The Australian government has put on the table a plan to loosen regulations around the AI data centres rapidly multiplying across the country. This week, prominent names from the worlds of music, publishing and journalism — among them Paul Kelly, Anna Funder and Peter Garrett — pushed back against the government in a public statement. At the heart of the dispute is a single question: who should bear the cost of AI investment.

What happened?

Big tech companies have already begun building data centres in Australia. On top of that, the government is now considering speeding up planning approvals and granting a copyright exemption for AI training. Leading figures from Australia's creative industries, writing jointly in the Guardian, called on the prime minister to make a clear choice: protect the creative sector, or bow to the demands of big AI companies.

Why does it matter?

This isn't a story unique to Australia. Ireland, thanks to low corporate tax rates and fast-tracked planning approvals dating back to the 2000s, became Europe's data centre hub; today, grid connections around Dublin are already constrained by the load data centres consume. Virginia's Loudoun County went down a similar path: investment drawn in by local tax breaks turned the region into an infrastructure hub carrying a significant share of global internet traffic. In both cases the pattern is the same: once the infrastructure is locked in, governments find it hard to tighten the rules afterward, wary of scaring off capital. Australia is at exactly that threshold now — companies have already started building while the government is still debating which concessions to grant.

The critical difference this time is that it's not just land, water and power allocation on the table. AI companies are also seeking a text-and-data-mining exemption under copyright law — the right to use copyrighted music, books and news content to train models without paying licensing fees. That cuts against Canberra's own track record: Australia set a global example with its news bargaining code, which forced news platforms to pay publishers. For the same government to now grant AI companies a blanket exemption would mean giving up that hard-won point of leverage with its own hands.

Who does this affect, and how?

PartyShort-term gainLong-term risk
AI companiesFast approvals, free training dataLocal backlash, reputational cost
Artists, writers, journalists—Loss of copyright income and data control
Local communitiesJobs, investment announcementsWater and power costs, grid strain
GovernmentInvestment headlinesErosion of tax and regulatory leverage

What's on the table isn't a single regulation but the combination of two separate concessions: ease for physical infrastructure and immunity for intellectual property. Granting both at once means the local economic benefit generated by a data centre and the income lost by copyright holders are never weighed in the same equation — the party making the decision and the party footing the bill are not the same.

What's next?

A similar front formed in the UK last year: musicians built public pressure against the government's proposed AI copyright exemption, and London shelved the plan — though without closing the door on it entirely. Australia looks likely to follow a similar trajectory — a formal consultation process, the creative sector exercising its political weight, and probably a watered-down, temporary exemption rather than an outright rejection. The real indicator to watch is what recommendations the government receives from independent review bodies (such as productivity-commission-style institutions), and when the prime minister breaks his silence on this public statement.

What the government stands to lose here isn't just a line item in a regulation, but the negotiating moment itself. Once data centres are built, walking back a concession becomes nearly impossible; the real leverage exists in these weeks before the concrete is poured. A government that fails to settle copyright and data rights within this window will sit down at the same table with the same companies tomorrow in a far weaker position.