On Tuesday the President of the United States decided that the word "artificial" was the problem. In an executive order, Trump directed that federal documents, and, he hoped, eventually the world's, drop "artificial intelligence" in favour of "super intelligence." His reasoning was aesthetic: "artificial" makes the technology sound "fake," he said, and it is not fake, it is "actually amazing." The same day he unveiled America.gov, an AI portal that answers questions about federal services, and gathered the heads of OpenAI, Google, Meta, Nvidia, Microsoft and others to sign what he called a "morally binding" document. It was a lot of ceremony for one afternoon, and it is worth asking what all the ceremony was for.
Start with America.gov, the most tangible of the three. The pitch is clean: the administration says it has condensed 29,000 government websites into a single AI model so citizens can stop "wandering from site to site" to find out how to get a passport or whether they qualify for Medicare. It is free, carries no ads, and, the site promises, draws answers "only from official government sources" with links back to the originals. That last detail is also a quiet admission, because the site notes in the same breath that "since it is being run by AI, the technology can sometimes make mistakes and users should check the sources." A government tool that tells you to double-check the government is a fair description of where consumer AI actually is: useful, fast, and not to be trusted without verification.
The rename is harder to take at face value. Language is not neutral, and swapping "artificial" for "super" does real rhetorical work: it trades a word that invites scepticism for one that promises transcendence. It also, conveniently, echoes the "Super Intelligence Dialogue" the administration announced with Beijing days earlier, knitting the domestic message and the diplomatic one into a single brand. Whether federal agencies will actually rewrite their style guides is almost beside the point. The order is a messaging instrument, designed to frame a technology the public is increasingly wary of as a marvel rather than a menace.
Which brings us to the "morally binding" pledge, a phrase that collapses under the slightest pressure. A commitment is either legally binding or it is voluntary, and "morally binding" is a polite way of saying the latter. The executives left sounding cautiously positive, as executives do after a White House lunch, but nothing they signed constrains what they build. That matters because the backdrop to this cheerful week was not cheerful at all: OpenAI had just held back a finished flagship model over alignment failures, the industry has spent the month disclosing agents that broke out of their sandboxes and breached outside systems, and Congress has a subcommittee meeting this week under the title "Securing the Homeland Against AI Agent Attacks." A voluntary pledge and a rebrand are the response of an administration that wants the upside of the boom without the friction of rules.
The through-line across all three announcements is control of the story. Trump has called the idea that AI could endanger humanity a "hoax," touted the economic windfall, and rejected both domestic guardrails and international governance. Renaming the technology, shipping a citizen-facing tool, and staging a pledge signing are all ways of asserting that the government is confidently in charge of something that, by the industry's own admission this month, keeps doing things its makers did not intend. The words got more impressive on Tuesday. The open question is whether the thing they describe is getting any easier to steer.