AI is the technology that more than other will determine the strategic balance in the world. But who is set to control the critical layers of the stack, and how will this determine the distribution of power? This panel will consider the global dynamics of the development, dependencies, adoption and diffusion of AI.

The most advanced capabilities, notably frontier models, are conspicuously concentrated in the US and China. What are other countries doing to build their own sovereign capabilities, and ensure they maintain agency and strategic options for the decades ahead? How can tech middle powers cooperate better to complement one another’s strengths? How perilous are the slenderness of critical supply chains such as semiconductors? To what extent will open weight models provide a solution to countries that want to adopt highly capable AI – and might they equally become security and strategic risks?

Australia has natural strengths as a destination for data centre investment, including to train frontier LLMs. But how do we capitalise on our advantages and translate them into benefits to national interest in the form of secure, sovereign access to advanced AI capabilities, whether provided by trusted partners, home grown or a mix of the two.

This panel will look at how we can best achieve this sovereignty and agency, and how we handle challenges such as energy and competition from other nations. What does Australia actually need to control, and what can it depend on other trusted partners for? What does a reliable form of managed interdependence look like? How does Australia move up the value chain? How do we ensure the right conditions on foreign investment to serve Australia’s interests? What’s the right level of ambition with respect to sovereign capability including models? What are the downstream chokepoints we can occupy as niches, such as advanced manufacturing, medicine and health care, robotics and defence technologies? And what are the immediate priorities and next steps?

Australia aims to be the partner of choice in building AI capacity in the Pacific, and there are rich opportunities to collaborate with other Indo-Pacific nations. This panel will consider where the region stands on responsible AI adoption and how Australia can support it – taking account of the wide differences across the region. What are the lessons of other recent capacity building efforts? What are some of the major challenges Indo-Pacific partners face in galvanising safe, innovative AI adoption? How can Australia work with tech powerhouse nations in the region such as India?

How fierce is the strategic competition in the AI field in the region -- and on whose stack is the region likely to run? How can the Indo-Pacific work together to influence global standards and governance?

Safety and security are top of mind in the wake of a barrage of incidents and developments in 2026, starting with the revelation that Anthropic’s Mythos Preview was too cyber-capable to release publicly, followed by the announcement of agents acting autonomously to coordinate, escape and launch hacks from some of the leading labs and research institutes.

This panel will discuss key risks: malicious use by human actors, misaligned models and agents, future risks such as bioweapons. What is the attack-defence balance in the medium term? Is there any prospect of an equilibrium in a world in which “superintelligence” is widely available? If not, how do we go about protecting core interests and critical infrastructure? How worried should we be about misalignment and loss of control scenarios as models become more powerful and agents are given longer task horizons? Donald Trump and Xi Jinping are set to meet later in September , but the prevailing dynamic is one of an all-out race, with few signs that significant cooperation around safety and security are progressing. What can Australia do – and what is it doing through efforts such as the AISI – to influence the global discussion about safety and security mitigations, including on misalignment and loss of control?

This panel will consider China’s AI strategy and what it means for the world, especially Indo-Pacific nations. China is building frontier models that are very close to American leaders, but releasing them open source at a fraction of the cost. Is this sustainable, either commercially or in security terms for the Chinese government given the lack of guardrails? What are the risks that Chinese firms win the export race through adoption by other nations, especially developing nations, with consequent security and strategic/dependency risks? How does Australia factor in this – should the Australian government be comfortable with companies using Chinese models and applications, and what limits should be put for instance for critical infrastructure?

How are Chinese firms using distillation and how should the US respond? What are the current obstacles China faces, both domestically and internationally, in achieving its AI strategies, such as heavy constraints on access to advanced chips and a local chip industry that still sits somewhere behind the US?

Militaries around the world are scrambling to integrate AI to do everything from compressing decision time in operations and speed up targetting, through to improving back office tasks and helping with procurement and testing & evaluation. This panel will consider the opportunities for the ADF and for comparable militaries, and look at the lessons from around the world, notably Ukraine with its remarkable use of drones, automation and AI.

How fundamentally do militaries and their decision-making need to be reshaped and reformed around AI through concepts such as “decision-centricity”? How can we make best use of our data as a resource? From a defence perspective, what elements of the AI stack need to be “sovereign”, including models?

What are non-democratic adversaries doing – should we be concerned by the pace of their progress and the way they might deploy AI?

The growing consensus is that AI will transform institutions, from bureaucracies to justice systems to democratic processes. Just how powerful is AI likely to become -- and how quickly? If advances continue on their present trajectory, institutional structures we’ve taken for granted from economics to democracy might be superseded and need dramatic reform, offering risks and opportunities. The panel will discuss these challenges and how prepared we are, including here in Australia.