Australia’s decision to establish an Office of AI is a significant step. It recognises that artificial intelligence can no longer be managed as a collection of separate technology projects spread across government departments.
AI is already influencing employment, education, creative industries, public services, national security and critical infrastructure. It is also creating growing demand for data centres, energy and water.
A coordinated national response is overdue.
But establishing an office is the easy part. The more difficult task is deciding what that office will stand for, and whose interests it will serve.
I believe the Office of AI should be judged against one central test:
Does Australia’s approach to AI create measurable public value, or does it simply make it easier for technology companies and infrastructure developers to operate?
Australia should welcome responsible investment. It should also recognise that investment alone is not a national AI strategy.
The public debate about AI is often framed as a choice between innovation and regulation.
That is the wrong choice.
Poor regulation can restrict useful innovation, but weak governance can also undermine it. Businesses need clear rules. Workers need confidence that technological change will not happen to them without consultation. Creators need certainty about how their work can be used. Communities need to know that new infrastructure will not transfer private costs onto the public.
Good governance is not the enemy of innovation. It is what gives innovation legitimacy.
The Office of AI should therefore be assessed against four principles: authority, transparency, participation and shared benefit.
The first question is whether the Office of AI will have the authority to influence decisions across government.
AI touches numerous portfolios, including industry, employment, education, defence, energy, the environment and the arts. Coordination will be valuable, but coordination without authority can easily become another layer of meetings, reports and voluntary guidance.
The office must be able to establish consistent expectations across departments. It should also have a clearly defined role in reviewing high-impact uses of AI, particularly when automated systems affect employment, healthcare, credit, insurance or access to government services.
If the office identifies an unacceptable risk, it must be able to do more than publish advice.
Australians should know when AI is being used to make or influence important decisions about them.
Government agencies and businesses should be required to disclose consequential uses of automated decision-making. They should explain what information is being used, how risks have been assessed and how affected individuals can challenge an outcome.
The proposed Australian standards for AI must also be public and understandable.
It is not enough to announce that standards will exist. Australians need to know whether they will be mandatory, which organisations they will cover, who will enforce them and what will happen when those standards are breached.
Standards without disclosure, enforcement and appeal rights risk becoming little more than reassuring language.
AI policy cannot be designed exclusively by government officials and technology companies.
Workers, unions, educators, researchers, creators, disability advocates, consumer groups and affected communities must have a meaningful role in shaping the standards.
This is particularly important in the workplace.
AI can improve productivity by reducing repetitive work and helping employees make better decisions. It can also remove roles, weaken entry-level career pathways and increase workplace surveillance.
Employers should be required to consult workers before introducing AI systems that significantly change roles, performance management or staffing requirements.
Australians should have a voice before these systems are deployed, not only after harm has occurred.
The government must be able to show how the economic benefits of AI will be shared.
Productivity gains should lead to better services, stronger businesses, higher-value work and improved living standards. They should not flow exclusively to technology providers, infrastructure owners and shareholders.
The same principle should apply to data centres.
Data centres may attract investment and support Australia’s digital capacity, but they can also consume substantial amounts of electricity and water. Faster approvals should be accompanied by clear obligations covering environmental performance, infrastructure costs and benefits for local communities.
If a project receives faster approval or public support, the public should receive a measurable return.
The treatment of Australian creative work will be an important early test of the new office.
Generative AI systems are trained using enormous quantities of data, potentially including copyrighted writing, journalism, music, visual art and other creative material.
Australian creators deserve clear rules about consent, attribution and compensation.
The government should resist framing copyright protection as an obstacle to innovation. Creative work has economic and cultural value, and its creators should not be required to subsidise the development of commercial AI systems without permission or payment.
A credible AI framework must protect innovation and creative rights at the same time.
The Office of AI should not focus solely on controlling risk. It should also help Australia identify where AI can produce genuine national benefits.
This could include better healthcare administration, more accessible public services, improved disaster preparation, support for regional industries and tools that help small businesses become more productive.
The goal should not be to adopt AI wherever possible. It should be to use AI where it solves a real problem and improves outcomes for Australians.
That distinction matters.
Technology should remain a means of delivering public value, not become an objective in itself.
The creation of an Office of AI gives Australia an opportunity to build a coherent national approach to one of the defining technologies of this century.
But the office should not become Australia’s AI marketing department. Its purpose cannot be limited to attracting investment, accelerating data-centre approvals or demonstrating that the government is keeping pace with technological change.
It must have real authority. Its decisions must be transparent. The people affected by AI must participate in its governance. The benefits must be shared.
Those four principles - authority, transparency, participation and shared benefit, provide a practical way to judge whether the Office of AI is succeeding.
The real measure of success will not be how quickly Australia adopts AI.
It will be whether Australia can shape AI in a way that earns public trust, strengthens the economy and improves people’s lives.