US Order 14434 Strips the Word "AI" Out of Executive Branch Documents - A "Super Intelligence Force" Follows in a Post

Executive Order 14434, signed September 29, 2026, tells executive agencies to write "Super Intelligence" and "SI" instead of "Artificial Intelligence" and "AI." The definition still borrows the existing statutory one for AI. A Super Intelligence Force was then announced in a post.

policy regulation United-States government terminology AI-governance

Executive Order 14434, signed on September 29, 2026, instructs United States executive agencies to write “Super Intelligence” and “SI” where they would have written “Artificial Intelligence” and “AI”¹. The scope runs across official correspondence, public communications, agency websites, reports and policy documents, plus other material that is not statutory.

The order’s policy statement goes further than swapping terms. To the greatest extent the law allows, the executive branch is to use those two terms and is not to acknowledge the use of “Artificial Intelligence” or “AI” in any applicable setting¹. The White House fact sheet summarizes it the same way, as directing every department and agency to stop acknowledging both terms².

Alongside that, President Trump used a post on his own social network to announce the formation of a “Super Intelligence Force” (SIF)³. TechCrunch describes it as a Sunday morning post, and the publication’s own dateline is October 4⁴.

What the order changes, and what it leaves

The instruction is drawn along document types. What changes is the non-statutory material listed above. Regulations already issued, prior presidential actions, contracts, grants and other historical documents are explicitly not required to be altered¹.

The definition does not move either. Under the order, both new terms denote whatever falls under “artificial intelligence” as 15 U.S.C. § 9401(3) defines it, and that holds until a later presidential action or an Act of Congress displaces it¹. The label is swapped; what it points at remains the statutory definition of AI.

The statutory side has its own instruction. Within 60 days of signing, the Assistant to the President for Science and Technology (APST) is to deliver proposed legislative language to the President for a federal definition of the two terms¹. The submission is to include an assessment of whether the new definition should modify, expand on or supersede the existing statutory one, any conforming amendments to current statutory references, and recommendations for further action needed to carry the definition through the executive branch. Counting from the signing date puts that deadline at November 28, 2026.

The closing section carries a clause stating the order creates no legally enforceable right or benefit for any party¹. Standardizing the vocabulary is an internal operating instruction, not an obligation placed on the private sector.

The rewriting already shows up in the fact sheet itself

How the substitution works in practice is visible in the fact sheet from the same day. Where it lists past initiatives, the generic noun has become “SI.” Biden administration policy is described as “harmful SI policies,” and for America’s AI Action Plan from July 2025, the proper name keeps its “AI” while the surrounding description talks about winning the SI race, accelerating SI innovation and building American SI infrastructure².

So within one document, the proper name says AI and the running prose says SI. Together with the order not requiring old documents to be revised, that suggests a split — programs and plans keeping their AI names while the text describing them says SI — will be around for a while.

The fact sheet also cites the National Super Intelligence Policy Framework unveiled in March 2026, and more than five billion dollars announced for the Genesis Mission, which it calls a nationwide push to use SI to speed up breakthroughs in medicine, energy, manufacturing and science².

What the post says the SIF will be

For the SIF, the only primary record available right now is the post itself. As of October 5, 2026, no corresponding executive order or memorandum appears in the White House list of presidential actions or turns up in a site search. Everything below is what the post states³.

The post gives the SIF’s task as coordinating the federal government’s effort so that America keeps leading the world in Super Intelligence. Five groups are named as the ones it will deal with on the government’s behalf: critical infrastructure operators and Super Intelligence companies, plus consumers, public interest organizations and religious bodies.

Four people are listed as leading it: Jay Clayton, the Director of National Intelligence; Andrew Ferguson, who chairs the Federal Trade Commission; Emil Michael, Under Secretary of War for Research and Engineering and also Chief Technology Officer; and Scott Kupor, who runs the Office of Personnel Management. It reports to the President, and to Susie Wiles as Chief of Staff.

The post also frames the SIF as following the White House Accord on Super Intelligence, which it says brought the leading technology companies together to affirm what they owe the American public³. No full text or summary of that accord appears on the White House site, so the signatories and the contents are not covered here.

What the listed titles suggest

The four titles in the post cover areas that barely overlap: the intelligence community, competition and consumer protection, defense research and engineering, and federal personnel. Read together with a counterpart list that stretches from industry and infrastructure operators to consumers, advocacy groups and religious bodies, this looks like a plan to consolidate the channel to industry under a cross-agency banner.

That said, the post says nothing about the legal basis for the body’s authority, its budget, its decision-making procedure or its relationship to existing bodies.

Reporting is filling in the rest for now. TechCrunch, citing the Wall Street Journal, says Clayton will chair the SIF, with Ferguson, Michael and Kupor serving as vice chairs. The publication further reports that the task force is said to have 120 days in which to write up where AI carries risk and where it offers opportunity, and that its charter is said to call for drawing up plans to answer SI-enabled threats to society, while guarding against regulatory capture and rules so heavy that innovation and competition suffer⁴. All of it is written as secondhand via the WSJ, and as of October 5, 2026 nothing in the White House presidential actions or fact sheets backs it up.

What bites when the vocabulary changes

Among this administration’s AI-related executive actions, Order 14409 of June 2, 2026 directed a voluntary pre-release review framework for models, and that framework had reached the point of being assembled in August. Order 14434 differs from those: it commands uniform vocabulary rather than the substance of any review or obligation.

What bites first in practice is probably searching and mapping. For anyone following federal procurement specifications, guidance and agency web pages, more of what gets published from here will not turn up under “AI.” Because old documents are not rewritten, a regulation written in AI terms and an explanation written in SI terms will describe the same program side by side. If internal documentation at a company references government material, widening it to catch both words is the safer move.

Regional divergence is also widening. California’s executive order and the EU’s guidelines on AI Act transparency obligations both build their regimes around the word AI. If only federal non-statutory documents move to a different word, the vocabulary of documents written about the same business will vary by who receives them.

What happens to the statutory language will take shape in the legislative proposal due November 28. Whether that proposal replaces the existing statutory definition or settles for carrying both will decide how long the doubled vocabulary lasts.