Australian researchers urge mandatory rules for data centres and AI safety

October 14, 2026, 01:43Source: ARC Centre of Excellence for ADM+S

The ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S) says voluntary commitments from technology companies are not sufficient for AI infrastructure and should be backed by mandatory safeguards. The centre made the argument in a submission to the Australian Government's consultation, Getting it right: Building AI infrastructure that works for Australia.

Key points

ADM+S made a submission to the Australian Government's consultation Getting it right: Building AI infrastructure that works for Australia.
The centre says voluntary commitments from technology companies are not sufficient and should be backed by mandatory safeguards.
It calls for mandatory transparency and reporting, environmental standards for energy, water and land use, and community participation in planning.
On AI safety, it recommends independent oversight, incident reporting, safety evaluations and accountability mechanisms.
The submission was led by Kimberlee Weatherall with Henry Fraser, Christine Parker, Jake Goldenfein and Jacky Zeng.

Growth must not come at the expense of communities

According to ADM+S, Australia has a critical opportunity to establish strong, nationally consistent standards for AI safety and AI infrastructure. The submission warns that rapid growth in data centres and AI development must not come at the expense of communities, the environment or public trust.

The centre responded to two parts of the consultation paper: Part 1, Managing growth in large data centres and the AI training opportunity, and Part 3, Conditions for AI training. ADM+S supports the development of national AI standards, but argues that they need real enforcement behind them.

The submission calls for the following measures:

mandatory transparency and reporting requirements
robust environmental standards for energy, water and land use
meaningful community participation in planning processes
stronger protections for communities affected by the rapid expansion of data centres and AI infrastructure

A broader view of AI safety

On AI safety, the researchers argue that Australia needs to look beyond mandatory incident reporting. In their view, the aim should be to ensure that emerging safety issues are mitigated and that harms are remedied.

The submission says AI safety means more than technical risks. It also covers privacy, discrimination, human rights, security and environmental impacts.

The authors also recommend moving beyond voluntary commitments by AI companies. They propose mandatory governance measures, including independent oversight, incident reporting, safety evaluations and accountability mechanisms.

Sovereign capability and who benefits

The submission also highlights the importance of building sovereign AI capability. It points to investment in Australian research, skills development and public-interest infrastructure as ways to ensure that the benefits of AI training and innovation are shared across Australian society.

It was led by Kimberlee Weatherall, with Henry Fraser, Christine Parker, Jake Goldenfein and Jacky Zeng. According to the centre, the document examines how Australia can balance investment in AI infrastructure with environmental sustainability, community interests and public accountability. It also proposes stronger governance of AI training through mandatory safety safeguards, transparency requirements and independent oversight, along with measures to ensure AI development delivers benefits for Australian research, innovation and workforce capability.

The full submission is addressed to the Department of the Prime Minister and Cabinet's consultation.

What it means

The submission adds a research-led voice to the debate over how Australia should regulate the build-out of data centres and AI systems. Its central position is that voluntary industry commitments should give way to mandatory rules covering transparency, environmental limits, community input and independent oversight. The source does not say how the government will respond to the consultation.