The Week the White House Renamed AI

On September 22, 2026, at the lectern of the United Nations General Assembly, President Trump told the room what he thought was wrong with the name of the technology everyone had been talking about. “The use of the word artificial makes intelligence fake, it makes it sound fake,” he said. “And it is not fake.”

From then on, he said, American documents would say “super” instead of “artificial,” and he hoped the world’s would follow. Then he added a line that read like a bet: “Let’s see if I have any power. Maybe I do and maybe I don’t. We’re going to find out pretty soon.”

One week later

Seven days later he signed Executive Order 14434, “Inaugurating the Era of Super Intelligence.” It tells federal agencies to use “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI” in their correspondence, websites, reports and policy documents. Its policy statement goes a step further: the executive branch “will not acknowledge the usage of ‘Artificial Intelligence’ and ‘AI’ in any applicable setting,” to the maximum extent permitted by law.

The same day, after a lunch at the White House, he stood outside with the leaders of six of the biggest AI companies: OpenAI, Anthropic, Google, Nvidia, Meta and SpaceXAI, Musk’s AI company. They had signed a short voluntary pledge, the White House Accord on Super Intelligence.

It promises four layers of oversight for their most advanced AI models: internal controls, an internal team that checks them, an independent outside auditor, and a committee of the company’s board. Trump called the pledge “morally binding,” The San Francisco Standard reported.

The third layer is an outside auditor, and each company is to “partner with” one. The accord doesn’t say how the auditor is chosen or whether the company can overrule its findings, and that is the question The Silent Exit Clause asks of any safety check.

Nothing in the text makes any of it mandatory. It says only that “over time, it may make sense to codify these steps into laws or regulations.”

The slip

Elon Musk, who signed for his AI company, was partway through his remarks when he used the old word. “SI, pardon me,” he said, laughing, according to The Standard’s account. Trump thanked him and patted his elbow.

How the room sorted itself

The order directs federal agencies, not companies, and The Independent reported that it has no legal effect on the private sector. So the leaders in the room were free to choose, and their choices differed.

Jensen Huang embraced it. He posted on X that superintelligence “offers an extraordinary opportunity to advance discovery, productivity, security, health, and prosperity,” The Standard reported, and the White House listed a post of his, along with two from Musk and others, including David Sacks, Trump’s former AI and crypto czar, as use of the phrase, The Independent reported.

Sundar Pichai was more careful. On X he referred to the accord by its title, which includes the new word, and said Google is focused on releasing models only after a thorough review.

At the press conference, The Standard reported, OpenAI’s Greg Brockman and Anthropic’s Dario Amodei both said “AI,” and Amodei was the only leader there who insisted that “the technology has real risks.”

Meta hadn’t needed an order. It named its AI unit Meta Superintelligence Labs in June 2025, and in a July 2025 memo Mark Zuckerberg wrote, “Developing superintelligence is now in sight.”

Who followed, and why

Tech insiders and lobbyists told Politico, some of them privately, that the rename is unlikely to catch on, The Independent reported. One lobbyist described the pattern in a sentence: “The government can change its vocabulary overnight and the private sector follows where it’s working with the government, but broader usage takes longer.”

Adam Kovacevich, who runs the Chamber of Progress, a progressive tech industry coalition, offered a sharper reading: “It is a way for the CEOs to signal fealty to the president.”

The Standard made a version of the same point more evenly. The leaders of the biggest labs may think the word is inaccurate, it wrote, but they have a vested interest in pleasing a White House that can regulate the industry and dole out contracts.

It doesn’t explain everyone. The leaders of Google, OpenAI and Anthropic all deal with the government too, and The Standard said none of them used “SI” at the press conference.

Then California answered.

On September 30, 2026, Governor Gavin Newsom signed Executive Order N-10-26, which directs the state agencies and departments under his authority to keep saying “Artificial Intelligence” and “AI,” “notwithstanding any rebranded or different terminology used by the federal government, unless inconsistent with the law.” Its preamble argues that a new name can’t distract from what it calls a failure to take action on well-documented security and safety risks.

Two governments, two vocabularies, one week.

Musk’s answer

On Sunday, October 4, 2026, Musk posted on X, in order, “No more AI,” then “SI,” then “It’s better,” then “SpaceX is a super intelligence company,” Teslarati reported. A user asked whether his AI unit, SpaceXAI, would become SpaceXSI. “Yes, we will make that change,” he replied, according to Reuters.

He gave no date, and as of Sunday the unit’s account on X still carried the old name. It would be the unit’s third name this year: SpaceX bought xAI in February 2026, the unit became SpaceXAI on July 6, 2026, and SpaceXSI would follow.

Five days earlier, at the White House, he’d caught himself saying “AI.” Now he was announcing there’d be no more of it.

So twelve days after the UN speech, Trump’s question had a partial answer. Inside the federal government, the order told agencies to change the word, and the White House counted posts from Musk, Nvidia’s Huang and others as taking it up.

The leaders of Google, OpenAI and Anthropic hadn’t used it at the press conference, The Standard said, and California had ordered the opposite.

Three jobs for one word

The order writes “Super Intelligence” as two words. The tech industry typically writes “superintelligence” as one, The Standard noted, and the one-word version has a history. The Standard calls it “the white whale that all the top labs have been chasing for more than a decade.”

Put the order beside two other uses of the word, and you can see three different jobs.

In the order, it’s mainly a label. For its own purposes, the order says the new terms mean “the technologies and systems encompassed by the term ‘artificial intelligence’” as federal law already defines it, which makes “Super Intelligence” a new name for the systems on the market today. But the order’s opening section leans further, saying that as these capabilities improve, they “increasingly represent not merely artificial intelligence, but a new era of Super Intelligence,” so even in the order the label comes with a claim about capability attached.

In Zuckerberg’s memo, it’s a goal, something that is “now in sight.”

And in the Senate, Bernie Sanders introduced a bill to ban “artificial superintelligence,” which The Standard says defines it as AI that matches or surpasses human intelligence, or that can “disempower humanity.” There, it’s a danger. Anthropic’s IPO Filing Warns About Its Own Models covers how one lab describes the risks of its own systems.

Researchers noticed the collision. David Krueger, an AI professor and founder of the nonprofit Evitable, told The Standard that superintelligence is “already a word, and it means something different than AI,” and that it means specifically AI that is “vastly superhuman and so very difficult, if not impossible, to control.” Adam Thierer of the R Street Institute told Politico that the term “means different things to different people, but it already had a general meaning in the literature and debates about advanced AI.”

Pedro Domingos, a computer science professor at the University of Washington, told DW that “AI is doing different things from what humans do, so calling it superintelligence is misleading.” He also said the US labs “have every interest in calling it super, hyper, superior and whatnot intelligence, because it’s playing up their capabilities.”

What it means, and what’s still open

For a company that deals with federal agencies, the order tells those agencies to write “SI” in their own correspondence and reports, so it may see the new term there. For everyone else the order has no legal effect, as The Independent reported, and the word a company uses is its own choice.

One piece of the story is still unwritten. Within 60 days of the order, the president’s science adviser has to send him proposed legislative language for a federal definition of “Super Intelligence,” including whether it should replace the existing one. It’s due around the end of November 2026.

Until then, expect the word to keep turning up with a different job each time, and sometimes with two at once, so a sentence that uses it is worth reading twice, once for what it says and once for which job, or jobs, it’s doing.