NextFin News - Australia’s copyright fight over artificial intelligence has become a test of whether creators will keep ordinary property rights, or whether large technology companies will win a cheaper route to the books, music, films and journalism they need to train their systems. Anna Funder’s warning is direct: the country’s fairness culture and legal framework should stop any attempt to gut copyright for the sake of AI scale.
The dispute is no longer hypothetical. Funder says her books have already been ingested into AI systems without consent and without payment, and she argues that technology companies are now pushing governments to make that conduct legal after the fact. The issue matters because copyright is not just a cultural concern. It is the payment system for publishing and other creative industries, and it determines whether AI developers must negotiate with rights holders or can treat creative work as free raw material.
That is why the policy stakes in Canberra are so high. The Australian Government has already said it is not considering a text and data mining exception in copyright law, after a Senate inquiry into artificial intelligence. It has also said it is working with stakeholders on licensing, clarity around AI-generated material and lower-cost enforcement. For creators, that is the crucial line: the state has not abandoned the principle that copyrighted material remains property, even as AI firms press for broader access.
Funder frames the argument in blunt commercial terms. Books are not just expressions of personality or culture; they are licensable assets that generate income through publishers, territories and formats. If an AI system can ingest them without paying, the creator’s bargaining power falls just as machine learning models become more valuable. That is why writers, musicians and other artists see the copyright debate as a fight over market structure, not a sentimental argument about the past.
The political pressure is obvious. AI developers need large quantities of high-quality text to build better models, and books are especially attractive because they are edited, verified and coherent. Yet the better the source material, the stronger the case for compensation. A system that relies on valuable copyrighted work should pay for that input the way other industries do when they buy essential supplies.
Funder also warns against a familiar policy trick: describing a rights grab as an innovation policy. She says the industry is trying to connect copyright reform with local data-centre investment, even though the real objective is to secure access to creative work at low cost. The technical term may be “training data,” but the economic issue is the same as in any other market: who owns the asset, who decides the terms, and who gets paid.
The government’s own response to the Senate inquiry shows that the current debate is not about whether copyright exists, but how it should function in the age of AI. It says Australia is invested in the success of the creative and media industries and that it wants development and adoption of AI technologies to build trust and confidence. It also says the government is working with stakeholders to find solutions that encourage innovation while protecting and supporting creators.
“The Australian Government is invested in the success of Australia’s creative and media industries.”
That line matters because it captures the policy balancing act. The state is not saying AI should be blocked. It is saying that AI policy should not come at the expense of creators’ rights. For authors and other rights holders, the next question is whether that principle survives the pressure of industry lobbying once consultation shifts from broad statements to concrete legal design.
The Market Logic Behind The Copyright Push
The copyright argument is really an argument about pricing. AI companies want access to a very large pool of text, images and audio because that improves model quality and commercial value. But if the companies do not pay for that access, the creative sector bears the cost while the technology sector keeps the upside. That is the core reason the debate has become so contentious.
For creators, a license is not an obstacle to innovation; it is the basis of a functioning market. Publishing already operates on that model. Rights holders permit use, set terms and receive payment. The same logic applies to music licensing, film rights and many forms of visual art. AI developers are asking to be treated differently from everyone else, even though they are using the work in bulk and at scale.
That makes the government’s rejection of a text and data mining exception especially important. Such an exception would have shifted the default from permission to presumed access. It would have made copyrighted material available for machine training without the ordinary requirement to ask first. The current government line leaves that door closed, at least for now, and keeps the debate centered on licensing rather than expropriation.
There is also a practical reason creators are resisting broader exceptions. Once the law permits uncompensated use at the training stage, it becomes harder to control downstream value. The model can be built once, then deployed across countless commercial products. That means the original creator could lose leverage not just over one license but over a stream of future uses. In a digital economy, the first transaction often determines the rest.
The Attorney-General’s Department says the Copyright and AI Reference Group is the main vehicle for consultation. Its current focus includes fair, legal avenues for using copyright material in AI through licensing arrangements, certainty around copyright law for AI-generated material and cheaper enforcement options for lower-value disputes. That agenda suggests the government understands a simple point: if copyright is to survive in the AI era, creators need both clear rights and practical remedies.
The enforcement issue is not trivial. Even a strong law can be weakened if rights holders cannot afford to assert it. That is why lower-cost enforcement matters as much as the underlying rule. A right that cannot be enforced efficiently can become a right in name only, especially when infringement is widespread and technically difficult to detect.
Creators counter that the clearance problem is being overstated. If rights holders can be found and paid in publishing, music and film, then AI developers can do the same. The challenge is not impossibility; it is that permission costs money. That is precisely why companies with large training budgets would prefer a legal shortcut. A licensing market is slower and more expensive than a free-riding model, but it is also fairer and more sustainable.
“The Australian Government is working with stakeholders to find solutions to encourage innovation while protecting and supporting Australian creators.”
That statement leaves room for compromise, but it also sets a boundary. Innovation is not meant to be a synonym for unpaid extraction. If the government keeps that boundary intact, the policy signal is clear: AI development must fit within copyright, not rewrite it.
Why Australia Is Being Watched
Australia matters because it is being watched as a test case for how democracies respond when AI companies ask for access to copyrighted material at scale. If a country with a strong legal tradition and a large creative sector weakens copyright for technology firms, the decision will reverberate well beyond Canberra. If it holds the line, it strengthens the global case for licensing-based AI development.
Funder’s appeal to fairness is therefore strategic, not just moral. She is arguing that Australia’s institutions should treat creative work as property, not as a pool of data to be mined whenever new technology wants faster growth. That argument is reinforced by the government’s existing position: it has already rejected a broad text and data mining exception and is now consulting on licensing, transparency and enforcement.
The pressure to soften that stance will not disappear. Technology companies want scale. Investors reward faster model training. Governments are tempted by promises of investment and jobs. Those incentives make it easy to dress up a transfer of value as modernization. The risk is that creators end up subsidising the next phase of AI growth without consent or compensation.
The next stage of the debate will depend on the details that emerge from consultation. Licensing structures, enforcement tools and the treatment of AI-generated material will determine whether copyright remains a live market right or becomes a weakened formality. For authors, the central demand is simple: if AI systems want the work, they should negotiate for it.
That is the point of the fight in Canberra. Australia can preserve a copyright system that treats creators as owners, or it can move toward a model where the most powerful companies in the market capture value first and argue about payment later. Funder’s warning is that fairness cannot survive that shift.
What happens next will show whether the government’s promise to support creators is a principle or a placeholder. In a copyright battle this important, the real test of fairness is whether the people who make the work still control what happens to it.
Explore more exclusive insights at nextfin.ai.
