Commentary on: Defending Australia

I'm fairly certain that this report will not figure in best sellers' lists. Whilst it addresses the social issue du jour, AI, it is not targeting a popular audience. Its strategic think tank
fodder, so politicians, analysts and some serious journalists will be forced to read it as punishment for their bad life choices. Its not about democratic possibilities, its about influencing government. I'm sure that many on the "progressive" side of politics will eschew the whole subject as distasteful, and the report itself as tedious given it does not
address questions of identity or sexuality. It doesn't even discuss copyright.

Coming as it does from the conservative right of the Labor Party, 'a priori' assumptions
provide a foundation for a dour, procedural continuation of social democratic capitalism with continuous invocation of Australian sovereignty as motivation for industrial policy on AI. As one might expect, a key focus is national security and defence capability.

What is a little surprising is an openness to recognise the divergence in interests between the US and Australia, without stating that explicitly. Examples of malfeasance
typically use examples of Chinese origin, but the report is assiduous in using the word "foreign" as a descriptor when analysing risk.

And given the increasing disruption of global trade and diplomatic relations concurrent with the massive investment in AI compute capacity, it is timely.

Awfully, I find myself in agreement with the recommendations of the paper, even though I might quibble around some analysis and the over emphasis on defence capability
even as my aging idealism is struggling to hold on. The forces arrayed against disarmament lead me to believe that a detour through establishing defence
self-reliance is a depressing necessity. I don't like this but my material analysis indicates that TINA.

TLDR;

• AI is here and is now essential to bureaucratic, financial, public utility infrastructure and defence
• Australia has no domestic AI capability
• Access to AI capabilities, including more advanced capabilities is controlled by corporations and governments that are only minimally subject to Australian political influence.
• There is a kill switch on Australian access (a kill switch is a method available to disable the system that can be operated by an external interest)
• The absence of a domestic AI capability increases vulnerability to antagonistic external actors
• dependence on foreign AI capability increases the political leverage of those governments and corporations to require alignment and extract financial imposts
• training of LLMs (large language models) on foreign sources for use in AI undermines and diminishes the idiosyncrasies of Australian culture
• Controls of AI can be manipulated to introduce subtle values emphases - without access to these controls they are opaque and do influence the political and ideological bias of AI systems
• Unionised workers are a key element in enabling democratic governance of AI
• the overriding recommendation is that government act to build a national sovereign AI Capability

Other comments:

What is perhaps a glaring omission is that there is no consideration of engagement with the Chinese AI capability. I assume that this is a tactical political decision to avoid incurring the wrath of the spooks at ASPI and or the US itself. There is no attempt to analyse the drift from a global rules based order, the "polycrisis" or China's emergence. These are tacitly accepted.

Of note for some will be the embedded defence of AUKUS as a strategic necessity to maintain access to US technology and assistance.

https://curtinrc.org/wp-content/uploads/2026/07/Defending-Australia-Policy-Report-No.-14-v3.2-2.pdf