Trump Super Intelligence AI: UN Speech Renames AI, Rejects Global Oversight

Only a quarter of voters approve of how Donald Trump is handling AI. That’s what a Quinnipiac University poll found, released the same day he signed an executive order renaming the entire field. Twenty-five percent approve. Fifty-eight percent disapprove. His worst mark on any issue the poll tested. And the order sitting next to those numbers? A directive that the executive branch stop saying “artificial intelligence,” start saying “super intelligence”, and stop acknowledging the old term at all.

This is the full arc of Trump’s AI rebrand: a social-media poll, a UN podium, a State Department memo, and a signed executive order, eleven days end to end. What the order changes, what it can’t touch, and why the rest of the world isn’t following.

Eleven Days: Poll, Podium, Memo, Order

The push to rename AI to “super intelligence” began with a poll. Not a national survey, a Truth Social poll, put to the president’s own followers, asking them to pick a replacement for “artificial intelligence.” The options: “Superior Intelligence,” “Extreme Intelligence,” “Supreme Intelligence.” His pitch: the current words were “inaccurate, and very ineloquent.” More than 27,000 people voted within a day, according to Anadolu Agency’s reporting. A poll of his own followers. The scientific gold standard, it was not.

Two days later came the runoff: “SUPER INTELLIGENCE” versus “SUPERIOR INTELLIGENCE.” That same weekend, Trump announced plans for an “AI Force” modeled on the Space Force and a new AI czar, “Only High I.Q. individuals need apply.” (The AI Force announcement is covered separately on this site.)

Then the podium. In Trump’s UN speech on AI at the General Assembly, he declared that US documents would “use the much more accurate term ‘super’ as opposed to artificial”, “So it’s super intelligence.” His reasoning was plain-spoken: “The use of the word artificial makes intelligence fake, it makes it sound fake and it is not fake. It’s actually amazing.” He even questioned his own authority to impose it: “Let’s see if I have any power. Maybe I do, and maybe I don’t.”

The memo followed within hours. Michael Drager, a deputy assistant secretary of state, emailed staff in the Bureau of International Organization Affairs, the office handling US missions to the UN and multilateral agencies, under the subject line “Change in Nomenclature AI to SI.” The instruction: change every reference of “Artificial Intelligence” to “Super Intelligence” in documents and positions, and use “SI” in remarks and press materials going forward. The Associated Press obtained the email. Whether other bureaus followed suit is unclear.

And then, eleven days after the first poll, the executive order, the moment Trump’s AI rebrand stopped being a branding exercise and started being law.

What the Trump Superintelligence AI Order Changes, and What It Can’t

For anyone tracking Trump’s superintelligence story, the operative text is the order itself, not the speech. The superintelligence executive order is titled “Inaugurating the Era of Super Intelligence,” and its language is blunter than anything said at the podium:

• Every executive-branch department and agency must use “Super Intelligence” and “SI” instead of “artificial intelligence” and “AI”, “to the maximum extent permitted by law.”
• The old terms don’t just get replaced. The order says the executive branch “will not acknowledge the usage of ‘Artificial Intelligence’ and ‘AI’ in any applicable setting.” An unusual line for a presidential directive.
• But the legal definition doesn’t move an inch. The order defines “Super Intelligence” as exactly what “artificial intelligence” means under 15 U.S.C. 9401(3), and says so explicitly. A speech can’t rewrite the statute, and neither can this order. That takes Congress.
• The White House science adviser gets 60 days to draft proposed legislative language for a federal SI definition. That document decides whether this stays branding or becomes regulatory vocabulary.

The same afternoon, Trump hosted the heads of Google, Anthropic, Meta, OpenAI, Nvidia, and Elon Musk at the White House, where all six signed what was described as a “morally binding” accord on self-policing: robust internal controls, an independent external auditor, board-level oversight committees. No new federal safety law. Self-regulation, with the Department of Justice watching from the sidelines, the same posture as the UN speech.

Watching the summit footage, what stood out to me wasn’t in any document. It was Musk, on stage outside the West Wing, catching himself mid-sentence: “the positive benefits of AI, SI, pardon me. Super intelligence.” Even the allies needed a beat to adjust.

So: the Trump superintelligence AI rename is now mandatory inside the executive branch, voluntary everywhere else, and legally it’s the same technology wearing a new label.

The Other Half: Washington Says No to Global AI Rules

Buried under the headlines about the rename of AI to super intelligence was the other half of Trump’s UN speech on AI, the actual policy move. “The United States also totally rejects any attempt to construct a globalist scheme to control” AI, he told the General Assembly. Not a hedged objection. Totally rejects.

The counter-argument was sitting in the same building. UN Secretary-General António Guterres: “National action is essential, but global coordination is indispensable.” His warning: “The danger is not technology. The danger is technology without accountability.” He wants guardrails that keep AI “safe, transparent, accountable, with human dignity at the centre”, plus direct contact mechanisms between the US and China, the two countries at the frontier.

The UN isn’t starting from zero. Its General Assembly has already established two AI-governance mechanisms: an independent scientific panel to assess what the technology can and can’t do, and a global dialogue where governments try to align their approaches. That’s the machinery behind the global AI governance push Washington just walked away from.

France’s Emmanuel Macron staked out a middle position at the same session: build domestic AI capacity and the ability to audit models independently, “rather than depend on a handful of private actors”, while cooperating internationally on the sensitive areas. (The UN Security Council’s own AI debate shows how fractured the room is.)

One frame captures the tension: 71% of American voters say they’d back a candidate supporting stricter AI guardrails, per Quinnipiac. The administration’s position is encouragement first, enforcement only if the DOJ decides it’s needed. Those aren’t the same conversation. That gap, between Trump’s superintelligence agenda and what voters say they want, is the story beneath the rebrand, and the sharpest divide in the global AI oversight debate: one side building institutions, the other declining to join them.

The Word Was Already Taken

There’s a wrinkle nobody at the summit mentioned. “Superintelligence” already means something specific: a hypothetical system that outthinks humans across virtually every domain of cognition. No such system exists. Researchers use the word for the thing they’re most worried about building.

So the administration has taken the name of AI’s most feared hypothetical outcome and applied it to today’s systems. Trump’s superintelligence AI label doesn’t describe new capabilities, it redescribes existing ones. Critics call that a category error that muddies an already confusing field. Supporters say the new label finally captures the technology’s ambition instead of diminishing it with the word “artificial.”

Does the rename change what the technology actually does? Not a thing. The models are the same. The companies are the same. Only the government’s vocabulary moved, which is all Trump’s rebrand has moved so far.

One correction worth making plainly, because it’s already circulating as fact: the executive order did not declare that machines have surpassed human intelligence. It renamed today’s AI. The superintelligence of research papers, the kind that keeps safety researchers up at night, remains hypothetical.

How It Landed

Start with the numbers. Quinnipiac didn’t just find 25% approval for Trump’s handling of AI against 58% disapproval. It found 71% of voters more likely to back candidates favoring stricter guardrails, 79% ranking safety above innovation leadership, and 73% opposing an AI data center in their own community. Polling analyst Tim Malloy’s summary: nearly eight in ten voters “favor safety over the race to dominate development of AI.” The Trump superintelligence AI push landed in a country that mostly wants guardrails, not new vocabulary.

Scrolling through the press reactions afterward, what struck me was how the split ran along lines the speech didn’t anticipate. The Wall Street Journal reported that some Trump allies dismissed the resignation of Anthropic researcher Jacob Coxon, whose post warning that the people building AI “earnestly believe that it could kill us all by the end of the decade” was viewed more than 170 million times, as an effective-altruism plot to force regulation, with a White House memo targeting Anthropic and the movement circulating internally. Le Monde went the other direction entirely, calling the whole AI-pocalypse framing “a mirage distracting us from AI’s unsustainable development.” Nature, meanwhile, ran an explainer asking whether the science behind the extinction claims holds up, and mostly answered no. Same resignation, same week, three different stories.

That’s the longer-running argument underneath all of this, the slowdown-pledge debate among AI leaders, and the rename didn’t settle any of it. It just gave everyone a new word to argue with.

What to Watch Next

Five things decide whether this sticks:

• The science adviser’s proposal. The superintelligence executive order’s 60-day clock is running. That draft definition decides whether “super intelligence” becomes regulatory language or stays a branding exercise.
• Agency uptake. So far the only confirmed compliance is one State Department bureau’s email. Watch whether the Trump superintelligence AI terminology spreads across agencies, or stalls at the memo stage.
• The AI czar. Trump said a name was coming within days. The pick signals how seriously the administration takes the governance side of its own rebrand.
• Congress. Conforming amendments to the statute need legislation. Without it, federal law keeps saying “artificial intelligence” indefinitely, and every contract, grant, and regulation written under the old term stays exactly as it is.
• Other countries. If no ally adopts “SI,” the rename lives and dies inside the US executive branch.

The practical takeaway: if you deal with federal paperwork, search both terms for the foreseeable future. The technology didn’t change, a label did. Whether the Trump superintelligence AI name outlasts the administration that coined it is now the open question. Presidents’ vocabularies sometimes do.

FAQ