I’ve just finished reading a fantasy novel written entirely by an AI. The Second Son of the House of Bells is the story of a young bellmaker whose keen musical ear makes him an agent of political revolution in a world in which legal transactions, and hence power, are mediated through song.

Like much about AI in 2026, the novel was imperfect but a big improvement on the technology from even a year ago and a harbinger of things to come. Crucially, the AI grasped that a good story needs a hero who overcomes obstacles in pursuit of a goal and changes both himself and the world along the way. In another five years, we could see an AI-authored novel that to an average reader resembles the heights of Moby-Dick or Middlemarch.

Australia’s content-creating industries should take note as they fight to preserve the country’s strict copyright laws at all costs. Right now, they’re battling to protect their modest turf as Neanderthals once tried to fend off encroaching Homo Sapiens. Their case is worthy but blinkered to the reality that the rapid advance of AI models isn’t going to stop. The better pathway for Australia is to establish itself as an indispensable part of the AI revolution and, in parallel, find new ways to protect artists while raising audiences that make intelligent, informed decisions about what they value in books, films, art, music and media.

The best way to carve out a role for ourselves is through the construction of data centres that train and operate AI models. Under Australia’s restrictive copyright laws, AI companies are reluctant to train their models here for fear of committing infringements and hence are cautious about investing.

Right now, our copyright laws aren’t helping anyone ^[1]. Tech companies can already include most Australian content in the enormous data sets on which AI models are trained to set their weights – the neural connections that enable the model to think. They’re just doing it in data centres overseas.

AI giants shouldn’t pay nothing. That would hasten the shift in balance from the already precarious economic potential of human creators to the growing economic momentum of AI creators. Whatever compromise arrangement the Australian government comes up with isn’t going to halt that global shift, but it will help buy our artists and media companies time while we figure out how to preserve our culture and modernise our information landscape for this new reality.

A framework in which AI companies pay into an Australian fund that is distributed among copyright holders by a government-backed agency makes most sense. The Australian public seems to get this. A survey commissioned ^[2] by the non-profit Good Ancestors and conducted by YouGov found that 61 percent of respondents supported changing copyright arrangements while still supporting Australian creators if that meant enabling AI training in Australia and hence encouraging investment. Only 15 percent wanted laws to be kept in their current form.

Encouraging data-centre investment for frontier AI will attract clusters of innovation, strengthen Australia’s strategic position in its neighbourhood by making it a compute hub, and earn it leverage in discussions about AI global governance. We can build applications at the higher end of the AI stack, but our immediate natural strength is computing infrastructure, given our land, renewable energy, political stability and geostrategic depth.

Some of the proceeds of the economic activity this generates – which consulting firm McKinsey & Company this month estimated at A$80 billion a year ^[3] from 2030 – could then be directed to support our creative industries.

Fixing the copyright issue to create a relatively low-friction environment in Australia for data-centre investment is therefore the number one priority.

Over the long term, the challenge is to recognise that these are categorically new problems, requiring a shift in mindset, laws and institutions. Increasingly, AI-generated output will compete directly with human creations in a content market.

Memorable though singer-songwriter Nick Cave’s description of composition as an ‘act of self-murder’ was, the audience also gets a vote. The literary world has already had several Turing-test moments, most famously the withdrawal of the US horror novel Shy Girl ^[4] after the publisher discovered AI was used in the writing.

Should we care whether a novel or a song arose from authentic human experience, including suffering, if it resonates meaningfully with an audience? Could we accept a chatbot posing as a human to extract leaked government information if it exposes a scandal in the public interest? What’s the future for elected politicians’ speeches if their staff are using AI in the drafting because AI can do a better job?

The answers to these and thousands of similar questions will depend on the values we apply, which in turn will vary from country to country.

The way we’re entertained, the way we deepen our understanding of ourselves and others by consuming art, the way we arm ourselves with information so that we can usefully participate in our democracy – all of this is changing.

Australia won’t protect itself by building a fragile moat through measures such as copyright. The sooner we accept that AI advances are real and continuing, and that this is going to fundamentally change content industries, the sooner we can prepare for that future – and do so from a position of relative strength by being part of the AI industry rather than resisting it.

This article was originally published in ASPI’s Cyber and Tech Digest ^[5].