UN chief calls for worldwide AI safety partnership

At a press briefing on Wednesday, United Nations Secretary‑General António Guterres sounded a rare alarm about the pace of artificial‑intelligence progress. He said the global community “cannot afford a race to the bottom on AI safety,” and that the time for isolated, national‑only measures has long passed.

Guterres noted that the people building today’s most powerful models are already sounding the alarm – they’re racing ahead of our understanding of the risks. “We need guardrails to build trust,” he said, “and those guardrails must make AI safe, transparent, accountable, with human dignity at the centre.”

His remarks echoed a proposal put forward by Dario Amodei, CEO of Anthropic, which calls for a coordinated slowdown among the biggest tech firms and governments. In recent days, OpenAI’s Sam Altman, Elon Musk’s SpaceX AI team, and several other industry leaders have publicly advocated for a pause or at least a more deliberate pace.

Not everyone is on board, though. Meta’s Mark Zuckerberg quickly distanced his company, insisting that each AI developer must ensure safety on its own. And in the United States, former President Donald Trump dismissed existential AI fears as a “hoax,” labeling regulatory attempts a “sick conspiracy.” Those divergent voices highlight how political and commercial interests are tangled up with the technology.

Still, Guterres highlighted the upside of AI – faster economic growth, better education tools, stronger health‑care systems – while warning that “runaway artificial intelligence” could become an existential threat. He said AI will dominate the agenda at the high‑level U.N. General Assembly session next week, where leaders from 193 member states will discuss the issue.

Since 2025, the U.N. has backed a 40‑member scientific panel on AI, the first of its kind. The panel’s July report warned that unchecked, large‑scale AI use could harm mental health, become a destructive weapon, and disrupt social, economic, and environmental systems. The findings have been a catalyst for the Secretary‑General’s push for a shared, international framework.

Guterres acknowledged the geopolitical friction that makes cooperation tough – especially competition with China – but urged the countries with the most advanced AI capabilities to start a dialogue. “We need to exchange information, set common guardrails, and avoid a race that could end in a gigantic disaster,” he said.

When asked whether the upcoming U.N. gathering could double as a summit on AI limits, Guterres admitted the timing was tight but expressed optimism. “It makes full sense at the right moment to have a serious international summit on artificial intelligence,” he said, adding that governments, companies, scientists and civil society all need to be at the table.

In short, the message was clear: technology will keep advancing, but it must do so under a framework that protects people, respects dignity, and keeps the planet safe.