What CAIRG actually is
Most reporting on Australian AI and copyright refers to CAIRG without saying what it is, so:
the Copyright and Artificial Intelligence Reference Group was announced on
5 December 2023 by the then Attorney-General, Mark Dreyfus, as an outcome of
the 2023 Ministerial Roundtables on Copyright. The department describes it as
a standing mechanism to engage with stakeholders across a wide range of sectors on issues
at the intersection of AI and copyright
.
It is a consultation body, not a decision-maker. That distinction does most
of the work in reading any news about it: CAIRG does not change the law, it feeds
policy advice to government
. Nothing it discusses is in force because it was discussed.
What it has actually consulted on
Two strands, both narrower than the public argument. The department says its initial focus
was on the use of copyright material as inputs for AI systems and on copyright-related AI
transparency issues
.
In May 2024 participants were asked to complete a questionnaire as part of
an exploratory mapping exercise on uses of copyright material as AI inputs
. In
September 2024 a discussion paper went out on transparency, seeking views on
whether there are specific copyright-related AI transparency issues where the government
should consider urgent action
.
There is a detail in the meeting record worth pausing on. At a technical session on
29 May 2024 organised by the Tech Council of Australia,
Google gave an overview of how Large Language Models (LLMs) are developed and trained
and Canva provided insight into the deployment of AI models in an Australian context
.
Whatever you make of that, it is a matter of public record that the developers explained the
training process to the group considering the copyright questions about it.
The line that answers the big question
On 28 October 2025 the department hosted a CAIRG meeting on three priority areas, and recorded this alongside it:
The department noted that the Attorney-General had announced that the government is not
considering a text and data mining exception in Australian copyright law
.
That is the load-bearing sentence on the whole page. A text and data mining exception is the mechanism by which several jurisdictions have carved out AI training from copyright infringement. Australia is not pursuing one. So the default position stands: using copyright material to train an AI system in Australia is governed by ordinary copyright law, with no AI-specific exemption, and none is being drafted.
The three things that are on the table
The same 28 October 2025 meeting set out the three priority areas the government asked the department to explore with CAIRG, referring to an Attorney-General's media release of 26 October 2025. They are:
encouraging fair, legal avenues for using copyright material in AI through examination of licensing arrangements
improving certainty on the application of copyright law to material generated through the use of AI
exploring avenues for less costly enforcement in relation to AI outputs-related infringement, including through a potential small claims forum
Read them together and the direction is coherent. If you rule out an exception, the
remaining levers are licensing (make lawful use purchasable), clarity
on outputs (whether and when AI-generated material attracts copyright at all), and
cheaper enforcement (because a right you cannot afford to enforce is close to
no right). The third is the one with the most practical bite for individual creators, and the
phrase to watch is a potential small claims forum
.
The department confirmed a consultation paper would be circulated to CAIRG participants after that meeting seeking written feedback on the three areas.
What none of this changes today
CAIRG has been running since December 2023 and has produced consultations, meetings and policy advice. It has not produced a change to the Copyright Act. If you are deciding whether to train on material you do not own, or wondering whether your own work is protected against it, the answer sits in ordinary copyright law rather than in anything CAIRG has published, and that has been true for the whole period the argument has been running.
About this page
Every quoted passage is verbatim from the Attorney-General’s Department’s own page on the Copyright and Artificial Intelligence Reference Group, checked character by character against it and read on 23 August 2026. We have quoted rather than paraphrased anything that carries a legal or policy consequence, particularly the statement about a text and data mining exception, which is the fact most likely to be misreported.
This is general information about a public policy process and is not legal advice on any specific use of copyright material. We have not read the Attorney-General’s 26 October 2025 media release itself; the description of the three priority areas is as the department records them on this page, and it is cited that way. Policy in this area is live, so check the department’s page against the day you need it. AI Geek holds no interest in any party to this debate.
Sources
- Attorney-General's Department, Copyright and Artificial Intelligence Reference Group (CAIRG) (read 23 August 2026): the 5 December 2023 announcement by then Attorney-General Mark Dreyfus and its origin in the 2023 Ministerial Roundtables on Copyright; the description of CAIRG as a standing engagement mechanism; the initial focus on copyright material as AI inputs and on transparency; the May 2024 questionnaire and exploratory mapping exercise; the September 2024 transparency discussion paper; the 29 May 2024 technical session with Google and Canva; the 28 October 2025 meeting, the three priority areas drawn from the Attorney-General's 26 October 2025 media release, the confirmation that a consultation paper would follow, and the statement that the government is not considering a text and data mining exception.
Spotted an error, or has the policy moved? Tell us and we will check it against the department's page and log the outcome here.