Donald Trump has extended to artificial intelligence (AI) his habit of renaming geographical features — the Gulf of America instead of the Gulf of Mexico or Lake America instead of Lake Ontario. In a display of originality, on this occasion he has not reused the initials by calling it American intelligence, but rather he is pushing to have this technology identified as superintelligence (SI). As with the Gulf and the Lake, he has embodied his will by signing an executive order that marks the beginning of the SI era.
A name change that goes back a long way
The president’s actions have caused a certain surprise, but the name change was not an improvised idea. In July 2025, during his remarks on the launch of the AI Action Plan, he already expressed his displeasure with the word “artificial.” In December of that year, at the ceremony signing the executive order intended to limit states’ ability to regulate AI, he stated that he had always thought it should be called “SI, supreme intelligence“, although he added that the decision lay with the industry. Nine months later there would be no room for that deference.
The final stretch of the operation began with a poll on the social network formerly known as Twitter. On X, with 233,673 votes, participants chose as the most appropriate term for the technology intelligence superior (41.5%), compared to the other two options of supreme intelligence (30.7%) or extreme intelligence (27.8%). New votes followed, until on September 21 Trump faced “superior” against a new option, “super”, which won by a wide margin.
The most striking thing is that Trump has been politically correct in introducing and ultimately backing the designation superintelligence rather than intelligent superior, which could seem like a slight to human intelligence. It seems, in any case, that the acronyms were decided before the word.
“The most striking thing is that Trump has been politically correct in introducing and ultimately backing the designation superintelligence rather than intelligent superior”
As with the renaming changes of the Gulf of Mexico and Lake Ontario, Donald Trump aims for global recognition of the new term for AI. It is likely that, since it is not a physical entity as in the two previous cases, the president trusted greater success in his objective. But if with the Gulf he managed to get Google or Apple to adopt the name within the United States, outside his sphere of power, reality again contradicts his wishes.
The glass is almost empty for the acceptance of the new term
First, there is a potential clash with a significant portion of the tech community. The term “superintelligence” already had a consolidated technical usage. The idea of a machine capable of surpassing us and of self-improving traces back to I. J. Good —a collaborator of Alan Turing—, who in the sixties spoke of an “ultraintelligent” machine. The term was solidified by Nick Bostrom, first in a 1998 article and later in his 2014 book. With that name, academia defined a system that outperforms humans in virtually all domains and that could design itself to be even better, leaving human intelligence outpaced.
And the industry itself had turned it into its great commercial aspiration: Meta named in 2025 its division as Meta Superintelligence Labs, Ilya Sutskever founded Safe Superintelligence and Microsoft speaks of a “humane superintelligence”, always as a stage above current AI. By contrast, the legal definition of AI —contained in the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. § 9401(3))—, which the executive order assigns verbatim to SI, encompasses much more modest systems in capability, from a frontier model to a spam filter, an online store recommendation system or the pricing algorithm for rentals.
Even the acceptance of the SI term by the tech oligarchs is nuanced. It is true that a portion of them signed the so-called “White House Agreement on Superintelligence”, but in its just over three hundred words, outside the title, not a single instance appears of the new name Trump wants to impose on AI: the text speaks of “frontier models”.
Nor has there been any meaningful step toward industry adoption of the term, aside from statements on the signing day, an apparent endorsement by Jeff Bezos on Fox News and a flattering message from Elon Musk on his social network in which he announced that SpaceXAI would be renamed SpaceXSI, a change that so far has not been reflected in corporate filings.
Also in internal terms, the substitution of the term is set to become a central issue in the US political debate. Within Congress, in the proposed laws to create a risk assessment framework for AI or to pause the development of superintelligence, the name superintelligence is reserved for an AI that allows devices or software to “match or surpass human performance and cognitive capabilities in most domains or tasks”.
In other words, again the name proposed by Trump is used for a technology with greater capabilities than the systems we currently know as AI. If everything becomes superintelligence, the attempts to curb only the most advanced systems become semantically diluted: appropriating the language of the risk debate is also a way to disable it.
Furthermore, it is not having an immediate positive reception on the international stage. It has not emerged that any other leader endorsed in his UN General Assembly speech the name-change proposal announced by Trump. Likewise, China has not joined. Trump boasted after his meeting with Xi that the Chinese leader accepted the term, but the Foreign Ministry merely stated that it “respects” the American formulation. In the subsequent joint statement, Beijing shuns the term — even putting “artificial intelligence” in the mouth of Trump himself — while the White House includes it.
“If everything becomes superintelligence, the initiatives to curb only the most advanced systems are semantically diluted”
Moreover, from the very day of the UN speech, the Trump Administration has ordered diplomats to use the new term. The U.S. delegation will thus be the only one to say SI in forums where everyone else says AI.
The first test will come with the G20 summit chaired by the United States this year: it will be interesting to see whether the term slips into its final declaration or into the documents of the G7 and OECD, and what formula of equivalence and interoperability between the two terms is applied.
What is at stake with the name
The ultimate objective of Donald Trump with the AI renaming proposal is uncertain. The executive order that opens the era of superintelligence is, on the surface, contradictory. While it orders the use of the new term instead of the one used until now “to the maximum extent permitted by law”, it requests, within sixty days, a definition for the term and an evaluation of the extent to which it expands or replaces the term AI. But the contradiction is only apparent: meanwhile, the order defines SI exactly as the 2020 law defines AI. It changes the label, not the thing, and postpones any substantive change to a future law. This ambiguity may also serve to provide legal certainty to contracts and obligations currently in force in the U.S. administration.
It is not negligible for analysis that Trump’s renaming proposal aligns with the feverish debate over AI regulation. Each time the president has spoken about the name change, he has done so alongside a deregulation milestone: the Action Plan in July 2025, the drive against state laws in December, the rejection of any international oversight at the UN, and a voluntary industry agreement on the same day as the order.
“The renaming proposal aligns with the feverish debate over AI regulation. Each time Trump has spoken about the name change, he has done so alongside a deregulation milestone”
Trump’s known fondness for blocking any rule that constrains the work of frontier-model developers remains. Nevertheless, it is unlikely that the name change will place U.S. frontier-model tech firms outside the existing AI laws —both at subnational levels in the US and in other areas such as the EU—. But it opens a debate about it.
One cannot rule out that this is all part of a Bannon-style tactic to flood the terrain —overwhelming the discussion with initiatives so that none receive thorough scrutiny— to gain some advantage in this field.
For now, Trump has already declared enemies on his Truth Social platform to anyone who uses the term AI instead of his proposed designation. The statement points to an attempt at total terminology replacement, but the history of the magnate’s second presidency invites us to await any outcome.
