The United Nations wants countries to work together on rules for artificial intelligence. Secretary-General António Guterres said global coordination is needed as AI technology keeps advancing.
His comments came about a week after a former researcher at the AI company Anthropic warned that AI could pose an existential threat to humanity. That warning was made public on September 8.
Guterres said national action alone is not enough. "National action is essential, but global coordination is indispensable," he said on Wednesday.
The UN has spent several years building a system for countries to discuss AI safety, human rights and access to the technology.
Two New Global Bodies
The UN's current work on AI started with the Global Digital Compact. Member states adopted it in 2024 as part of the UN's Pact for the Future.
The Compact called for two new tools. One is an independent scientific panel to study AI risks and opportunities. The other is an annual dialogue for governments to coordinate on AI rules.
The UN General Assembly formally created both groups in August 2025.
The Independent International Scientific Panel on AI gives governments an outside look at what is known, and not known, about the technology.
The Global Dialogue on AI Governance brings governments and other groups together. They share experience and try to make their different AI rules work with each other.
The Scientific Panel released its first report on July 1. It found that AI is improving fast in areas like reasoning, coding and science.
The report also listed risks. These include misinformation, discrimination, privacy violations, cyberattacks and systems described as "superintelligence" that could become hard to control.
Calls for International Guardrails
Guterres said the world needs a shared understanding of how to develop AI safely. He said stronger safeguards may be needed as systems grow more powerful.
He said voluntary efforts to slow AI development will not work if countries act alone, if the efforts cannot be checked, or if they are applied unevenly.
Guterres compared the AI race between countries to the Cold War arms race between the United States and the Soviet Union.
He called on countries leading in AI, including the United States and China, to set up contact and share information to avoid what he called a race to the bottom.
He also called for an AI Child Safety Pledge and a Global Fund for AI. The fund would help developing countries build AI infrastructure.
Guterres said a global summit on AI regulation cannot happen right away but should take place soon, with input from public, private, scientific and civil society groups.
AI will be a major topic when Guterres meets with world leaders during the UN General Assembly's high-level week this month.
The UN's SDG Media Zone will host an AI event on September 21. A separate discussion on AI and human development is set for September 23.
