Trump’s ‘super intelligence’ order changes AI’s name, not its definition

Donald Trump’s new promise to ban the term “artificial intelligence” follows a federal order already changing government language. That order, however, gives “super intelligence” the same legal meaning as AI: it is a rebranding rule, not evidence that machines have attained a new level of ability.

In a Truth Social post on 27 October 2026, Trump called AI “a very bad and inaccurate name” and promised to ban the term in “all relevant documents, and everything else”. He pressed The Wall Street Journal to use “SI”, describing it as “the name which virtually everyone else is using, or soon will be”.

The post did not explain what would be added to the renaming order he signed on 29 September.

What the September order covers

Executive Order 14434 directs federal executive departments and agencies to use “Super Intelligence” and “SI” instead of “Artificial Intelligence” and “AI” in official correspondence, public communications, websites, reports and policy documents, to the extent permitted by law.

It does not require previously issued regulations, presidential actions, contracts, grants or other historical documents to be rewritten. Nor does it impose a naming rule on private newspapers such as the Journal.

There is already a visible effect. The National Institute of Standards and Technology, or NIST, has a topic page headed “Super intelligence”, uses SI in programme descriptions and explicitly says it is updating its communications as directed by the order. Some AI references remain on the same page.

Trump’s campaign over the terminology also predates his latest pledge. In a separate post on 8 October, he said the White House considered anyone using “Artificial Intelligence” rather than “Super Intelligence” to be “THE ENEMY!”

A name is not a capability test

The September order argues that the new wording better captures the technology’s “promise, potential, and rapidly advancing capabilities”. Yet its definition of SI points directly to the existing federal AI statute.

That statutory definition describes machine-based systems capable of making predictions, recommendations or decisions for human-defined objectives. It does not require them to outperform humans. The order also calls for proposed legislative wording within 60 days, including possible changes to that definition; the order itself does not rewrite the statute.

The OECD’s guidance published in February 2026 similarly defines AI broadly, covering systems that generate predictions, content, recommendations or decisions. It recognises differing levels of autonomy and adaptability.

A proposed framework by Google DeepMind researchers, by contrast, places artificial superintelligence, or ASI, at its highest combined level of performance and breadth: a system able to perform a wide range of non-physical tasks at a level no human can match. That is a capability threshold within the framework, not a blanket synonym for AI.

The new wording has reached the private sector, but adoption need not mean replacement. Microsoft opens an Azure Maia infrastructure announcement with “As we move into the era of Super Intelligence”, then describes the programme as part of its “AI infrastructure strategy” and repeatedly uses AI elsewhere in the same announcement.

These examples show genuine uptake, but do not substantiate near-universal adoption. NIST is following a government instruction; Microsoft is using both terms. Trump’s accompanying suggestion that others will soon follow looks ahead, rather than establishing present usage.

For federal agencies, the instruction is to rename the same statutory category of systems, not to certify those systems as superintelligent. What Trump’s latest pledge adds to that rule — particularly for private publishers — remains unclear.