• Trump announced during his September 22, 2026 UN General Assembly address that US government documents will use 'super intelligence' instead of 'artificial intelligence.'
• The White House's own transcript is the only public documentation of the change; no executive order or federal directive accompanies it.
• Superintelligence already has a specific technical meaning in AI research — systems exceeding human cognitive performance across essentially all domains — a threshold no deployed AI system meets.
• Trump used the same remarks to reject international AI oversight, calling proposals for global coordination a 'globalist scheme of control.'
• The rename carries no operational impact on AI builders, evaluators, or regulators unless and until it is codified in binding policy.

President Donald Trump used his September 22, 2026 address to the UN General Assembly to announce that US government documents will replace the term 'artificial intelligence' with 'super intelligence,' arguing the word 'artificial' makes the technology sound less real than it is. The announcement carries no operational weight for anyone building or deploying AI systems: it is a labeling preference delivered from a podium, not a rule, standard, or funding decision — and it lands on top of a term, superintelligence, that AI researchers already use for something specific and, by every available benchmark, not yet built.

What Trump Actually Said

The line arrived inside a broader rejection of international AI oversight. In the White House's own release of the remarks, Trump stated: 'The United States totally rejects any attempt to construct a globalist scheme of control for the Artificial Intelligence being spoken of so much now — hereinafter officially called Super Intelligence.' The rename, in other words, functioned less as a scientific correction than as a rhetorical vehicle for rejecting the kind of coordinated, multilateral AI safety and risk-classification frameworks other governments and UN-adjacent bodies have proposed.

A Term That Already Has a Definition

Superintelligence is not a neutral rebrand — it is a term of art in AI research, denoting a system that exceeds human cognitive performance across essentially every domain, not a chatbot or a coding assistant. Safe Superintelligence Inc., the startup Ilya Sutskever co-founded in 2024 after leaving OpenAI, built its entire premise on that distinction: superintelligence as a future, unbuilt capability class that requires dedicated safety work before it arrives. Applying the label to every current AI system, from spam filters to production LLM deployments, erases the distinction the term was coined to protect.

No Executive Order, No Enforcement Mechanism

As of publication, the only public documentation of the change is the White House's own transcript of the speech — no accompanying executive order, OMB memorandum, or Federal Register notice establishing 'super intelligence' as required terminology across federal agencies has surfaced. That distinction matters for anyone tracking US AI policy: a floor speech is not a procurement standard, a NIST framework revision, or a legal definition, and nothing in existing federal AI documents changes on the strength of it alone.

Why the Terminology Still Matters

Terminology fights are not purely cosmetic in AI policy: risk frameworks, liability rules, and export controls hinge on how a document defines 'AI' versus more speculative categories like AGI or superintelligence. If US agencies eventually did adopt 'super intelligence' as a blanket substitute, it would blur the line policymakers use to scope narrow rules for today's deployed models against the separate, open question of frontier-risk regulation for systems that don't exist yet — the opposite of the precision safety researchers have spent a decade building into that vocabulary.

Call it what it is: a branding move, not a policy. Nothing about model capability, safety evaluation, or deployment risk changed on September 22 — only a word Trump would prefer federal documents use. Treat 'super intelligence' in any future US filing as administration house style, not as a claim that a system has crossed the technical threshold researchers still reserve the term for.