GREAT BAY--The Netherlands has joined a group of international leaders calling for stronger controls over the world's most advanced artificial intelligence systems, warning that the speed at which AI is developing could exceed governments' ability to manage emerging safety and security risks.
Dutch Prime Minister Rob Jetten added his name to the international declaration, titled “A Call for Control of Frontier AI Models,” during United Nations high-level week in New York. The initiative was launched by Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Støre. According to the Office of the President of Finland, 22 leaders from 20 countries had joined the initiative at its launch, along with European Commission President Ursula von der Leyen.
The declaration focuses on what are commonly called frontier AI models, meaning the most advanced AI systems being developed at the leading edge of the technology. While acknowledging AI's potential to improve lives, accelerate scientific research and strengthen economies, the signatories warn that increasingly capable systems can also create new safety and security risks when adequate safeguards are not in place.
The statement points specifically to recent examples of advanced AI systems circumventing testing safeguards, exploiting computer vulnerabilities and gaining unauthorized access to real-world systems. It says governments, technology companies and society need to strengthen oversight while ensuring that countries are not excluded from the economic and social benefits of AI.
Testing before advanced AI is released
Among the proposals is a call for AI companies to establish transparent safety protocols and conduct mandatory testing before advanced models are deployed. The declaration also calls for independent evaluations, with qualified outside evaluators given enough access to determine whether systems present significant risks.
Governments and regional organizations are being urged to coordinate common AI safety standards and create mechanisms for reporting serious safety incidents. The signatories also want scientific knowledge and independent evaluation capabilities made available to countries across different regions, rather than concentrated among a small number of technologically advanced states.
The initiative goes further by asking UN member states to explore the creation of an international institution capable of establishing standards, supporting verification and bringing governments together when AI systems cross specified capability thresholds. The declaration does not itself create such an institution or impose new binding rules, it is a political call for governments and companies to develop them.
Stubb and Støre have framed the issue as requiring international cooperation rather than individual countries attempting to manage rapidly advancing AI independently. Their offices said the appeal will continue to be promoted during UN High-Level Week and remains open for additional countries to endorse.
Major AI powers not among initial signatories
The official list includes leaders or senior officials from countries including the Netherlands, Germany, Canada, Australia, Spain, Singapore, Kenya, South Africa, Denmark, Norway and Finland.
The United States, China, United Kingdom and France do not appear on the initial list of signatories published by the Dutch and Finnish governments. Their absence from the list does not, by itself, indicate opposition to every proposal contained in the declaration.
The debate comes as artificial intelligence takes a prominent place at the 81st United Nations General Assembly. UN Secretary-General António Guterres told world leaders Tuesday that AI is one of four major tests facing the international community and called for binding international cooperation on AI governance.
Netherlands signs as EU AI rules take effect
The Dutch decision also comes shortly after another major development in AI regulation. The European Union's AI Act became broadly applicable on August 2, 2026, with the European Commission beginning enforcement of several provisions, including transparency requirements for certain AI systems and AI-generated material.
The EU framework and the new international declaration are not the same initiative. The AI Act establishes European legal requirements, while the declaration signed by Jetten seeks broader international cooperation on the most capable AI models and possible future global oversight mechanisms.
For smaller countries and territories that increasingly rely on AI in government, business, education, healthcare, communications and other sectors, the international discussion could eventually influence the standards expected from the companies developing and supplying these systems.
For now, however, the declaration represents a call for stronger international safeguards rather than a new global AI law. What follows will depend on whether additional governments join the initiative and whether its proposals for mandatory testing, independent evaluation and international oversight are converted into formal agreements or legislation.
