The tension between Australia’s creative economy and the global technology companies racing to build ever larger artificial intelligence models has moved firmly into political territory, with a federal minister publicly warning the biggest players in the industry not to treat local intellectual property as free raw material.
The warning, reported this week, lands at a moment when governments around the world are being pushed to decide whether the text, images, music and code scraped to train generative models should be paid for, licensed, or simply taken. The Australian position, as the minister framed it, is that the country’s writers, musicians, photographers, journalists and software developers should not have their life’s work absorbed into commercial products without their knowledge or a cheque.
How the fight got here
For most of the past three years, the large language and image models that power tools like ChatGPT, Gemini and Midjourney have been trained on vast troves of material harvested from the open web. Much of that material is protected by copyright, and very little of it was licensed in advance. The companies building these systems have generally argued that ingesting publicly available content to train a model is a transformative act that should sit outside traditional copyright rules, an argument that has been tested in courts in the United States and Britain but never comfortably resolved.
Creators see it very differently. To a novelist whose backlist appears in a training set, or a session musician whose recordings help a model generate a passable imitation, the process looks less like inspiration and more like uncompensated extraction. That gap in worldview is what the minister was speaking to, and it explains why the language has hardened from polite consultation to something closer to a line in the sand.
Australia has been circling this question for more than a year. The Productivity Commission floated the idea of a text and data mining exception to copyright law, a change that would have made it easier for AI developers to use protected works, and the reaction from the creative sector was immediate and hostile. Peak bodies representing authors, musicians and screen producers argued that such an exception would effectively legalise the very practice creators were fighting, and would hand the economic upside to a handful of offshore technology giants while leaving Australian artists with nothing.
Two sides, dug in
The creative industries have been consistent. Organisations such as APRA AMCOS, which represents songwriters and composers, and the Australian Society of Authors have spent the past two years pressing for a consent, credit and compensation model, sometimes shortened to the “three Cs”. Their case is straightforward: if a company profits from a model trained on protected work, the people who made that work should be asked first and paid fairly. They point to the licensing deals that some news publishers and record labels have already struck with AI firms as proof that a market can exist when the will is there.
The technology industry counters that the sheer scale of modern training data makes individual licensing impractical, and that an overly restrictive regime would push AI development, investment and jobs to jurisdictions with looser rules. On that reading, a hard Australian line risks leaving the country as a consumer of foreign AI rather than a builder of its own, a concern that carries weight given how little frontier model development happens onshore. The industry’s preferred outcome is a broad exception with, at most, an opt-out mechanism, which creators reject because it places the burden on them to police a global scraping operation they cannot see.
The minister’s intervention signals that the government is, for now, siding with the creators rather than the platforms, at least rhetorically. Whether that translates into legislation, a licensing framework, or simply a firmer negotiating posture with the big American labs remains the open question.
What it means for Australia
The stakes here are unusually concrete for a country of Australia’s size. Local creative industries employ hundreds of thousands of people and contribute billions to the economy each year, and much of that value sits in exactly the kind of content that trains generative models. If the material can be taken without payment, the risk is that Australian creators subsidise the profits of overseas technology companies while watching AI-generated substitutes compete with their own output.
There is also a sovereignty dimension that fits a pattern FluentSea has tracked repeatedly this year, from debates over sovereign large language models to the scramble to build domestic data centre capacity. A nation that cannot protect the value of its own cultural and intellectual output has less leverage in the broader contest over who controls the AI stack. For news publishers in particular, the licensing question is existential, since their journalism is both highly valuable to model builders and easy to scrape at scale.
At the same time, the government has to weigh the reality that Australia is a small market largely dependent on AI tools developed elsewhere. Push too hard and the fear, voiced quietly across the local technology sector, is that global firms simply exclude Australia from early access to new products or scale back investment. The minister’s challenge is to defend creators without pricing the country out of the technology that is reshaping every industry it touches.
What happens next
The immediate test will be whether the warning is followed by policy. Options on the table range from a statutory licensing scheme that would compel AI firms to pay into a collective pool, to transparency rules forcing companies to disclose what they trained on, to a narrower copyright reform that codifies consent as the default. Each carries trade-offs, and each will be fought over hard by both camps.
For now, the message from Canberra is that the era of quietly hoovering up Australian creativity is meant to be ending, and that the largest AI companies in the world should not assume the country’s intellectual property is theirs for the taking. The far harder work, turning a warning into an enforceable framework that satisfies both artists and an industry moving faster than any regulator, is still ahead.
Sources: GNews · AI copyright-creators AU.
