Trump rebrands AI as 'Super Intelligence' by executive order and signs a voluntary accord with six AI CEOs

FACT (CONFIRMED). On Tuesday 29 September 2026, President Donald Trump signed Executive Order 14434, "Inaugurating the Era of Super Intelligence." The order directs the executive branch, "to the maximum extent permitted by law," to use the terms "Super Intelligence" and "SI" in place of "Artificial Intelligence" and "AI" in official correspondence, public communications, websites, reports, policy documents and other non-statutory documents — and states the executive branch "will not acknowledge the usage of 'Artificial Intelligence' and 'AI' in any applicable setting." It was published in the Federal Register on 2026-10-02 at 91 FR 63129 (document 2026-20321).

FACT (CONFIRMED). The same day, at a White House tech summit (closed-door luncheon in the East Room, followed by ~30 minutes of questions in the West Wing driveway with House Speaker Mike Johnson), Trump and six AI-company leaders signed the "White House Accord on Super Intelligence — Joint Commitment on Frontier Responsibilities": Sundar Pichai (Google), Dario Amodei (Anthropic), Mark Zuckerberg (Meta), Greg Brockman (OpenAI), Elon Musk (xAI, "now part of SpaceX") and Jensen Huang (Nvidia). Trump posted the signed document on Truth Social late on 2026-09-29; the White House communications team emailed it out shortly after signing.

FACT (CONFIRMED). Trump characterised the accord as "morally binding," said "It's almost like a constitution, in a way. And the biggest people in the world signed that, and I signed it as president. And it really is a form of protection," and said he was considering a ~10-person board to "watch over the whole enterprise," plus a named official to oversee the agreement. He also said: "I think I'm seeing tremendous self-policing. And they understand that they have to self-police."

FACT (CONFIRMED). The accord's signature block misspells the country as "President of the Unites States" under Trump's signature. Bloomberg's David Gura appears to have been first to notice it; Gizmodo's Matt Novak also flagged it early (per TechCrunch). It remained visible as of the morning of 2026-10-01.

FACT (CONFIRMED). On 2026-10-03 (WSJ exclusive, relayed by Reuters/CNBC) and via a 2026-10-04 Truth Social post, the administration named DNI Jay Clayton to lead the "Super Intelligence Force," a task force with 120 days to report on AI risks and opportunities and on the federal government's responsibility. Other named members: FTC Chairman Andrew Ferguson, Pentagon CTO Emil Michael, OPM Director Scott Kupor. It reports to Trump and chief of staff Susie Wiles.

CONTEXT (CONFIRMED). The same day also brought: an America.gov AI chatbot launch ("Streamlining Access to Government Services Through America.gov," a separate presidential action of 2026-09-29), at which Trump said "I will never stifle the growth of a technology that will be bigger than the industrial revolution," and a daylong "Golden Age" event where officials used "Super Intelligence" terminology. Trump had already pre-announced the rename at the UN on 2026-09-22 ("the Artificial Intelligence being spoken of so much now — hereinafter officially called 'Super Intelligence'").

FACT. Federal vocabulary changed by executive order: across the executive branch's non-statutory output, the government is to write "Super Intelligence"/"SI" rather than "AI"/"Artificial Intelligence." Statutes are untouched — the EO defines "Super Intelligence" and "SI" as exactly what 15 U.S.C. § 9401(3) means by "artificial intelligence," and does not require altering previously issued regulations, contracts, grants or historical documents.

FACT. For the first time, the US government has a signed, published, industry-wide voluntary frontier-AI safety accord rather than a rulemaking: four layers of controls and audits (internal controls during training/deployment covering cybersecurity, biosecurity and chemical threats and unintended system access; an empowered internal team; an independent external auditor/evaluator; an independent board committee), plus a commitment that participating companies "will meet regularly to establish standards and best practices."

COMPANY CLAIM / GOVERNMENT CLAIM. The administration frames the package as recognition that these technologies "far exceed" what "Artificial Intelligence" implied, and as proof that self-regulation works ("tremendous self-policing").

INTERPRETATION. Two different kinds of instrument were used deliberately: an order with real administrative force over language (agencies must comply in non-statutory documents) and a document with no legal force over safety practice (the accord). The binding tool was spent on rebranding; the substantive safety commitment is voluntary. That inversion — enforceable terminology, unenforceable safety — is the actual change.

FACT — the EO's mechanics.

• Sec. 1 (Purpose): policy statement — the executive branch shall use "Super Intelligence"/"SI" in place of "Artificial Intelligence"/"AI" "to the maximum extent permitted by law" and "will not acknowledge the usage of 'Artificial Intelligence' and 'AI' in any applicable setting."
• Sec. 2 (Implementation): applies to agencies' official correspondence, public communications, websites, reports, policy documents and "other non-statutory documents"; explicitly does not require altering previously issued regulations, presidential actions, contracts, grants or other historical documents. Statutes and regulations already on the books therefore keep saying "AI."
• Sec. 3 (Definition): for the order's purposes, "Super Intelligence"/"SI" mean what section 9401(3) of title 15, US Code means by "artificial intelligence" — so nothing semantic changes yet; only the label does. Within 60 days, the Assistant to the President for Science and Technology (APST) must submit proposed legislative language for a federal definition reflecting the capabilities described in Sec. 1, an assessment of whether it should modify/expand/supersede the statutory AI definition, proposed conforming amendments to statutory references, and recommendations for further presidential/executive action.
• Sec. 4 (General Provisions): standard savings clauses; implementation "subject to the availability of appropriations"; and an explicit statement that the order creates no right or benefit enforceable at law or in equity by any party.

FACT — the accord's mechanics (four layers, verbatim structure). Each participating company should:

1. "Implement robust internal controls to monitor the capabilities and alignment of its models during training and deployment around areas like cybersecurity, biosecurity, and chemical threats, and to ensure that its models do not hack or access technical systems in unintended ways."
2. "Empower an internal team to ensure all of the controls, monitoring, and detection are operating as intended, and that any issues are remediated."
3. "Partner with an independent external auditor or evaluator to carry out independent assessments of whether the controls, monitoring, and detection are operating as intended."
4. "Designate an independent committee of the board of directors to oversee and receive reports from the teams operating the controls and the internal and external auditors and evaluators, as well as to ensure any issues identified are remediated."

Plus: participating companies "will meet regularly to establish standards and best practices"; "Over time, it may make sense to codify these steps into laws or regulations"; and regardless of legal requirement, "each of our companies are committed to doing this." There is no defined scope of covered models, no metric, no reporting cadence, no penalty, no signatory beyond the seven named, and no duration or withdrawal clause in the published text.

FACT — the oversight mechanism announced after the fact. The "Super Intelligence Force" is a coordinating task force (not a regulator) that will reach out to consumers, public-interest groups, religious organisations, critical-infrastructure providers and AI companies, and report to Trump and Susie Wiles; per WSJ/Reuters/CNBC it has 120 days to report. Trump separately floated a ~10-person board to oversee AI safety, with members unnamed inside the window.

INDEPENDENT EVIDENCE. The accord's operative wording was independently checked across four renderings (White House-issued PDF as transcribed by Wikisource, AP, BBC, Forbes — all quote the same four layers and the codification sentence), and the EO's text was compared word-for-word between whitehouse.gov and the Federal Register PDF (23 of 24 sentences match verbatim; the one mismatch is a text-extraction artefact, not a wording difference).

INTERPRETATION. The design is a "commitment plus light-touch scaffolding" model: a public pledge whose only real teeth are reputational (association of seven of the most valuable companies' names with the four layers) and whose only stated pathway to force is the final sentence inviting future legislation.

--

• FACT: The US federal government has chosen voluntary self-policing as its frontier-AI instrument while spending its one clearly binding instrument (an executive order) on terminology. AP explicitly notes that "some of the actions outlined in the accord are measures these companies are already taking in some form or have previously committed to doing."
• INDEPENDENT EVIDENCE: Reactions are split. Bloomberg Government reports technology lobbyists, executives and safety advocates divided over the pact. USC's Shri Narayanan told AP the balance favours innovation but includes safety-improvement effort; USC's Robin Jia said it is risky to put "so much faith in self-policing" and that lab intentions must be "well-aligned with the broader interest of the American public," pointing to recent hacking incidents; Berkman Klein Center's Alex Pascal said the only way to reduce AI harms "is through robust legal liability, regulation and fundamentally changing the race dynamics." BBC quotes Weatherall: "The accord should be ignored, for the distraction it is," arguing it lets firms "define what counts as safety" and states no consequences for breaches.
• FACT: Public opinion, per Reuters/Ipsos (17–20 September 2026, cited by Reuters): 73% worry AI companies have not done enough to prevent serious harm; 55% say it would be good to slow AI development. The Guardian cites a NYT/Siena poll (61% of voters opposed data-centre construction) and SSRS/CNN (71% say government is not doing enough to regulate AI). The accord lands against that backdrop and ahead of the November midterms (Bloomberg Government frames it as a way for Trump to address risk concerns "during a pivotal midterm campaign season" without new rules).
• INTERPRETATION: The rebrand is not cosmetic plumbing: it reshapes how agencies talk about (and therefore budget, procure and prioritise) the technology, while the accord converts what could have been regulation into a press event with industry names attached. Enterprises now face a governance landscape where the federal signal is "self-police," states and agencies retain their own tools, and the only new federal body is a 120-day task force.
• INTERPRETATION: The typo is small but symbolically loud: a document whose entire premise is careful stewardship of a transformative technology was released with the country's name misspelled under the president's signature, and it became the most-shared detail of the day (Guardian, TechCrunch, CBS Austin, WCVB all led with it; California Gov. Newsom mocked it).

• FACT (possible today): Federal agencies must now write "Super Intelligence"/"SI" in non-statutory documents, and vendors selling into government can expect rebranded solicitations, reports and websites; the statutory definition is unchanged, so contracts and compliance obligations still key off 15 U.S.C. § 9401(3).
• FACT (possible today): Signatory companies have a public, named commitment to external audit and board-level oversight committees they can cite in procurement and investor materials — and, symmetrically, non-signatories can be asked why they did not sign.
• PREDICTION: The accord's "codify over time" sentence becomes the anchor for later legislative drafting — the APST's 60-day definition proposal (~2026-11-28) is the first place to watch, and its conforming-amendment list will reveal how much statutory "AI" language the administration is willing to reopen.
• PREDICTION: "Independent external auditor" and "independent board committee" become the two fastest-productising phrases in AI governance consulting, because they are the only concrete obligations with a name (and they are already common in adjacent assurance regimes).
• PREDICTION: States, Congress and agencies fill the vacuum the accord leaves: expect more state-level AI measures and FTC/agency activity framed explicitly as the counterweight to voluntary federal policy (the FTC chair's presence on the task force cuts both ways).
• PREDICTION: The terminology creates a durable compliance edge case — every government dataset, API spec and RFP now needs dual-term handling ("AI" in statutory/regulatory text, "SI" in new prose), and tools that normalise between them are a small but real category.

1. What does the APST's 60-day legislative proposal (~2026-11-28) actually say — does it seek to amend statutory definitions, and which conforming amendments are proposed?
2. Who are the "~10 people" on the oversight board Trump said he would name, and what powers would they have over accord compliance?
3. Which independent auditors will the six signatories engage, under what standards, and will audit summaries be public?
4. What are the "regular" standards meetings producing — a shared eval suite, incident taxonomy, or nothing public?
5. Will the "Super Intelligence Force" (120 days from ~2026-10-03/04) recommend codification, and does its report carry any mechanism beyond persuasion?
6. Do non-signatories (Microsoft, Amazon, Palantir and others present at or near the event — ABC reported Bezos, Karp, Sankar attending the summit) face any pressure to sign, and did any of them?
7. Does "no longer acknowledge the terms 'Artificial Intelligence' or 'AI'" survive contact with statutes, procurement law and public-comment requirements — and how will agencies handle public-facing text where clarity matters?
8. Will the Federal Register version of the accord itself ever be published (the Guardian noted it was unclear whether the document would appear in the FR), and does the typo appear in any official reprint?

• FACT : EO 14434 published 2026-10-02 (91 FR 63129); "Super Intelligence Force" announced 2026-10-03/04 with 120 days to report, reporting to Trump and Susie Wiles; members: Jay Clayton (DNI), Andrew Ferguson (FTC), Emil Michael (Pentagon CTO), Scott Kupor (OPM).
• FACT : The EO sets a 60-day clock (from 2026-09-29) for the APST to submit proposed legislative language for a federal "Super Intelligence"/"SI" definition plus conforming statutory amendments and recommendations for further executive action.
• FACT : Trump said he would name a person to oversee the accord "in coming days" after consulting with industry and floated a ~10-person safety board; neither was named inside the window.
• FACT : The accord commits signatories to meet "regularly" to establish standards and best practices, and explicitly contemplates future codification into law or regulation.
• PREDICTION: The two dates to watch are the APST proposal (~2026-11-28) and the task-force report (~late January 2027); both will show whether the accord was a staging document for codification or a substitute for it.
• PREDICTION: Until then, expect the terminology migration to surface in federal RFPs and sites, vendor questionnaires citing the four layers, and continued state/agency action as the practical governance layer.

--

The binding order was about words; the safety commitment was about vibes — and both were signed the same afternoon. EO 14434 makes the US government write "Super Intelligence" instead of "AI" while defining the new term as exactly the old statutory one, and dispatches the APST to propose a real definition within 60 days. The White House Accord on Super Intelligence asks six of the world's most powerful companies to do four sensible things — monitor, staff, audit, oversee at board level — with no metrics, no penalties, no scope and an explicit invitation to turn it into law "over time." Trump calls it "morally binding" and "almost like a constitution"; independent experts call it a distraction that lets companies define their own safety. The confirmable facts are unusually solid (two official renderings of the EO match nearly word-for-word, the accord text and signatories are corroborated across four independent outlets), and the emblem of the whole package is a one-letter error — "President of the Unites States" — printed under the signature on an image-only PDF. The story to tell: this is what US frontier-AI governance looks like when the enforceable instrument is spent on branding and the substantive instrument is spent on goodwill, with a 120-day task force and a 60-day definition proposal as the only scheduled tests of whether either becomes real.