Trump Calls It 'Superintelligence,' Rejects Global AI Controls as 22 Nations Push for Oversight

World leaders spent this week arguing over who gets to control artificial intelligence, and they can't even agree on what to call it.

At the UN General Assembly on Tuesday, Secretary-General António Guterres opened with a warning shot, according to CTV News and Euronews, both of which carried nearly identical accounts of his farewell address. "Let us resolve that life-and-death decisions must never be surrendered to machines. Killer robots must have no place in our future," Guterres said. He pointed to a new UN science report he says shows a "slight but real potential of catastrophic and irreversible harm" from AI.

Less than an hour later, President Trump took the same podium and rejected the premise. "The United States also totally rejects any attempt to construct a globalist scheme to control for the artificial intelligence being spoken of so much," Trump said, adding that what people call AI should now be called "superintelligence." He mocked climate scientists in the same breath, saying the same people who warned the world would end in 12 years from global warming are now warning about AI.

Trump's comparison doesn't match the facts. The 2018 UN report he was referencing warned that breaching 1.5°C of warming would bring more dangerous extreme weather, not planetary extinction, and estimated cutting pollution could prevent 100 million premature deaths over 82 years, according to CTV News. NOAA data cited by CTV News shows the planet has warmed 0.45°C over the past 12 years, not cooled.

CEOs Ask Governments to Rein In Their Own Industry

A day after Trump's speech, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei and Hugging Face CEO Clément Delangue addressed the UN Security Council directly, according to Linkdood Technologies. Amodei warned poorly managed AI could pose risks to humanity broadly. Altman raised the possibility of "losing control of the future to AI."

The room split fast. UK Energy Secretary Ed Miliband argued AI can't be left to the companies building it and said Britain will put AI governance at the center of its G20 presidency in 2027. White House science adviser Michael Kratsios pushed back, arguing that fear of losing control shouldn't be used to justify pausing development or building a new global governance structure. Chinese Ambassador Fu Cong warned against forcing the world into an "us-versus-them" tech bloc and pitched open-source AI as the better path.

Kratsios has a real point. Handing multilateral bodies veto power over the technology America currently leads is a competitiveness risk, not just a safety measure, and China isn't exactly a neutral safety regulator either. But dismissing the CEOs' own warnings as theater ignores that the men running the two most advanced AI labs on earth just told the Security Council they're worried about losing control of what they built.

22 Countries, No U.S., No China

Separately, 22 nations, including Finland and Norway, adopted a declaration Monday, September 21, stating AI "must remain under human direction, oversight and control," according to Politico Europe. Finnish President Alexander Stubb, one of the organizers, told Politico he wants to build a new international body, potentially modeled on the International Atomic Energy Agency, to oversee frontier AI development.

The declaration cites a cyberattack by rogue AI agents on the Hugging Face platform as justification for mandatory pre-deployment testing and independent safety audits. But it stops short of calling for any slowdown or pause in AI development. It has no enforcement mechanism, no penalties, and neither Washington nor Beijing signed on. Stubb told Politico he's hoping to get both superpowers on board eventually, calling the effort a matter of building "momentum," not a finished framework.

Behind the scenes, according to the Wall Street Journal (via Dow Jones Newswires) and separately reported by Hyper.ai, the leaders of Canada, France and Norway coordinated over a group chat after an island retreat, texting rapid affirmations like "We're in" to align on AI principles before bringing the effort to New York. That's diplomacy by text message, not treaty language, and it hasn't produced anything binding yet either.

No AI treaty exists. No enforcement body exists. What exists is a UN Security Council debate, a 22-country statement with no U.S. or Chinese signature, and public dueling speeches. Politico Europe notes the push comes just ahead of a scheduled meeting between Trump and Chinese President Xi Jinping in Washington today, Thursday, September 24, where AI governance is expected to be on the table alongside broader trade issues.

Whether that meeting produces any movement toward a U.S.-China understanding on AI rules, or whether Trump uses it to reinforce his rejection of "globalist" oversight, is the next thing to watch. Nothing announced so far commits either government to anything.