The United Kingdom must cultivate its own artificial intelligence capabilities to safeguard against dependence on foreign systems, according to George Williamson, the Chief Executive of the Alan Turing Institute. Williamson emphasized that "national resilience" is a primary focus for the institute, particularly given the leadership of the United States and China in the AI sector. He cautioned against relying solely on AI systems that the UK cannot inspect, control, or that could be subject to external restrictions or shutdowns.
Williamson highlighted that such dependence carries risks, especially as AI integrates into critical services like defense, national security, critical infrastructure, and healthcare. He pointed to instances where political and commercial pressures have already impacted access to AI. For example, Washington temporarily blocked the release of Anthropic's latest AI models outside the US in June. Furthermore, leading AI laboratories have reportedly been asked to withhold their models from testing by the UK's AI Security Institute until their US counterparts have examined them.
The Alan Turing Institute, which serves as the UK's national institute for data science and artificial intelligence, is undergoing a strategic shift under its new leadership. This redirection follows demands from the UK government, the institute's primary funding source, for a greater focus on strategic priorities. Williamson indicated that the emphasis on national resilience allows the institute to continue work on areas like climate, which he considers a national security issue, alongside defense and security.
While acknowledging that the UK is unlikely to match the funding of the largest AI laboratories globally, Williamson proposed that the country could focus on developing its own intellectual property and smaller, specialized AI models accessible to startups. He questioned whether Chinese-made AI systems would align with British values, citing their responses regarding recent Chinese history as an example.
The institute plans to contribute to national resilience by securing critical systems against disruption, countering AI-enabled threats, and helping to shape the UK's strategic AI capability. This includes developing AI applications for public good, such as weather forecasting and farming, and assisting in identifying disinformation. Williamson stressed that state-backed disinformation campaigns pose a threat to democracy.
The Alan Turing Institute has previously outlined a practical agenda to address frontier AI risks, identifying ways AI agents can behave contrary to organizational intentions and setting out principles to govern agent behavior. The institute's work in assurance, verification, and cybersecurity is intended to support a broader national effort, collaborating with entities such as the AI Security Institute and the National Cyber Security Centre. Williamson emphasized that discussions about advanced AI should not be confined to technology companies or extreme scenarios, but should involve a wider public discourse on the moral, ethical, and policy implications.