Prime Minister Anthony Albanese is expected to unveil a new “Office of AI” in a major address in Sydney today, establishing a coordinating body inside the Department of the Prime Minister and Cabinet to manage AI standards development across the entire federal government. The move, confirmed by Reuters ahead of the speech, marks Australia’s most concrete institutional step yet in a policy debate that has been building for two years: how to attract AI investment and adoption without leaving the public exposed to the technology’s risks, in a country that has deliberately chosen not to write a dedicated AI law.

A Coordinated Front Door for AI Policy

Albanese’s framing of the new office is explicitly about fixing fragmentation. “Up until now, our response has been issue-by-issue, sector by sector,” he is expected to say, drawing a historical comparison to how earlier governments built coordinated national responses to civil aviation in the 1920s and genetics in the 1990s. The Office of AI will sit at the center of government rather than inside a single department, intended to give AI policy a “whole-of-government” thread running through ministries that have so far handled AI issues separately, from financial services to health to competition policy. Australia is describing the approach as a world-first, and the government’s own framing suggests the office is meant to function as much as an investment-attraction tool, giving companies a single, clearer point of contact for approvals, as it is a governance mechanism.

National AI Plan’s on Investment Over Guardrails

The Office of AI builds directly on the National AI Plan, unveiled December 2, 2025, which marked a deliberate pivot in Australia’s regulatory posture. Earlier government consultation, dating back to a September 2024 proposal paper, had floated mandatory guardrails for AI use in high-risk settings, covering governance, risk management, data quality, testing, and human oversight. The National AI Plan stepped back from that mandatory framing, opting instead to continue building on Australia’s existing legal and regulatory frameworks rather than creating a new standalone regime, a shift the IAPP has described as a move from a previously more safety-oriented posture toward one weighted more heavily toward investment and economic opportunity.

The centerpiece of that shift is the Australian AI Safety Institute, which became operational in early 2026 with roughly AUD $29.9 million in government funding. Rather than enforcing binding rules, AISI’s role is to monitor, test, and analyze advanced AI capabilities, risks, and harms, feeding independent technical advice to regulators and ministers rather than issuing rulings itself. That advisory model reflects a broader judgment call the government has made explicitly: the Productivity Commission’s own review this year recommended against AI-specific regulation except as a last resort, warning that overly cautious rules could leave Australia economically worse off relative to faster-moving competitors, a concern echoed throughout the National AI Plan’s emphasis on data center investment, currently forecast to exceed AUD $100 billion, and Australia’s positioning as a regional AI and data hub.

The Tension Underneath the Consensus

Not everyone in Australia is convinced this lighter-touch approach is the right one, and the disagreement runs along fairly predictable lines. Treasurer Jim Chalmers rejected union demands for immediate workplace AI regulation in mid-2025, telling the Australian Financial Review the government was “overwhelmingly focused on capabilities and opportunities, not just guardrails” language that captures the philosophy behind the National AI Plan about as directly as anything in the plan itself. Unions have pushed back on that framing, and Australia’s Senate Select Committee on Adopting AI separately recommended the country move toward mandatory guardrails for high-risk AI applications, a more cautious position than the government’s own current stance.

Public sentiment adds another layer of tension to the debate. Survey data cited in recent regulatory analysis found only about 30% of Australians believe AI’s benefits outweigh its risks, while nearly 80% express worry about its negative impacts. That gap, between a government strategy explicitly betting on investment and adoption, and a public that remains substantially more cautious than its own government, is arguably the real story underneath today’s Office of AI announcement: an institutional fix for policy fragmentation doesn’t by itself resolve the deeper question of whether Australia’s regulatory philosophy actually matches what most Australians want.

What’s Already Legally Binding vs What Isn’t

It’s worth being precise about what’s actually enforceable in Australia right now, since the National AI Plan itself creates no new legal obligations. AI use is currently governed through a patchwork of existing, technology neutral laws: the Privacy Act 1988, the Australian Consumer Law, and the Online Safety Act, overseen by a distributed set of regulators including the OAIC, ACCC, ASIC, and APRA, each applying their own sector-specific expectations rather than a unified AI standard. The most concrete new binding obligation on the horizon arrives December 10, 2026, when Privacy Act amendments take effect requiring organizations to clearly explain automated decisions that significantly affect individuals’ rights or interests, covering AI use in hiring, lending, insurance, and customer analytics specifically. Everything else, the National AI Plan, the Voluntary AI Safety Standard, the AI Ethics Principles, remains guidance rather than law, which is precisely the model Australia’s government has chosen to build around rather than legislate directly.

How Australia’s Approach Compares

Australia’s bet stands in fairly sharp contrast to the European Union’s AI Act, which established binding, risk-tiered obligations for high risk AI systems directly in statute. Australia has instead pursued compatibility, not full alignment, with international frameworks, building bilateral AI safety cooperation with the US, UK, Singapore, India, and South Korea, and signing a trilateral technology partnership with Canada and India this month. That positioning, engaged internationally, but resistant to binding domestic legislation, mirrors the same investment-first philosophy driving the National AI Plan and, now, the Office of AI.

The Bottom Line

What makes today’s announcement worth watching isn’t the creation of another government office, it’s what the office is explicitly not: a regulator with new enforcement powers, or a step toward the mandatory guardrails the Senate committee and various unions have pushed for. Australia has made a clear strategic choice to compete for AI investment through coordination and clarity rather than through binding rules, and the Office of AI is best understood as the administrative machinery built to execute that choice more smoothly, not a signal that stricter regulation is coming. Whether that bet pays off will depend less on how efficiently the new office coordinates ministries and more on whether the gap between government confidence and public wariness, roughly 50 percentage points wide by the polling cited above, narrows or widens as AI’s presence in daily Australian life continues to grow.