Question to the Cabinet Office:

To ask the Minister for the Cabinet Office, what assessment her Department has made of whether domestic resilience measures can protect the UK from risks arising from loss of control of advanced AI systems in the absence of international agreement limiting the development of such systems.

Domestic resilience measures are an important part of the Government's approach, but a range of responses will be needed to address the risks from loss of control of advanced AI. The most capable AI systems are developed by a small number of companies, largely outside the UK, and the risks they pose cross borders.

That is why the Government works on these risks at source as well as at home. The AI Security Institute (AISI) works directly with frontier AI developers to test their most advanced systems, including for capabilities relevant to loss of control, and to strengthen the safeguards that protect against them. The UK also works with allies and international partners to build a shared scientific understanding of these risks and of the measures needed to manage them.

Domestically, the Government prepares for and monitors a wide range of national risks, including scenarios where AI could act as a catalyst or introduce novel risks, through the National Risk Register and the Chronic Risks Analysis. We keep this approach under review as the technology and the evidence develop.