ANALYSIS Two democracies moved within days of each other to protect creative intellectual property from generative AI — and chose opposite mechanisms, revealing a strategic fork that will shape how the global creative economy absorbs the technology.
Why it matters
Australia's recording industry is drawing a line at the chart door: wholly AI-generated tracks are out, effective with the ARIA Chart dated Monday, August 31, 20264,5. Japan's Cabinet Office, meanwhile, is drafting a principle-code that would require generative AI businesses to disclose their training data and collection methods under a "comply or explain" framework6,7. ◆ One country is gatekeeping the output; the other is regulating the input. Together they sketch the two poles of an emerging global debate over how creative industries coexist with generative AI.
The big picture
The Australian Recording Industry Association updated its Charts Code of Practice to exclude music that has used generative AI unless it is "substantially human-made" and "raises no stream or chart manipulation concerns". Recordings that use generative AI in a supporting role remain eligible. Under the new rules, AI assistance can still be used, but humans must have written the song and performed the lead vocal and primary instruments3. AI can still be used for mastering songs and for using drum machines and auto-tune.
The trigger was concrete: an AI cover of Madonna's Like a Prayer by Josh Fawaz peaked at No. 2 on the ARIA chart, spent 16 weeks in the top 10, and has been streamed more than 48 million times on Spotify alone2. Fawaz later added generative AI credits to the song after a backlash. The track also appeared on commercial radio as an Australian song, helping stations fulfil obligations to promote 25 per cent Australian music — without paying the royalties.
ARIA CEO Annabelle Herd framed the policy around human artistry: "These changes reflect our intent to remain dynamic and promote the human nature of artistry in what is — to say the least — a rapidly developing space. Artists already use AI tools in their work, the Charts can and should evolve to keep room for that, but music generated wholesale by services built on artists' recordings is a different matter altogether". ARIA said it was "not interested in promoting or celebrating the success of AI-generated music that does not contain human artistry".
Japan's approach targets a different layer. Its draft principle-code would apply to generative AI developers and providers, including businesses outside Japan whose systems or services are available in Japan. The code asks firms to avoid crawling pirate sites, respect access restrictions such as paywalls, and increase transparency around how they manage IP risks. The draft was drawn up amid growing concerns that texts and images may be used to train AI models without consent and could potentially infringe copyrights. A government panel broadly approved the plan on August 18.
ANALYSIS Australia's move is enforcement at the point of recognition — charts, awards, radio play — rather than at the point of creation. ARIA said it could retrospectively adjust chart positions if music was later found to be largely AI-generated, and if a piece topped the charts, ARIA may request that awards given for reaching Number One were returned. Artists will be required to declare AI use when they submit a song for chart consideration. Yet the ARIA Charts Code of Practice is voluntary. That gap — mandatory disclosure paired with voluntary compliance — limits the policy's teeth to reputational pressure and chart exclusion rather than legal sanction.
Japan's code carries a similar structural softness. The government will use a comply-or-explain approach, setting out a nonbinding code and allowing firms to choose either to comply or publicly explain why they will not. ◆ Both regimes therefore rely on transparency as the enforcement lever: Australia demands that artists disclose AI use; Japan demands that AI firms disclose training data. Neither imposes binding penalties for noncompliance.
The Fawaz episode illustrates why output-side gatekeeping alone may prove insufficient. The track exploited a regulatory seam — qualifying as "Australian" content for radio quotas while sidestepping royalty obligations. Banning it from charts does not close that radio-quota loophole.
The international dimension is accelerating. Earlier this year a song was banned from Sweden's music charts because it was created by AI. In July, the International Federation of the Phonographic Industry said AI guidelines would be rolled out in Latin America, the Middle East, Africa, and Southeast Asia for use in their official charts. ARIA confirmed its updated code aligns with the labelling standard announced by the global music community on July 10.
What's next
ARIA's new rules take effect with the chart published Friday, August 28. Japan's draft code will proceed through further government review following the August 18 panel approval. ANALYSIS The two frameworks are not mutually exclusive — a jurisdiction could eventually pair input-side training-data disclosure with output-side chart eligibility rules — but for now they represent distinct bets on where the leverage point lies. The speed at which IFPI is exporting chart-eligibility standards across regions suggests the Australian model may propagate faster, while Japan's comply-or-explain template offers a governance blueprint that could appeal to countries reluctant to amend copyright statutes outright.