Below is the opening statement delivered by Darcy Allen (also on behalf of Chris Berg) to a public hearing of the House of Representatives Select Committee on Cyber Security for Small to Medium Sized Businesses and Organisations on 19 October 2026 at 10:00am AEDT.
Our submission, “How Australia’s AI and technology policy shapes cyber security for SMEs” is available here: https://www.aph.gov.au/DocumentStore.ashx?id=94b2b82d-c825-4e9c-92c4-3359371e1aa3&subId=801652
—
Thank you, Chair. My name is Associate Professor Darcy Allen, and I am here with my colleague Professor Chris Berg. We are academic economists at RMIT University appearing in our private capacities.
Much of the discussion about AI and cybersecurity understandably focuses on attackers. But AI can also help defenders detect threats, get advice, and respond to incidents.
Realising those benefits of AI depends on people being able to access and use capable tools.
We have experience working with teams in the cryptocurrency industry who have proactively and successfully used frontier models to identify critical threats to user funds before they were exploited, and would be happy to share those experiences with the committee.
The core of our submission is that Australia’s broader technology policy, including our emerging AI policy, shapes our defensive capabilities.
Our submission discusses examples overseas where safeguards on frontier models have blocked legitimate defensive analysis. Partly this is because defending yourself can involve discussing or analysing the same malicious code or commands used by an attacker.
Indeed, as we appear today there is an ongoing incident with the AI platform Opencode that illustrates important challenges using commercial models, open models, and cybersecurity specific program access that could inform this committee’s thinking.
That illustrates a broader concern that we have. Decisions about how we regulate AI can affect the tools people have available to protect themselves and their businesses. Policy settings that make those tools more expensive, less capable or harder to use can weaken defensive capabilities. We urge the committee to consider whether seemingly-unrelated technology policy settings risk placing our defenders at a disadvantage.
Our recommendation is to make strengthening defensive AI capability a central consideration in this inquiry. Australia’s broader technology policy should support access to capable tools and their use for legitimate defensive purposes.
Thank you very much. We welcome your questions.
