Loading page…
The U.N. secretary-general wants leading AI powers to share information and establish a common safety floor. But he says the upcoming General Assembly session is too soon for a properly prepared international summit.
Listen to this story
The U.N. secretary-general is deferring a proposed international AI summit because the upcoming General Assembly session cannot be prepared properly. Instead, António Guterres wants high-capacity countries to establish contacts and align on a baseline of safety, transparency, accountability, and human dignity. A 40-member scientific body created in 2025 already warned in July about mental-health, destructive-use, systemic, and control risks.
Guterres says any summit should include governments, AI companies, scientists, and civil society, but offers no date.
The July preliminary report from a 40-member body lists mental-health, destructive-use, systemic, and control risks.
Anthropic CEO Dario Amodei proposed a coordinated slowdown; Meta CEO Mark Zuckerberg has distanced Meta from that approach.
António Guterres is urging countries and AI companies to adopt common safety guardrails, warning that global competition must not become a race to the bottom. But the U.N. secretary-general said there is not enough time to prepare a serious international summit for the upcoming General Assembly session.
Speaking to U.N. reporters, Guterres said governments have a first responsibility to protect civilians from AI threats. National action is essential, he said, but global coordination is indispensable.
His proposed baseline is broad rather than technical: guardrails that make AI safe, transparent and accountable, with human dignity at the center. He urged countries with the highest AI capacity to establish contact, exchange information and agree on protections.
The world cannot afford a race to the bottom on AI safety.
António Guterres, U.N. secretary-general
AI will be a major subject in Guterres’ discussions with leaders at next week’s annual General Assembly gathering. He said there was not enough time to properly prepare a summit, though a serious international meeting would make sense “at the right moment.”
Guterres said a summit should put governments, technology companies, scientists and civil society at the same table. The U.N. will continue exploring practical approaches in the immediate days ahead, he said.
At Guterres’ behest, the General Assembly established a 40-member global scientific body on AI in 2025. Its preliminary July report warned of risks from rapid, unchecked use at scale.
Guterres’ appeal arrives amid disagreement over whether safety should slow AI development. Anthropic CEO Dario Amodei has proposed a coordinated slowdown among leading technology companies and governments. Meta CEO Mark Zuckerberg has said Meta must make its AI safe while distancing the company from that approach.
President Donald Trump has called claims that AI could destroy humanity a hoax and opposed efforts to limit the technology. Guterres is instead asking leading AI countries to begin contact and agree on guardrails before competition lowers safety standards.
Loading discussion...
Join the conversation
Explain what could make cross-border cooperation work.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.