Introduction by Croakey: We recently published an in-depth examination of gaps and opportunities in the Federal Government’s plans for national standards for artificial intelligence (AI), to be driven by an Office of AI established in the Department of Prime Minister and Cabinet last week.
The analysis included comments from Professor Fran Baum, Rachel De Sain, Dr Elizabeth Deveny, Associate Professor Zerina Lokmic-Tomkins, Tan Nguyen, Professor Peter Sainsbury, Professor Nick Talley, and Adjunct Professor Tarun Weeramanthri. Publications from the Lowitja Institute, CSIRO and the Royal Australasian College of Physicians were also cited.
This week we are also covering the Royal Australian and New Zealand College of Radiologists conference on AI – #Intelligence26, which Marie McInerney previewed here.
In the Q&A with Croakey below, Kent Buse, Professor of Health Policy at Monash University Malaysia, and Co-founder and Co-CEO of Global 50/50, which works for global gender justice, stresses the importance of wide engagement with the Government’s development of AI regulation.
“The consultation on the legislation, and the National Cabinet process, will be dominated by industry and investors unless public health, climate, and community voices insist on a seat,” he says.
“…the fights for health, for decent work, for creative livelihoods, and for a safe climate are a single struggle over who sets the terms of a technology, not four separate ones.”
Interview with Kent Buse
Croakey: What else would you have liked to have heard from the PM in his speech?
Credit where it is due: the Prime Minister has acted early, and the obligations on data centres to underwrite new renewable generation and pay for their water infrastructure appear world-leading.
However, health appeared in the speech only in passing, as a cancer-screening tool, and climate only as an energy problem to be managed.
AI is fast becoming a determinant of health in its own right: it will shape who is diagnosed and who is missed, whose work is dignified and whose is degraded, what information people receive about their bodies, and whose communities carry the environmental costs of the infrastructure.
I would have liked to hear that the new Australian Standards will be subject to independent health, equity, and climate impact assessment, and that the Office of AI will carry an explicit mandate for health and equity. Standards written without a health lens will still determine health outcomes.
Finally, I would also have liked to hear who the framework will answer to. The speech promises consultation with industry and trading partners, and ‘a streamlined process for verifying compliance’, but says nothing about co-development with workers, communities, and civil society, and nothing about independent monitoring or public reporting once the rules are in place.
Standards without accountability are aspiration.
Croakey: What are the issues you think we should highlight around AI regulation and wider public health and climate health issues?
Three, all of them about power.
First, concentration: a handful of largely foreign corporations control the models, the data, and the compute, and the question for public health is the one we always ask, who stands to gain, who loses and what issues are being kept off the agenda entirely.
Second, invisibility in the data: AI systems are trained on data that under-represent women, Indigenous peoples, and marginalised communities, so they reproduce and automate existing inequities in diagnosis, triage, and access. What is not counted does not count.
Third, the commercial determinants: the same health-harming industries that perfected targeted advertising will use AI to market alcohol, gambling, junk food, and misinformation with unprecedented precision – and likely help serve the manosphere too with considerable health harms for young men and women too.
On climate, the data-centre rules are welcome, but the standards should reach the full lifecycle, including the minerals, the embodied emissions, and the global supply chains whose health costs land on communities half a world away, with no seat at the table.
Power, again: the gains are captured here, whilst the costs are exported.
Croakey: How would you like to see the public health community engage with this announcement and the Government’s plans for AI regulation?
Engage, not spectate. The consultation on the legislation, and the National Cabinet process, will be dominated by industry and investors unless public health, climate, and community voices insist on a seat.
The community should:
- make submissions arguing that health and equity be written into the mandatory standards, not bolted on afterwards
- demand transparency and disaggregated reporting so that the effects of AI on different populations can actually be seen
- build alliances with the unions, the creative sector, and the climate movement, because the fights for health, for decent work, for creative livelihoods, and for a safe climate are a single struggle over who sets the terms of a technology, not four separate ones.
At Global 50/50 we have shown that independent benchmarking of powerful organisations changes their behaviour.
The same accountability infrastructure, independent, public, and comparative, should be built around AI from day one.
Croakey: Are there any particular books, journal articles or publications on this topic to which you would refer the PM and co?
The WHO guidance on the ethics and governance of artificial intelligence for health (2021, with 2024 guidance on large multimodal models) sets out exactly the health governance architecture the speech lacked.
Its companion for the people who will actually write the rules is the European Observatory on Health Systems and Policies’ Demystifying artificial intelligence in health: what health policy-makers need to know (2026), edited by del Rey Puech, Saund, Panteli, and Martin McKee, a primer on the promise, the pitfalls, and the questions officials should be asking.
The Lancet and Financial Times Commission, Governing Health Futures 2030: growing up in a digital world (2021), makes the case for putting public value and children’s rights at the centre of digital governance.
Kate Crawford’s Atlas of AI (2021) covers the material costs, the energy, water, minerals, and labour, that connect AI directly to planetary health.
Shoshana Zuboff’s The Age of Surveillance Capitalism remains a great account of the business model underneath it all.
And Buolamwini and Gebru’s ‘Gender Shades’ study (2018) showed empirically how AI systems fail women and people of colour first.
Q: Any other issues or questions on your mind?
Two. Beyond a historical nod to suffrage, the speech was silent on gender, and our work at Global 50/50 has shown, year after year, that when the governance of powerful sectors is designed without explicit commitments to equality, it is led by, and works for, the already powerful. There is no reason to expect AI to be different unless the standards say otherwise.
And second, the global dimension: Australia is right that setting standards first confers influence, as the social media age restrictions showed.
The opportunity now is to make health-centred AI governance a global public good, working through WHO and the multilateral system, rather than a national carve-out in a race whose rules are otherwise written by the companies. It is not a zero-sum game.
• Previously at Croakey by Kent Buse: An update on the power and politics of health policy – and a call for action on health equity
See Croakey’s archive of articles on AI and health
Add Comment