Vision

A new name for AI

What President Trump said about calling AI super intelligence, what his order of 29 September requires of US agencies, and what it changes for a company.

  • Analysis
  • 12.10.2026
  • 5 min read

On 29 September, President Donald Trump signed an executive order that tells US federal agencies to write Super Intelligence where they used to write artificial intelligence. Since then the news has travelled in a shorter form: AI is now called superintelligence.

Here is what was said and ordered, what the words mean, and what changes for a company.

What was said and ordered

The dislike of the name is older than the order. At an AI summit in Washington on 23 July 2025, the president said of the word artificial: “I can't stand it. I don't even like the name.” He added: “We should change the name.”

On 19 September 2026 he returned to it on Truth Social. Many people, he wrote, think the words artificial intelligence are inaccurate, and he asked his followers to choose between Superior Intelligence, Extreme Intelligence and Supreme Intelligence. Three days later, at the UN General Assembly in New York, he settled on super. The word artificial, he said, makes intelligence “sound fake”, and all US documents, and hopefully the world's, would now say super.

A week later the word became policy. Executive Order 14434, Inaugurating the Era of Super Intelligence, tells agencies, to the maximum extent permitted by law, to use Super Intelligence and SI in place of Artificial Intelligence and AI in correspondence, websites, reports and policy documents. Earlier regulations, contracts, grants and historical documents need not change. On 8 October the president wrote that the White House considers anyone who still uses the term artificial intelligence the enemy.

So the claim holds, within limits. The president did say that AI should be called super intelligence, and his order makes it the language of the US executive branch. It does not bind companies, states or other countries, and it does not change the law: the statutes still say artificial intelligence. The order also writes the term as two words, where researchers and labs write superintelligence as one.

What the words mean

Artificial intelligence is the older name. It goes back to a summer research project at Dartmouth College in 1956, organised by John McCarthy, which Dartmouth calls the birth of the field.

In law the term is plain. US federal law defines artificial intelligence as a machine based system that can, for objectives set by people, make predictions, recommendations or decisions that influence real or virtual environments. The EU AI Act defines an AI system in similar terms. Neither definition claims that the system is smarter than a person.

Superintelligence means something else. The philosopher Nick Bostrom, whose book of 2014 made the word widely known, defines it as any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest. It is the top rung of the ladder we described in our piece on AGI, one step beyond general intelligence.

The order keeps the old meaning under the new name. For its purposes, Super Intelligence covers the same technologies as the legal definition of artificial intelligence. The label points to the top of the ladder, while the definition still covers every system it covered before.

Why the name matters

In law, what counts is the definition, because definitions decide who must do what. The order gives the Assistant to the President for Science and Technology 60 days, until 28 November, to propose a legal definition of Super Intelligence, including whether it should modify, expand or replace the existing definition of artificial intelligence. The law firm Wiley points out that contract terms, reporting duties and grant programmes often refer to the current definition, so a new one could change what they cover. Until then, as an analysis for the IAPP puts it, people in AI governance work with two vocabularies that may refer to the same systems.

In industry, the word was in use long before the order, mostly as a statement of ambition. Meta gathered its AI teams into Meta Superintelligence Labs in June 2025, and a month later Mark Zuckerberg wrote that developing superintelligence is now in sight. A company called Safe Superintelligence describes building safe superintelligence as the most important technical problem of our time. Such names signal to investors and recruits how large the bet is.

Some companies have followed. Salesforce renamed its AIForce platform SIForce. Anthropic, which signed a voluntary White House accord on Super Intelligence on the same day as the order, still describes itself as an AI company, Newsweek reports.

The largest effect of a name is on expectations. Checking the UN speech, PolitiFact noted that people in the industry do not consider superintelligence an accurate description of the AI systems available today. Amazon Web Services still describes artificial superintelligence as a theoretical concept in AI research. A tool sold as super intelligence will be judged against that word, by customers and by boards.

What it changes for a company

For most companies, nothing changes in law. A Swiss company selling into the EU still works with the AI Act's definition of an AI system, and can keep the words it uses.

Three things are still worth doing. If you sell to US federal agencies or hold their grants, expect forms and solicitations to say SI, and ask the agency to confirm that the requirements behind the new word are unchanged, as Wiley advises. In your own contracts and policies, describe systems by what they do rather than by what they are called, and keep one glossary, so that AI and SI are never read as two different things.

And when a supplier puts super intelligence on the box, ask what is inside: which task, measured how, with which error rate, and who answers when it fails. A new label does not change what a model can do on Monday morning. Nor is there a need to rename your own products: the vocabulary may move again once a legal definition is proposed.

Our own line reads: Human led. AI powered. Autonomous next. Whatever the technology is called, the useful questions stay the same: what it does reliably today, and who answers for the result.