CNAS’s Janet Egan on AI security and Australia’s big opportunity

Janet Egan is a senior fellow and deputy director of the Technology and National Security Program at the Center for a New American Security, and one of the most respected thinkers on AI and national security — globally and from an Australian perspective. She joins STW at the perfect moment after PM Anthony Albanese’s AI speech and the big news about OpenAI’s agents going rogue.The conversation opens with the OpenAI hack of online AI repository Hugging Face—which Janet calls a very big deal—and works through the implications for cybersecurity attackers and defenders. She addresses the argument that safeguards on frontier models are restricting their use and handing an advantage to China, weighs the merits of open source and open weight models, and reads the clues in Xi Jinping's recent AI speech about China's current strategy.The conversation then turns to Australia. Janet offers her assessment of Albanese's AI speech and examines the case for Australia to build data centres capable of training frontier models like Claude, GPT and Gemini. She covers the fraught AI copyright debate before finishing with thoughts on global AI governance and the prospects for US-China cooperation to reduce the most catastrophic risks.