On 29 September 2026, President Donald Trump signed an executive order directing federal agencies to use “super intelligence” — abbreviated “SI” — in place of “artificial intelligence” and “AI” in official government communications. That instruction is confirmed fact, from the order’s own published text. What the term is claimed to mean is separate. The order presents “super intelligence” as a better description of today’s most capable AI systems. In AI research, “superintelligence” has a much narrower meaning: an intellect greatly exceeding the best human minds in practically every field — debated, contested and not established as achieved. The two meanings are not the same, and confusing them is the central misunderstanding this explainer clears up.
This article keeps three things apart: what the order actually says, what the White House claims the term captures, and what the technical terms mean in research. “Super intelligence” as a federal label is not a technical standard, changes no law, and sets no definition — yet. A definition proposal is due in late November 2026; until it appears, the order defines the new term by pointing back at the old one.
What the executive order says
Executive Order 14434 is titled “Inaugurating the Era of Super Intelligence”. It is short, and its operative content can be stated in full.
Section 1 states that American scientists first named artificial intelligence and that American companies are again pushing the frontier. It claims frontier systems “far exceed what was envisioned when the term ‘Artificial Intelligence’ first came into use”, doing “much more than imitate or automate discrete aspects of human intelligence”. From that claim it draws the policy: “to the maximum extent permitted by law, the executive branch shall use the terms ‘Super Intelligence’ and ‘SI’ in place of ‘Artificial Intelligence’ and ‘AI’ and will not acknowledge the usage of ‘Artificial Intelligence’ and ‘AI’ in any applicable setting.”
Section 2 directs executive departments and agencies, to the maximum extent permitted by law, to use “Super Intelligence” and “SI” in “official correspondence, public communications, websites, reports, policy documents, and other non-statutory documents”. The “non-statutory” qualifier matters: the instruction covers the executive branch’s own documents, not laws passed by Congress, and nothing in it “requires the alteration of previously issued regulations, Presidential actions, contracts, grants, or other historical documents”.
Section 3, on definition, is the most important part. For purposes of the order, “Super Intelligence” means “the technologies and systems encompassed by the term ‘artificial intelligence’ as defined in section 9401(3) of title 15, United States Code”, unless superseded by later presidential action or an Act of Congress. In plain terms, “super intelligence” currently covers exactly whatever the statutory AI definition already covers — no new threshold, test or category.
Section 4 provides the order “is not intended to, and does not, create any right or benefit, substantive or procedural, enforceable at law or in equity by any party against the United States”. It directs the executive branch’s own language, creating no enforceable rights and imposing no obligations on companies, states or Congress.
What the White House claims the term means
The Section 1 claim does the order’s persuasive work. Its three steps: frontier systems do more than “artificial intelligence” originally suggested, amplifying ingenuity rather than merely imitating it; the old name therefore understates them; and “super intelligence” better captures their promise and advancing capabilities.
Each step is a judgement about language, not a measurement: the order offers no benchmark by which a system could be shown to have crossed into “super intelligence”. Readers should recognise the claim as the Administration’s characterisation, not a verified finding.
That matters because “super” implies more than human. A reader seeing a federal document describe a chatbot as “super intelligence” could reasonably take it to mean the system is superintelligent in the technical sense. On the order’s own definition, it means only that the system falls within the statutory AI definition and is described with the new label.
On 8 October 2026, President Trump wrote on Truth Social, as reported by the Philippine News Agency: “The White House considers anyone that uses the term, ‘Artificial Intelligence,’ as opposed to the highly accepted new and more accurate term, ‘Super Intelligence,’ THE ENEMY!” It is political rhetoric, not a legal instrument: using “artificial intelligence” carries no legal consequence. The safety accord signed the same day is a different document — see our explainer on the White House AI safety pact and Super Intelligence accord; neither should be cited as evidence for the other.
A short history of the two terms
“Artificial intelligence” is conventionally credited to the computer scientist John McCarthy, who proposed the term for a 1956 workshop at Dartmouth College. For seventy years it has named a broad field and technology class, from early chess programs to modern models — not a claim about how intelligent any particular system is.
“Superintelligence” — one word in research, though the White House writes “Super Intelligence” as two — comes from philosophical writing about what might follow human-level AI. Its best-known formulation is by the Oxford philosopher Nick Bostrom, who in his 2014 book Superintelligence defined a superintelligence as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills”. The definition demands superiority across practically every field, not excellence at one task: superintelligence is a hypothetical future condition, not a demonstrated system.
The order collapses that distance, applying a word from the futures literature, as federal house style, to the entire existing field. That is its prerogative for executive-branch communications, but it risks misunderstanding, addressed next.
Narrow AI, AGI and superintelligence: the ladder
AI discussions use three rungs that are often confused. Keeping them separate is essential to reading any claim about “super intelligence”.
| Term | What it means | Status |
|---|---|---|
| Narrow AI | Systems that excel at specific tasks — translation, image recognition, game-playing, code completion — without general ability across tasks | What exists today: every deployed system falls here |
| AGI (artificial general intelligence) | A hypothetical system with broadly human-level ability across most tasks, transferring learning between domains as humans do | Hypothetical and contested: no agreed threshold, no demonstrated system |
| ASI / superintelligence | A hypothetical system greatly exceeding the best human intellect in virtually all domains, including scientific creativity and social skills (Bostrom) | Hypothetical: a possible future condition, not a demonstrated capability |
The order skips the middle rung, moving terminology from the broad field to the top rung’s label without any test for passing through general intelligence — and its own definition confirms the skip: “super intelligence” currently means whatever artificial intelligence already meant in statute.
Current systems are still impressive — frontier models meet or exceed human baselines on some demanding benchmarks — but benchmark excellence is not Bostrom’s test. Superintelligence claims breadth and superiority across practically every field, and no current system has been demonstrated to meet it. A programme to develop superintelligence is an ambition, not a report of achievement.
What the order does not change
It does not change the law. Statutes saying “artificial intelligence” still say so; the order applies to non-statutory executive-branch documents and borrows the statutory AI definition. Only Congress can change a statutory definition — the IAPP calls the immediate change linguistic, “not a wholesale rewrite of U.S. laws governing AI”.
It does not set a definition. No federal definition exists yet. The order requires a proposal within 60 days, but a proposal is not a definition and may not be adopted.
It does not declare that superintelligence has arrived. Nothing in the order states that any system meets Bostrom’s definition or any other technical test. Its language about frontier capabilities is general, cites no benchmark or study, and is not a technical assessment.
It does not bind private companies, the press, researchers or the public. The instruction runs to executive departments and agencies; companies may still call their products AI. The “will not acknowledge” language describes the executive branch’s own communications, not a prohibition on anyone else.
It does not rewrite history. Previously issued regulations, presidential actions, contracts, grants and historical documents need not be altered. Federal records will show both vocabularies: older documents saying “AI”, newer executive-branch documents “SI”, statutes and contracts “AI”.
It does not create enforcement or rights. Section 4 provides it creates no enforceable right or benefit; no penalty exists for an agency that keeps writing “AI”.
The 60-day proposal: what to watch
The order’s potentially most substantive part is Section 3(b). Within 60 days of 29 September 2026 — late November 2026 — the Assistant to the President for Science and Technology must submit proposed legislative language for a federal definition of “Super Intelligence” and “SI”, assessing whether it should modify, expand or supersede the statutory AI definition, with conforming amendments and recommendations.
Three outcomes are possible: a narrow definition (a subset of AI systems meeting a capability threshold), a broad one (the current AI definition under a new name), or one expanding or superseding the AI definition, changing which systems federal programmes cover. None can be known until the proposal appears.
The IAPP highlights the practical problem meanwhile: organisations dealing with the federal government “may soon encounter ‘SI’ in federal agency materials while statutes, state laws, contracts, internal policies, vendor questionnaires and governance frameworks continue to use ‘AI'”. Its advice: manage two vocabularies — anchor legal references to each instrument’s own language, add a crosswalk where needed, and watch the proposal.
Why the name matters, and why it does not
Terminology is never purely cosmetic: calling a technology “super intelligence” invites the public to think of it as already superhuman, inflating expectations and dampening scepticism, and confuses partners translating between the federal vocabulary and the one used everywhere else.
At the same time, the name changes nothing about what the technology can do, what laws apply, or what safeguards exist: systems called “super intelligence” in a federal press release are the same systems called “artificial intelligence” in a statute. Whether renaming becomes more than presentational depends on the late-November definition proposal and on Congress. Until then, use the context-appropriate term and say which meaning is intended when it matters. For the political context, weeks before the 2026 midterm elections, see our report on the Reuters/Ipsos poll and the generic ballot.
Frequently asked questions
What is “super intelligence” in the Trump executive order?
A new federal label for the technologies covered by the statutory definition of “artificial intelligence” (15 U.S.C. § 9401(3)). Executive Order 14434 (29 September 2026) directs agencies to use “super intelligence” and “SI” instead of “artificial intelligence” and “AI” in executive-branch communications and other non-statutory documents.
Does the order mean superintelligence has been achieved?
No. In research, superintelligence means an intellect greatly exceeding the best human brains in practically every field (Bostrom) — hypothetical, undemonstrated. The order sets no test, claims no system has met it, and defines its term by borrowing the AI definition.
Does the order change the law or ban the term “artificial intelligence”?
No. Statutes say “artificial intelligence”. The order covers executive-branch documents, does not require rewriting existing regulations, contracts or historical documents, creates no enforceable rights, and imposes nothing on private companies, researchers or the public.
What happens next on the definition?
Within 60 days of 29 September 2026 (late November 2026), the President’s science and technology adviser must submit proposed legislative language for a federal definition, assessing whether it should modify, expand or supersede the AI definition. Until it appears — and unless Congress acts — the borrowed definition governs.
Updated: 28 October 2026. The definition proposal required by Executive Order 14434 was still pending when this article was written. This explainer will be updated if the proposal is published.
Sources
- The White House, Executive Order 14434, “Inaugurating the Era of Super Intelligence”, 29 September 2026 — https://www.whitehouse.gov/presidential-actions/2026/09/inaugurating-the-era-of-super-intelligence/
- IAPP, “AI by any other name? White House directs federal agencies to enter the era of ‘super intelligence'”, 30 September 2026 — https://iapp.org/news/a/ai-by-any-other-name-white-house-directs-federal-agencies-to-enter-the-era-of-super-intelligence
- Infosecurity Magazine, “Trump, Six AI Giants Sign ‘Super Intelligence’ Safety Accord”, 30 September 2026 — https://www.infosecurity-magazine.com/news/trump-ai-giants-super-intelligence/
- Philippine News Agency, “Trump calls users of term ‘artificial intelligence’ the ‘enemy'”, 14 October 2026 — https://www.pna.gov.ph/article/trump-calls-users-of-term-artificial-intelligence-the-enemy