Australia has become a decisive testing ground in the global contest over who controls copyrighted work used to train artificial intelligence. On October 19, the International Confederation of Music Publishers said Australian songs had already appeared in AI training data without permission or payment, while technology companies continued pressing policymakers for a training exception or an opt-out system.
The immediate dispute is centered on music, but its consequences extend to visual art, photography, illustration, books, film and every practice that circulates as a digital file. The essential question is not whether artists should use AI. It is whether access to creative work should begin with consent and licensing, or with a presumption that a work may be taken unless its creator finds a way to refuse.
A policy debate with global consequences
Australia’s current copyright framework does not contain a broad text-and-data-mining exception for commercial AI training. OpenAI and Anthropic have argued that a carefully designed exception would make the country more attractive for model development. Creative organizations have countered that changing the law would transfer negotiating power from rights holders to technology companies.
The government has not enacted an opt-out regime. A parliamentary inquiry is still examining AI policy, and Prime Minister Anthony Albanese has publicly said creators should retain control and receive payment. That distinction matters: the present story is a live policy contest, not a completed legal change.
What makes the Australian debate unusually important is its international reach. A model trained in one jurisdiction can produce and distribute outputs everywhere. If a country becomes a permissive training venue, rights holders elsewhere may face practical difficulty enforcing the protections their own laws provide.
Why opt-out reverses the burden
In an opt-in licensing system, a developer must identify a rights holder and secure permission before using protected material. In an opt-out system, the creator must discover that a use is possible, understand the relevant mechanism and communicate a refusal in a format the developer will honor.
That burden is especially heavy for independent artists. A large publisher or collecting society can maintain rights databases, monitor model providers and negotiate licenses. A painter, photographer or small studio may have thousands of images spread across portfolio sites, galleries, social platforms and archives, each with different metadata and technical controls.
Opt-out also raises a verification problem. Even when a creator registers a refusal today, there may be no reliable public record showing whether an earlier training dataset already contained the work. This is why transparency about datasets, auditable licensing records and workable remedies are as important as the legal label attached to an exception.
What artists should preserve now
Artists do not need to wait for a final policy outcome to strengthen their position. Keep dated source files, export histories, exhibition records, invoices and contracts. Embed authorship and contact information in image metadata where practical, and retain high-resolution originals that were never published online. These records cannot prevent every unauthorized use, but they can establish provenance and support a later claim.
Before uploading work to a platform, read the terms governing machine learning, sublicensing and derivative services. Screenshots or archived copies of the terms accepted on a particular date can be valuable if those terms later change. Artists represented by galleries, agencies or publishers should also confirm who is authorized to grant AI-training licenses and how any revenue would be divided.
Collective action may become increasingly important. Copyright societies, professional associations and sector-wide licensing systems can reduce transaction costs without erasing consent. The strongest licensing framework will be one that allows technology to develop while keeping creators visible, compensated and able to say no.
The principle beneath the politics
Australia’s decision will not settle the global argument, but it could influence how other governments frame the balance between innovation and creative rights. For artists, the central issue is straightforward: innovation should not depend on making individual creators responsible for policing every use of work they already own.
The next stage to watch is the parliamentary committee’s report and any concrete proposal that follows. Until there is statutory language, claims that Australia has already opened copyrighted work to AI training are premature. What is real now is the pressure on policymakers—and the opportunity for artists to insist that transparency, consent and payment be built into the system before the rules are rewritten.
ATIM approaches artificial intelligence with curiosity, independence, and balance. We explore its extraordinary potential to empower artists while examining the questions it raises around authorship, copyright, consent, creative identity, transparency, and human expression.
We are neither advocates for AI nor opponents of it. Our focus is how the technology is used—and what that means for artists.
Above all, our commitment is to artists. We look at every development through the lens of what can support, protect, and empower the creative community—identifying opportunities, examining potential risks, and providing artists with the information they need to make informed decisions about their work, their rights, and their future.