Miliband Tells UN Security Council to Act on AI Founders’ Warnings

UK Foreign Secretary Ed Miliband told the United Nations Security Council on September 23, 2026, that governments must act together on warnings coming from the founders of artificial intelligence, laying out an agenda of safety testing, transparency from AI companies and national resilience. The Foreign, Commonwealth & Development Office published the transcript, exactly as delivered at the United Nations in New York, the same day.

UN Meetings Coverage recorded the session as the Security Council’s 10,228th meeting, an afternoon session held under the title “Artificial Intelligence and International Security.” Miliband opened by thanking the conveners of the session and the expert speakers who, he said, had made an illuminating contribution.

Founders’ Warnings and the Duty of Government

Miliband told the Council that AI undoubtedly has the power to transform the world for the better, something he said is visible in daily life in every country represented in the room. But, he said, the founders of the technology are now coming to governments with a stark warning: they fear its ability to do harm, and they are saying that it cannot be left to them alone to prevent disaster.

He said governments cannot ignore those warnings, because the founders are right. Governments, he said, cannot abdicate their responsibility to their citizens, and they cannot hand to private companies the first duty of government, which is to protect the public, whatever the issue.

Miliband said the Security Council was founded to address the greatest collective challenges of the time, and that the obligation now falls to governments to work together, and together with industry, in the interests of the people they represent. He said the UK government is committed to doing so.

Safety, Transparency and Resilience

Miliband set out three areas of work. The first, he said, must be safety: working harder to ensure frontier AI models are rigorously tested, and building a system of assurance around model development so that governments can establish trust that models are safe.

Second, he said, is transparency. The companies building the models cannot simply be left to their own devices, and governments must have the visibility to understand and assess what AI companies are doing. Miliband noted that the leading AI companies have committed to provide that visibility, and he described it as an offer that governments and the companies should follow through on.

Third, he said, is resilience: hardening defences to reduce the impact of AI-enabled incidents. Even if every model is tested, he said, governments must be prepared for AI to be misused. That includes boosting cyber defences, protecting critical infrastructure and reinforcing financial systems, as well as supporting populations to manage and adjust to the disruption AI will bring to how people live and work, while harnessing its benefits.

The AI Security Institute and the G20 Presidency

Miliband said the UK will play its full part in the process. He said the UK established the first AI Security Institute, which he described as sharing its expertise, testing capabilities and research while working closely with the United States and the leading labs on AI safety.

The institute describes itself as a research organisation within the Department for Science, Innovation and Technology that tests leading AI systems before they are released publicly and open-sources evaluation tools such as Inspect. It says its research spans cyber misuse, safeguards, alignment, control, autonomy, human influence and societal resilience, supported by £66 million in funding per financial year, a technical staff of more than 100 and pre-deployment access to leading AI models.

Also on September 23, the Cabinet Office announced that Dr Jade Leung will become Vice-Chair of the AI Security Institute and Security Advisor to the government’s AI Taskforce. She will step back from her full-time roles as the Prime Minister’s Adviser on AI and the institute’s Chief Technology Officer at the end of September 2026 because of personal circumstances, and a new Adviser on AI to the Prime Minister will be appointed in due course. Leung will also take up a part-time position as a Distinguished Visiting Fellow at Stanford University’s Hoover Institution.

Miliband also said that, as Prime Minister Andy Burnham had set out on September 22, 2026, the UK will put AI at the heart of its G20 presidency in 2027, building on its record of bringing together government, industry, researchers and civil society on the issue. The Prime Minister’s Office announced on September 22 that the UK-hosted G20 Leaders’ Summit will be held in Manchester in November 2027, and that the UK will take over the presidency from the United States following the G20 summit in Miami in December 2026. The same release announced a UK-US AI and Autonomy partnership pairing the Ministry of Defence’s Rapid AI Delivery Taskforce with the US Department of War’s Chief Digital and Artificial Intelligence Office, work the release said would accelerate the delivery of trusted, interoperable capabilities and help protect critical national infrastructure, including sea lanes and airspace. The partnership builds on next-generation technology already being developed through AUKUS, the release said.

Global Standards and the Closing Appeal

Miliband pointed to work towards agreeing a single set of global principles and standards, which he said would ensure the transparency and access necessary to increase preparedness and ensure that AI development is safe. Crediting a point made by a previous speaker, he said AI’s potential must be seized for the Global South and the Global North alike, across issues ranging from the future of jobs and protecting children to cyber security, biosecurity, national security and financial stability.

On September 21, 2026, UN Secretary-General António Guterres issued a statement welcoming the Call for Control of Frontier Artificial Intelligence Models, launched by the President of Finland and the Prime Minister of Norway. The initiative calls on UN member states to build on existing international mechanisms and explore creating an international institution able to set standards, enable verification and convene states when capability thresholds are crossed. Guterres also welcomed a thematic brief on AI safety and the risks from loss of human control from the Independent International Scientific Panel on AI, and he encouraged external experts, including from frontier AI labs and AI safety institutes, to engage with the panel.

Miliband closed by rejecting the idea of a choice between progress and safety. Innovation and the benefits of AI, he said, will be captured only if people have confidence that the right safeguards are in place, and governments must take whatever measures are necessary so that AI is secure, safe and trustworthy while still advancing its potential to drive growth, cure disease and transform people’s life chances. “We can’t say we haven’t been warned,” he said. “We have been warned.” Governments, he said, must act on the founders’ warnings, and must act on them together.