ARIA Sets New Rules for AI Music: Why Disclosure Is Becoming Part of Music Infrastructure

Australia’s official music charts have drawn a new line around generative AI. But behind the question of chart eligibility lies a broader challenge for the music industry: how should AI involvement be disclosed, recorded and supported by evidence?

The Australian Recording Industry Association (ARIA) has introduced new eligibility rules for recordings created with generative AI, marking an important development in how the music industry responds to AI-generated and AI-assisted content.

Under the updated ARIA Charts Code of Practice, recordings that are wholly or primarily AI-generated will not be eligible for the ARIA Charts. Recordings created substantially by humans, with generative AI used in a supporting role, can remain eligible, provided they meet the other chart requirements. The changes apply from the ARIA Chart dated 31 August 2026.

Importantly, ARIA describes the change not as a blanket ban on AI music, but as a distinction between AI-generated and AI-assisted recordings.

According to ARIA’s guidance, an AI-generated lead vocal or key instrumental performance can place a recording in the AI-generated category. By contrast, a human lead vocal with AI-generated backing vocals may be considered AI-assisted. Uses such as AI mastering, stem separation, reverb and certain production tools do not by themselves make a recording ineligible.

This distinction reflects a broader reality emerging across the music industry: AI involvement is not necessarily binary.

From AI Labels to AI Disclosure

Perhaps the most significant part of ARIA’s new framework is not simply which recordings qualify for the Charts. It is the information now required to make that distinction.

Every release submitted to the ARIA survey must now include a declaration about the use of generative AI. ARIA states that it will work from this declaration in the first instance. Where a credible concern is raised, the rights holder may be asked to provide further information and evidence before ARIA makes an eligibility decision.

This creates a simple but important information flow:

Declaration → Information → Evidence → Decision

ARIA ultimately makes the decision about chart eligibility. But before that decision can be made, the relevant information about AI involvement needs to exist. That raises a broader technical question for the music industry:

How should AI involvement in music be described, recorded and supported by evidence?

A simple label such as “AI-Generated” or “AI-Assisted” can communicate an outcome clearly. But different industry processes may require more granular information underneath that label. For example: Which creative elements involved AI? What role did AI play? To what extent was it involved? Which recording or digital content does the declaration refer to? And what information or evidence supports the declaration? These are increasingly questions of music metadata, identity and information infrastructure, rather than labels alone.

Recording the Facts Without Making the Decision

There is an important distinction between recording information and making a judgement based on that information.

A chart organisation may use AI disclosure to determine eligibility. A rights organisation may need it for rights review. A distributor or platform may use it for labelling or policy compliance. A listener may simply want greater transparency about how a recording was created.

Those decisions do not necessarily need to be made by the disclosure infrastructure itself.

The underlying technical role can be much narrower: to record relevant facts in a consistent form, associate them with the correct music and digital content, and make the information available to the appropriate systems or decision-makers.

This distinction becomes increasingly important as different organisations develop different policies around generative AI.

Where ISRC-AAM-CID Fits into the Discussion

These are questions that the Australian AI Music Alliance and OZBeat AI, a full member of DDEX and a contributor member of C2PA, have been exploring through the ISRC-AAM-CID framework, first published in February 2025. The original framework started with Sound Recordings and brought together three information components:

ISRC — Which recording?

The established identity of the Sound Recording.

AAM — How was AI involved?

AAM, or AI Application Music, provides a methodology for describing AI involvement across different creative roles and expressing that involvement in a structured form.

CID — Which digital content?

A content reference or anchor that can associate the disclosure with specific digital content. The purpose is not to determine whether AI use is good or bad, whether a work qualifies for copyright protection, who owns the rights, or whether a recording should be eligible for a particular chart. Instead, the framework explores a more fundamental technical problem:

Can information about AI involvement be recorded in a structured way and connected to existing music identity and specific digital content?

That distinction matters. ARIA determines what qualifies for the ARIA Charts. Other organisations will make their own decisions according to their own policies and responsibilities. A neutral disclosure layer can instead focus on making relevant information available to support those different processes.

The Next Question: What Should the Music Supply Chain Carry?

ARIA’s new rules also highlight why the conversation around AI disclosure is likely to move beyond simple classification.

If AI declarations become relevant to charts, rights management, distribution, licensing, platform policies and consumer transparency, the industry will increasingly need to consider how that information travels.

Some questions remain open:

What level of AI disclosure is useful across the music supply chain?
Which information should travel directly with music metadata, and which information should remain externally referenced?
What evidence can meaningfully support an AI-use declaration?
How should AI disclosure connect across Sound Recordings, Musical Works, parties and specific digital content?

These are also among the questions being explored as ISRC-AAM-CID moves into its next phase, with greater attention to structured disclosure, content binding, provenance and interoperability with existing music identity and metadata systems.

ARIA’s decision does not answer all of these questions, nor should one organisation be expected to.

But it demonstrates something increasingly important:

As AI becomes part of music creation, disclosure is becoming part of music infrastructure.

The challenge ahead is not simply deciding whether a recording is “AI” or “human”. It is developing reliable ways to describe what happened, connect that information to the right content, and allow different parts of the music ecosystem to use those facts according to their own responsibilities.

That is where the next stage of the AI music standards conversation is likely to become increasingly important.