AI’s rapid rise and online harms with eSafety Commissioner Julie Inman Grant

On this episode of Technology & Security, Dr Miah Hammond-Errey is joined by Professor Elke Schwarz, Professor of Political Theory at Queen Mary University of London and author of Death Machines: The Ethics of Violent Technologies. When a machine helps decide who dies, who is left holding the moral responsibility? Elke explains how AI-enabled targeting acts as a cognitive prosthesis, why operators working within a 20-second decision window tend to over-trust system outputs, and how the moral friction between a decision and its consequences is being eroded. The conversation examines why "meaningful human control" has given way to a more technical logic "context-appropriate human judgment and control," and what an expanding targeting appetite means for civilians in conflict. Elke traces how venture capital's growth-driven timelines are pushing defence procurement toward speed and scale, and asks who bears the cost when defence starts to look like Silicon Valley. The episode closes on what it takes to build a more intentional, deliberate AI future, and why philosophy, ethics and social science belong in the room for security leaders, not on the margins. Elke Schwarz is Vice-Chair of the International Committee for Robot Arms Control and a current British Academy Global Innovation Fellow at the German Council on Foreign Relations. Her research examines the ethics of war and technology, including how venture capital dynamics are shaping military norms and defence spending.

On this episode of Technology & Security, Dr Miah Hammond-Errey speaks with Rohan Samaraweera, former AI and analytics lead at the Department of Home Affairs. The conversation opens with how AI conversations inside government have shifted since the arrival of ChatGPT. They discuss Hermes: how predictive analytics built for the international mail stream helped identify hundreds of millions of dollars in illicit drugs, and how they calculated the platform's broader social impact at roughly two billion dollars. They discuss why AI's return on investment so often lands on someone else's balance sheet, what Robodebt teaches government about governance and accountability, and whether Australia's new Office of AI can join up that picture across agencies. Rohan makes the case that access to frontier AI is a dependency, not an asset, and explains what genuine sovereign capability requires beyond a licence or a locally built data centre — drawing on his own experience at ASD and Home Affairs reducing reliance on contractors and vendors. He and Miah cover the diffusion of power between states and multinational technology companies, the opportunity for Australia to build small, reliable, high-trust AI systems rather than compete at the frontier, and what the country needs to invest in now to close the gap with allies. The conversation closes with why human expertise remains indispensable to AI deployment, illustrated by a striking anecdote from an international mail centre, and Rohan's take on the "one yes versus one no" problem facing anyone trying to innovate from inside a large institution. Rohan Samaraweera was the enterprise AI and analytics lead at the Department of Home Affairs, and a 16-year veteran of the Australian Signals Directorate. He recently founded Emrest AI and spent a year at Harvard Kennedy School's Belfer Center researching AI and national security.

On this episode of Technology & Security, Dr Miah Hammond-Errey speaks with Professor Jon Whittle, founding CEO of Goldilocks AI and professor at the University of Melbourne's Institute for Digital Innovation and AI. With nearly 30 years in AI spanning a technical lead role at NASA, founding the National AI Centre, and five years leading CSIRO's Data61, Jon brings a long view and the practical experience of having worked on these questions from inside government, research, and industry. The conversation opens with what AI sovereignty actually means across the full stack, from data centres and foundation models through to inference, applications and responsible AI, and why focusing only on data storage creates a false sense of security. The conversation moves to the economic and geopolitical case for Australia building its own foundation models, why the argument that we are too small and too far behind deserves more scrutiny, and how being cut off from models like Fable has made that conversation more urgent. Jon and Miah discuss the gap between personal productivity gains from AI and the organisational productivity gains that remain elusive, the dimensions that actually determine enterprise AI success (most of which have nothing to do with technology), and the leadership anti-patterns holding Australian companies back. This episode also covers the rise of Chinese models, the risk of ‘Uberization’ of the Australian economy, what meaningful AI oversight requires from directors and boards, and why the conversation Jon hopes we are not still having in five years is the one we are having right now. Professor Jon Whittle is founding CEO of Goldilocks AI, professor at the University of Melbourne, and author of AI for Business. He is a former director of CSIRO's Data61 and founder of the National AI Centre.

On this episode of Technology & Security, Dr Miah Hammond-Errey is joined by Nicolas Miailhe, expert at the intersections of AI Safety and governance. Velocity shock refers to the gap between how fast frontier AI is being deployed and how slowly the institutions designed to govern it are moving. This episode explores what meaningful sovereignty is and what it means across the AI stack, the reorganisation of alliances now two dominant AI powers have solidified, and what it means for middle powers including Australia, Canada and the EU. The conversation covers why rigorous evaluation of frontier AI remains scientifically unsolved, why voluntary safety frameworks from labs are insufficient under competitive pressure and practical steps toward closing the technology use and governance gap. Nicolas Miailhe is co-founder of AI Safety Connect, CEO of Prism Eval, and former founding CEO of The Future Society. He is an AI expert for the OECD and UNESCO's AI ethics group.

Cybersecurity is often framed as a technology problem. Andrew Reeves argues it is fundamentally a human one. Dr Miah Hammond-Errey sits down with Dr Andrew Reeves, Deputy Director of UNSW's Institute for Cyber, registered organisational psychologist and cybersecurity leader, to examine what psychology reveals about attack, defence and victimhood in cybersecurity. From the power of cyber deception, to why security awareness training can backfire, to what cognitive load means for the people defending our networks, this is a conversation about the human factors that determine whether cybersecurity actually works. Andrew and Miah discuss the collaborative research project between UNSW's Institute for Cyber and Strat Futures, mapping the cybersecurity implications of emerging technologies in Australia over the next two to five years. Andrew shares what the data is revealing about the convergence of AI, biotechnology and brain-computer interfaces, why the most critical developments will come from how technologies interact rather than any single breakthrough, and what the tension between sovereign capability and international collaboration means for Australian organisations. They also discuss cyber security lessons from the golden age of piracy, the psychology of leadership under fatigue, and what forest bathing has to do with making better decisions.