a human performance.

The rule also connects AI to chart manipulation

ARIA’s framework goes beyond creative authorship. The organization has also emphasized that recordings must comply with rules around legitimate consumption and chart manipulation. That is significant because AI-generated music and streaming fraud have increasingly appeared in the same policy discussions, particularly as platforms receive extremely large volumes of low-cost generated uploads.

An AI-assisted record is not automatically fraudulent, but the economics of automated music production can make large-scale catalogue manipulation easier. Regulators are therefore dealing with two separate questions at once: who created the recording, and are the streams themselves legitimate?

Smartphone music streaming representing chart and radio consumption

What producers should take from the case

The new Australian position does not mean producers need to abandon AI tools. It does mean that transparency is becoming more important. Credits, metadata and documentation of how a recording was made may increasingly influence whether a track qualifies for charts, awards, licensing opportunities or platform labels.

For independent artists, the safest approach is to treat generative tools as part of a documented production workflow rather than as an invisible replacement for the artist. If a system generates a lead vocal, signature instrumental performance or central composition, that may eventually trigger different rules than using AI to clean noise or suggest processing settings.

A precedent other markets may study

Australia is unlikely to be the last territory to formalize this distinction. Streaming services, collecting societies, awards bodies and chart organizations all face the same basic problem. They need rules that permit legitimate technological innovation without making human authorship meaningless.

The Fawaz story matters because it moved the debate from theory to a real hit. Once AI-assisted records can compete at the top of radio and streaming charts, policies that once seemed optional become operational necessities.