Why Singapore and 19 Other Countries Called for Global Oversight of Frontier AI

Rapid advances in frontier AI prompted leaders to call for safeguards that no single company or country would have to manage alone. Launched during UN General Assembly week in New York on September 21, 2026, A Call for Control of Frontier AI Models seeks to keep people in charge while allowing countries to benefit from AI.13133

Why Singapore backed the call

Prime Minister Lawrence Wong framed Singapore’s position as a balance: AI holds tremendous promise, but it must remain under human control and oversight. Singapore was the only South-east Asian country in the initial joint call, which paired that principle with international cooperation on safeguards.617

The declaration likewise recognises AI’s potential to improve lives, advance science and strengthen economies. Its concern is that the most advanced models are developing quickly enough to create safety and security risks that national measures alone may not adequately address. It calls for AI to remain under human supervision and to be developed and used in accordance with international law.1331

What the declaration asks for

The call sets out work for companies, governments and UN member states—not an already-operating global regulator:

• Companies should establish transparent safety protocols, including pre-deployment testing and independent evaluations. Qualified evaluators should have sufficient access to assess risks.30
• Governments and regional organisations should coordinate common standards and improve transparency, including the sharing of serious safety-incident reports. The reported proposals also seek wider access to scientific expertise and trusted evaluations.2530
• UN member states should build on existing mechanisms and explore an international institution able to set standards, enable verification and bring states together when AI capability thresholds are crossed. That institution remains a proposal, not a body created by the declaration.2936

This is a call for testing, verification and shared oversight rather than an explicit demand in the cited proposals to halt frontier AI development.293036

Who joined—and who did not

Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Støre launched the initiative. Finland’s presidency lists the initial supporters as Finland and Norway, plus Denmark, Iceland, Estonia, Latvia, the Netherlands, Ireland, Germany, Spain, Moldova, Türkiye, the United Arab Emirates, Bahrain, Kazakhstan, Singapore, Kenya, South Africa, Canada and Australia, alongside the European Commission. Its announcement also reports a larger number of supporters joining subsequently, so the launch roster should not be mistaken for a fixed final list.33

The United States and China were not part of the initial declaration. Stubb said he hoped to bring more participants on board, but their absence alone does not establish either government’s reason for staying out.9 UN Secretary-General António Guterres welcomed the Finland–Norway initiative and specifically supported exploring standards, verification and a way to convene states when capability thresholds are crossed.36

The declaration arrived amid broader attention to frontier-model safety. Microsoft had published guiding principles for its cutting-edge models, while Anthropic announced an independent-evaluation partnership with Accenture days before the joint call. Those developments provide context, not proof that either prompted the governments’ decision.3940