Photo: Australia’s Assistant Minister for Science, Technology and the Digital Economy, Dr Andrew Charlton MP and UK Minister for AI and Online Safety, Kanishka Narayan MP (Photo: Australian Government, Department of Industry, Science and Resources)
The UK and Australia are deepening cooperation on artificial intelligence (AI) security, with a new agreement designed to help both countries respond to fast-moving risks from powerful AI systems.
The Memorandum of Understanding (MoU), announced as ministers meet in Canberra, will bring together the UK AI Security Institute and the Australian AI Safety Institute. The partnership will focus on frontier AI capabilities, including how advanced systems could be used in cyber-attacks, as well as how AI could strengthen cyber defence.
The agreement comes at a time when governments are increasingly concerned about the speed at which AI systems are developing. While AI is expected to support economic growth, improve public services and strengthen national capability, ministers are also warning that the same technology could be used to scale cyber threats, automate malicious activity or accelerate the discovery of new vulnerabilities.
Under the new partnership, the two institutes will share information on AI capabilities, collaborate on research into emerging risks and work together on international best practice for testing and evaluating AI systems. The agreement will also allow for staff exchanges between the two institutes, building day-to-day cooperation between teams working on AI safety and security.
Ministers seek to stay ahead of fast-moving risks
UK AI Minister, Kanishka Narayan signed the agreement alongside Australia’s Assistant Minister for Science, Technology and the Digital Economy, Dr Andrew Charlton, in Canberra.
Narayan said the UK and Australia had “always worked closely” to protect their citizens, adding that the partnership “matters more than ever in the age of AI”. He warned that the technology is moving quickly, particularly in areas such as cyber security, and said: “No country can tackle that alone.”
His comments reflect a growing international consensus that AI safety and security cannot be managed by individual countries acting alone. As frontier models become more capable, governments are seeking to develop shared approaches to evaluation, risk monitoring and technical testing.
The UK has placed significant emphasis on AI safety since hosting the first global AI Safety Summit at Bletchley Park in 2023. The AI Security Institute has since become a central part of the UK’s approach to understanding the risks posed by advanced AI systems and informing policy decisions.
The new agreement with Australia adds another layer to that international work, strengthening bilateral cooperation with a close ally while contributing to wider efforts to build common standards for frontier AI assessment.
Frontier AI and the cyber security challenge
The Government said the agreement is being driven partly by new research from the UK AI Security Institute, which indicates that advanced AI systems are improving rapidly in their ability to carry out complex cyber-attacks. That creates risks for businesses, critical infrastructure and the wider public, but it also opens up opportunities for defenders to use AI to identify vulnerabilities, detect attacks and strengthen resilience.
This dual-use challenge is one of the central questions facing policymakers. The same capabilities that could help a security team analyse code, monitor threats or respond more quickly to incidents could also be misused by malicious actors to automate parts of the cyber-attack chain.
The UK and Australia’s approach appears to be based on the view that technical cooperation is essential. Rather than treating AI security as a purely domestic policy issue, the two countries will work together on testing, evaluation and research, helping to ensure that both governments have a stronger evidence base as the technology develops.
“This is exactly the kind of international cooperation we need to see more of. The UK has already established itself as a leader in AI security, and by working closely with trusted partners like Australia we can build the coalitions needed to stay ahead of fast-moving risks while ensuring AI is developed safely and responsibly.”
Tim Flagg, CEO, UKAI
Building international best practice
The partnership will also support work on international best practice for AI evaluation. That is likely to become increasingly important as governments, companies and research bodies seek clearer ways to assess whether powerful AI systems behave as intended.
The Australian Department of Industry, Science and Resources said MOU will provide a framework for technical cooperation and promote the safe and responsible use of artificial intelligence. It said the agreement would help both countries share expertise on emerging AI capabilities and risks, conduct joint research, and support international work on AI measurement, evaluation and science. The department added that the partnership would help both countries “keep pace with change” while ensuring communities share in the benefits of AI.
The UK AI Security Institute already works with research bodies across major economies through the International Network for Advanced AI Measurement, Evaluation and Science, as well as through bilateral partnerships. The Australia agreement adds to that growing network of cooperation.
AI security enters a new phase
The UK-Australia pact signals a shift in how allied governments are approaching AI risk. The pact demonstrates how governments are looking to move away from high-level principles or broad ethical statements. Increasingly, governments are moving towards practical technical cooperation: sharing research, testing models, evaluating capabilities and building institutional links between AI safety and security bodies.
That reflects the pace of change in the technology itself. As AI systems become more capable, the risks they pose are becoming more operational, particularly in cyber security. The challenge for governments is to keep pace without slowing the benefits AI can bring to innovation, economic growth and public services.
The agreement between the UK and Australia is therefore both a security measure and a strategic partnership. It aims to ensure that two close allies can better understand emerging AI risks, strengthen their defences and help shape the international standards that will guide frontier AI development in the years ahead.