October 16, 2026

They Renamed AI "Super Intelligence." I'm Still Calling It AI.

Federal agencies now have to call artificial intelligence "super intelligence," and this week Washington started pushing the name abroad. The new label claims something the technology hasn't earned, so AI Next Wave is keeping the old one.

By Jay Cadmus

A federal office desk at dawn with a stack of official documents and two blank brass desk nameplates, a closed laptop in soft focus behind them, and an American flag by the window.

The Short Version

On September 22, speaking to the United Nations General Assembly, President Trump said the word "artificial" makes the technology "sound fake and it is not fake," and announced that U.S. documents would switch to "super intelligence." A week later, on September 29, he signed an executive order telling federal agencies to use "Super Intelligence" and "SI" in place of "Artificial Intelligence" and "AI" to the maximum extent permitted by law. The State Department had already told diplomats in its international organizations bureau to use the new term in all communications. On October 6, the Justice Department told its staff to do the same in most circumstances, including court filings when appropriate. This week the push went global. A Treasury official told Reuters on Thursday that the U.S. is negotiating to get "super intelligence" into a statement from the International Monetary Fund's steering committee, due Saturday in Bangkok. I cover the same technology those agencies are renaming. Here, it's staying AI.

The Word Already Had a Job

Superintelligence had a meaning long before Washington borrowed it. Inside the field, it's a term of art for a hypothetical machine that outthinks people at nearly everything. Philosopher Nick Bostrom's widely cited 2014 definition is "any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest." Researchers still use the word that way, and they describe the thing it names as theoretical.

Members of Congress use it that way too. In September, Senator Bernie Sanders and Representative Greg Casar unveiled a bill that would permanently ban artificial superintelligence, which it describes as a system that "exceeds human cognitive ability" or could plan humanity's destruction. Their bill treats superintelligence as a danger to prevent before it shows up.

The executive order concedes the gap in its own definitions section. For now, its "SI" covers exactly the technologies already covered by the legal definition of artificial intelligence in Title 15 of the U.S. Code. Same chatbots, same image generators, same recommendation engines. The order changed the label and borrowed the old definition to fill it.

That's the whole problem in one move. A word that meant a specific, frightening, not-yet-real thing now also means the chatbot that writes your meeting summary. Put the two documents side by side and it gets absurd: one proposal would outlaw superintelligence, while a federal order hands the same name to customer service bots. When a federal filing says "SI," a reader has no way to tell which meaning it carries. That kind of confusion is what you get when a name is picked for how it sounds over what it means.

What the Order Does, and What to Watch

To be fair to the stated reasoning, the order argues that today's frontier systems go well past imitating or automating human intelligence, and the administration's case is that a term coined about 70 years ago no longer fits them. That part has something to it. A phrase from the era of room-sized computers now covers software running on ordinary phones. Names can outgrow their origins.

The order itself reaches less far than the speech did. Trump told the U.N. that all U.S. documents would change, and he hoped the rest of the world's would too. The signed order covers new official correspondence, public communications, websites, reports and policy documents. Statutes keep their wording, and agencies are spared from rewriting old regulations, contracts or grants. Private companies sit outside it entirely, and they've split. SpaceX renamed its AI division SpaceXSI, Business Insider reported, while OpenAI's Sam Altman said of a rebrand, "No plans to do that."

The part worth watching has a date on it. The order gives the president's science and technology adviser 60 days from September 29, which lands around November 28, to send proposed legislative language defining "Super Intelligence." That proposal has to say whether the new definition should modify, expand or supersede the definition of AI that Congress already wrote into law. A proposal binds no one on arrival. If Congress or a later presidential action adopts a definition that differs from the current one, though, the rename stops being cosmetic. Contract language, procurement rules and compliance paperwork key off legal definitions. That's where a branding exercise becomes a regulatory fact, and it'll happen in documents almost nobody reads.

Washington Is Taking the Name Abroad

Thursday's move shows the administration wants other governments using the label too. Erin Browne, Treasury's undersecretary for international affairs, told Reuters in Bangkok that the U.S. is negotiating with other IMF shareholders to use "super intelligence" in the statement the fund's steering committee is drafting. "President Trump has been really clear on this," she said, and she argued the new term fits the technology better. She declined to say whether Washington would withhold its sign-off without the change.

That insistence raised eyebrows among some other members, according to the same Reuters report, and a person familiar with the talks said the final statement might sidestep both terms and refer to "new technology." The split showed up in public on Friday in Bangkok. IMF Managing Director Kristalina Georgieva kept saying AI on a panel, where she warned that without safeguards its risks could outweigh its opportunities. Federal Reserve Chair Kevin Warsh offered the American version: "In the US we now pronounce it SI."

Other finance officials picked a side the same day. Germany's finance minister, Lars Klingbeil, told Reuters, "I will continue to call it artificial intelligence." Bank of England Governor Andrew Bailey joked that he wasn't sure anyone was still allowed to say AI. By Friday evening, Washington had started an international naming dispute and had yet to win one.

Back home, the White House moved its AI policy website to si.gov this week, and White House science policy director Michael Kratsios posted, "The Super Intelligence era is here." So for the next few months, a reader following global economic news may see one technology under three names: AI from the IMF, SI from Washington, and possibly "new technology" from diplomats trying to keep everyone signed on.

Why I Won't Use It

Words in this field carry weight, and "artificial intelligence" describes what these systems are: intelligence simulated by machines, built by people, trained on human work. The "artificial" in the name is the honest part, and it insults nobody. These systems produce fluent, useful, sometimes remarkable output. They also make things up and take actions nobody asked for. One Anthropic model sent a false homicide tip to Philadelphia police, Engadget noted on Thursday, in a story about the FTC sending civil subpoenas to OpenAI and Anthropic that seek their records on security incidents and safeguards. Calling that "super intelligence" is advertising.

I don't take naming instructions for the technology I cover. The federal government can set the vocabulary of its own documents; that's its right, and agencies will follow the order. AI Next Wave answers to its readers, and readers need the plain name. My job is to read past the headline, and a government-mandated rename is exactly the kind of headline that needs reading past. Underneath it, the technology stayed the same on September 29. Only the sales pitch changed.

So my policy here is simple, and it's permanent. I'll keep calling AI what it is, and I always will. If a federal document says "SI," I'll translate it for you, and I'll tell you when the November definition arrives and whether it changes anything that touches your money, your work or your rights. The name was fine for seventy years. It's fine now.

Reading an SI Document Without Getting Lost

Picture Dana, who runs a 12-person software firm that bids on federal work. In December she opens a solicitation that asks vendors to describe their "SI capabilities." Before she changes a word of her proposal, she checks which definition the solicitation cites. If it points to the order's definition, "SI" means the same tools her team already sells, and her standard AI language answers it once she adds a sentence mapping one term to the other. If it cites a new definition that came out of the November proposal, she reads that definition against her product before claiming anything, because a company that says it sells "super intelligence" is making a promise it may have to prove. Her safeguard is small and repeatable: one page that maps each agency's term to the legal definition it cites, updated when the November proposal is published, and checked by whoever signs the bid. Where it can fail is a solicitation that uses "SI" with no definition at all. That's a question for the contracting officer, in writing, before the deadline.

What You Can Do With This

If you read government documents for work

When you see "SI" in a federal memo, website or court filing, read it as AI unless the document cites a different definition. The order's own definitions section says that's what it means for now. Keep the current Title 15 definition bookmarked so you can spot the day a document cites something new.

If you follow AI safety news

Translate as you go. The White House task force called the Super Intelligence Force held its first meeting on Thursday and adopted its charter, Semafor reported, and FTC Chair Andrew Ferguson, one of its vice chairs, will lead a new consumer-protection committee with the Justice Department. Its demand that companies notify regulators and fix problems when incidents happen is about today's chatbots and agents, whatever the letterhead calls them.

If you sell AI tools

Your product name can stay as it is, because the order binds federal agencies and leaves companies alone. Be careful about the reverse move. Calling your product "super intelligence" to match federal language invites a claim you'd have to back up, and the FTC is already asking AI companies for their marketing materials.

If you watch the international side

Read the IMF steering committee's statement when it comes out Saturday. Whether it says AI, SI or "new technology" will show how far other governments are willing to follow Washington's vocabulary.

The Bigger Picture

Names in technology policy end up doing legal work. Today the rename is mostly vocabulary: same systems, same statute, new letters on federal letterhead. The November proposal could change that, and the IMF talks show Washington wants other governments to adopt the word before anyone has settled what it means in law. A label that promises more than human intelligence makes every chatbot sound more capable than it is, at the same moment regulators are subpoenaing records about what these systems got wrong. My prediction, and it's only a prediction: the late-November definition will matter more than the rename itself, because that's where "SI" either stays a synonym or starts meaning something new on paper that contracts and courts have to follow. Until then, the most useful thing I can give you is a plain name. It's AI, and I'll keep calling it that.