Australia’s intelligence and security officials have raised concerns about the vulnerability of the country’s outdated technology to artificial intelligence (AI)-driven cyberattacks, as the government moves to establish regulatory frameworks amid rapid AI advancements.
Abigail Bradshaw, director general of the Australian Signals Directorate, spoke at the Sydney Dialogue summit in Canberra on Monday, highlighting the challenge of tracking AI agents operating online. “I think that prospect of getting some sort of definition or calculation of how these agents are out there is really remote,” she said, noting the presence of “a lot” of such agents. Bradshaw warned that significant investments would be needed to upgrade Australia’s aging digital infrastructure, cautioning that taking vulnerable systems offline could disrupt services Australians rely on.
The discussion coincided with calls from AI industry leaders for greater oversight and regulation. Dario Amodei, CEO of Anthropic—the developer of the AI model Claude—has advocated for independent evaluators to scrutinize AI systems and for U.S. government-sanctioned coordination among leading AI companies to avoid anticompetitive practices. Jeff Bleich, Anthropic's chief lawyer, welcomed Australia’s engagement, suggesting that global AI regulations should not be determined solely by corporate executives or governments of only a few nations. “It won’t be easy to get these things done but it actually makes it better to come to Australia,” he said, emphasizing the need for rules within months to keep pace with AI’s exponential growth.
Despite these warnings, the Australian government’s approach has been cautious. Labor assistant technology minister Andrew Charlton acknowledged the seriousness of the challenges but resisted placing too much weight on the demands of individual company leaders. He underscored that forthcoming guidelines would seek to address both the potential risks and benefits of AI. Legislation setting mandatory safety and transparency standards for advanced AI systems is expected to be introduced early next year. However, a bid by the Greens to require immediate safety testing and transparency obligations was blocked by Labor senators.
Discussions at the summit also highlighted intellectual property concerns related to AI training data. Industry Minister Tim Ayres stressed the government’s commitment to protecting copyright, even as companies like OpenAI expressed interest in expanding AI model training within Australia. OpenAI’s vice-president for global policy, Ann O’Leary, noted ongoing talks with Australian officials aimed at balancing copyright protections with the company’s desire to diversify its data sourcing. “We very much recognise that having a diversification of where we train is going to be important as you move forward,” she said.
The question of maintaining sovereign AI capabilities became a focal point for some participants. Andrew Hastie, the Coalition’s industry spokesperson, warned that without domestic frontier AI technology, Australia risked losing strategic autonomy. He advocated for stronger ties with the United States, including through the Aukus security partnership, to encourage major AI companies to establish operations on Australian soil. “We need to work on a way to get a hyperscaler down under,” Hastie said, underscoring the national security implications of advanced AI development.
Meanwhile, officials noted that Australia is already using Anthropic’s Claude model within government sectors, including defense, but some experts expressed skepticism about the country’s ability to independently develop cutting-edge AI technologies in the near term.
As the AI landscape evolves rapidly, Australian authorities face complex decisions balancing innovation, security, intellectual property rights, and global cooperation in shaping the country’s AI governance framework.