Trump Renamed AI to "Super Intelligence" Across the Federal Government. Nothing About AI Actually Changed.

On September 29, 2026, President Donald Trump signed Executive Order 14434, titled "Inaugurating the Era of Super Intelligence." It directs every executive department and agency to replace "artificial intelligence" and "AI" with "Super Intelligence" and "SI" in official correspondence, public communications, websites, reports, policy documents, and other non-statutory documents, to the maximum extent permitted by law. The order was published in the Federal Register on October 2.

It is one of the strangest federal technology policies of the year, and one of the most revealing. (For a substantive version of this accountability impulse, see the White House's mandatory AI incident reporting order.) The order does not regulate AI systems, fund them, restrict them, or define a new class of capability. It changes what the government calls them.

Then, on October 8, Trump escalated the rhetoric on Truth Social, writing that people who use "artificial intelligence" instead of "Super Intelligence" are enemies. The post did not clarify whether that applied to government personnel or to everyone. That same day, AI.gov, the administration's AI strategy portal, swapped its logo for "SI GOV" and changed its browser title from "AI Strategy" to "SI Strategy."

What the order actually requires

The operative part of EO 14434 is narrow. Section 2(a) tells agencies to use the new terms in forward-looking communications and documents. Section 2(b) explicitly says agencies do not have to rewrite previously issued regulations, presidential actions, contracts, grants, or other historical documents. Section 4 states the order creates no right enforceable against the government in court. There are no penalties anywhere in it.

Most importantly, Section 3(a) defines "Super Intelligence" and "SI" as the technologies and systems covered by "artificial intelligence" as already defined in federal law, at 15 U.S.C. 9401(3). That statute, from the National Artificial Intelligence Initiative Act of 2020, covers machine-based systems that make predictions, recommendations, or decisions. In other words, SI is whatever AI already was. There is no new capability threshold, no test, and no new category of technology. Anything that counted as AI on September 28 counts as SI on September 30, nothing more and nothing less.

Freshfields, the international law firm, noted in its analysis that the order targets non-statutory documents only. Existing statutes still say AI, and statutes outrank executive orders. Congress would have to act for the legal definition to change.

The order renames an existing legal category without changing what sits inside it. The vocabulary is new. The law, the systems, and the obligations are not.

Why the administration says it wants the name

Trump's objection to the term predates the order. In July 2025 he complained publicly that "artificial" was an unsuitable description of technology he regarded as genius. On September 19, 2026, he floated alternatives on X, including Superior, Extreme, and Supreme Intelligence. The administration's formal fact sheet, released with the order on September 29, frames SI as language that better expresses technological progress and American leadership.

The stated logic is partly about branding: "artificial" sounds dismissive of the technology's value, while "super" conveys ambition and national competitiveness. It is worth noting the term has also been politically useful at a moment when public anxiety about AI is high. A Plymouth Union poll of 1,000 likely voters conducted September 25-28 found that about 62 percent said AI presents a risk to humanity and 72 percent somewhat or strongly supported slowing AI development. A new name does not address any of those concerns, but it does move the conversation onto friendlier terrain.

The awkward part: the rollout is already half broken

Government vocabulary changes live or die on implementation, and implementation is visibly messy. AI.gov's rebrand on October 8 was incomplete: the site still runs at ai.gov, its tab title still begins with "AI.Gov," its search metadata still says "artificial intelligence," and its organizational name is unchanged. Links labeled "AI Education" and "Presidential AI Challenge" retained the old terminology. The domain itself would require an act closer to legislation than rebranding to change, since .gov domains are managed under federal authority and renaming ai.gov would break every link, contract reference, and bookmark pointing at it.

Elsewhere, adoption has been uneven. The State Department's Bureau of International Organization Affairs moved first: according to AP reporting, deputy assistant secretary Michael Drager emailed staff under the subject line "Change in Nomenclature AI to SI," directing the bureau to use SI in all remarks, positions, and press material. Because that bureau oversees U.S. missions to the UN and other multilateral bodies, the instruction reaches into international diplomacy, where every treaty, trade standard, and allied framework says AI, not SI. It was unclear, as of mid-October, whether other bureaus would follow.

What comes next: the 60-day clock

The order's one substantive forward-looking move is in Section 3(b). It directs the Assistant to the President for Science and Technology, Michael Kratsios, to submit proposed legislative language for a federal definition of "Super Intelligence" within 60 days of the order, a deadline that lands around November 28, 2026. That proposal is supposed to address whether the new definition should modify, expand, or supersede the statutory definition of AI.

This is where the rebrand could turn from theater into policy. If Congress were to adopt SI as a statutory term, every federal AI statute, and potentially references in regulations, procurement rules, and standards, would face revision. As of October 13, there is no indication such legislation has been drafted, and congressional calendars make action before the deadline unlikely. For now, the 60-day deliverable is a proposal, not a law.

There is also a geopolitical angle. An October 12 Washington Examiner op-ed argued that abandoning the "AI" label effectively surrenders a globally established brand to Beijing, since the rest of the world, including China and the EU, will keep saying AI. The piece proposed alternatives like "Accelerated Intelligence," which would keep the A. That argument has a practical core: in international standards bodies, the terminology is settled, and a unilateral American rename creates friction without leverage.

The bigger pattern: language as the policy itself

EO 14434 arrived on the same day as the White House Accord on Super Intelligence, a voluntary agreement signed September 29 by CEOs of Google, Meta, Anthropic, OpenAI, Nvidia, and xAI, committing to internal controls, independent assessments, and board-level oversight for frontier models, with no enforcement mechanism. The accord, like the order, runs on pledges and labels rather than rules.

Researchers and policy experts have been skeptical. Elon University polling director Lee Rainie and other scholars have argued that rebranding does not touch the actual sources of public worry: job loss, privacy, and safety. Words do not patch security holes or set enforceable rules. The term "superintelligence" also carries a specific meaning in AI research, where it has long described hypothetical systems that eclipse human intelligence across the board, a category today's models do not occupy. Using it for current systems risks confusing the public about what the technology can actually do.

There is a real policy question underneath the branding fight: whether the federal government should speak about AI with the same vocabulary as everyone else, or assert its own. For now, the answer is that it is trying to assert its own, unevenly, at the level of paperwork. The machines remain what they were on September 28. Only the memos are super.