The question of who gets paid when an artificial intelligence system learns to write like a novelist or compose like a songwriter has moved from theory to a genuine fight, and Australia’s creative community has decided it wants a cheque.
Authors, musicians and the organisations that represent them are pressing the federal government to force AI developers to compensate creators whose work has been used, without permission or payment, to train the large language and generative models now reshaping entire industries. The demand, reported by the Sydney Morning Herald, boils down to a blunt principle: if a machine profits from a lifetime of somebody’s craft, that person deserves a share.
How the argument arrived here
For most of the past two years, the debate about generative AI in Australia has been dominated by the technology itself, the data centres being built to run it, the productivity gains ministers hope it will deliver, and the jobs it might reshape. The creative sector has watched from a distance as models trained on vast troves of text, images and audio grew capable of producing passable prose, illustration and music in seconds.
Much of that training material was scraped from the open internet, and a good deal of it was copyrighted. Investigations overseas have shown that pirated book libraries and lyric databases found their way into the datasets behind some of the best-known systems. For writers and musicians, the implication is uncomfortable. The very output that now competes with them was built, in part, on their own back catalogue.
That grievance has now crystallised into an organised push. Peak bodies representing authors and composers argue that the current arrangement amounts to the largest uncompensated use of creative work in living memory, and they want the law changed so that permission and payment become the default rather than the exception.
The news
The specific demand is for a licensing and remuneration regime, a framework under which AI companies would have to seek consent to use copyrighted material for training and pay for the privilege. Supporters point to the collecting-society model that already governs radio play and public performance, where musicians receive royalties every time their work is broadcast. Extend that logic to AI, they argue, and creators would finally be paid when their work becomes the raw material for a commercial product.
The timing is deliberate. The federal government has been weighing how copyright law should adapt to AI, and creators are determined not to be an afterthought in a process they fear is being shaped largely by the technology industry and its promises of economic growth. Their message to Canberra is that innovation cannot be funded by quietly transferring value from artists to platforms.
Two sides of a hard question
On one side sit the creators and their representatives, who frame this as a straightforward matter of property and fairness. Their view is that copyright already protects a book or a song from being copied and sold, and that feeding it into a training set is simply a new form of use that should carry the same obligations. Without payment, they warn, the economics of a creative career, already precarious for most, become untenable.
On the other side sit AI developers and some economists, who argue that training a model is not the same as copying a work, and that requiring individual licences for billions of data points would be impossibly complex and would hand an advantage to overseas competitors operating under looser rules. Some have pushed for a text and data mining exception that would allow training on copyrighted material without permission, a proposal that has alarmed the creative sector. The productivity case is real, they contend, and locking up training data risks leaving Australia behind.
Between those positions lies a genuine policy dilemma. Move too far towards the creators and Australia could make itself an unattractive place to build AI. Move too far towards the developers and the country risks hollowing out the cultural industries that give it a distinct voice.
What it means for Australia
This is not an abstract international dispute imported for local colour. Australia has a substantial creative economy, and its writers, composers and performers already operate in a small market where incomes are thin and streaming has compressed returns. If generative systems can produce serviceable substitutes for commercial music, advertising copy or genre fiction, the competitive pressure falls hardest on exactly the people whose work trained those systems in the first place.
There is also a sovereignty dimension that fits the broader national conversation about whether Australia is building AI or merely hosting it. Australian stories, idiom and music carry cultural weight that generic, globally trained models do not reliably capture. A licensing regime would not only compensate creators, it would give local rights-holders leverage to keep Australian voices in the loop as the technology matures. The alternative is a future in which the country’s cultural output is absorbed into overseas models and sold back to Australians, with none of the value returning to the people who made it.
The question also lands amid an already tense debate inside government about copyright and data centres, one that has exposed differences within Labor over how hard to lean into the AI build-out. Adding the demands of authors and musicians to that mix sharpens the choice ministers face between courting investment and protecting creators.
What is next
The immediate battleground is the government’s copyright and AI policy work, where creators will push for consent and payment to be written into any framework, and technology interests will argue for exceptions that keep training data flowing. Expect the collecting societies to press their case that a royalty model is both workable and precedented, and expect developers to counter that the administrative burden would be crushing.
Whatever Canberra decides will be watched closely overseas, where similar fights are playing out in courts and parliaments. For Australia’s authors and musicians, the goal is simpler than the politics surrounding it. They want the principle established that when a machine learns from their work and a company profits, someone writes them a cheque.
Sources: Sydney Morning Herald.
