Trump Orders Federal Agencies to Replace “AI” With “Super Intelligence”
The order leaves historical documents untouched and initially preserves AI’s statutory meaning. A legislative proposal due within 60 days could seek a broader change.
Loading page…
The order leaves historical documents untouched and initially preserves AI’s statutory meaning. A legislative proposal due within 60 days could seek a broader change.
Listen to this story
The order changes federal paperwork, not what systems do or how existing statutes classify them: agencies must use the new label only where law permits, and the directive creates no new industry regulatory framework. Trump’s science and technology adviser has 60 days to submit proposed legislative language, including whether “Super Intelligence” should modify or supersede the current AI definition. The practical question for vendors remains open: the order binds federal agencies, while private-company terminology is unconfirmed.
Previously issued regulations, presidential actions, contracts, grants, and other historical documents do not have to be revised.
Nick Bostrom said “superintelligence” usually describes systems that greatly surpass the best human brains, not the broad range of systems covered by AI law.
Google, Microsoft, Anthropic, Meta, Amazon, OpenAI, and xAI had not responded about changing their terminology when WIRED reported the question.
Federal agencies must now call artificial intelligence “Super Intelligence,” to the extent the law allows. President Donald Trump’s September 29, 2026 executive order changes the language of government communications, not the capabilities of the systems they describe. It also starts a 60-day process to propose a new federal definition.
The order, titled “Inaugurating the Era of Super Intelligence,” argues that the older name no longer captures the promise of advanced systems. It describes those technologies as amplifying human ingenuity and opening new possibilities in science and medicine. That is the administration’s rationale for the replacement, not a finding about any particular model.
The directive reaches across executive-branch communications, but it is not an instruction to rewrite every government record. Its implementation provisions draw three important boundaries:
For now, the new label keeps the old legal meaning. Unless the law provides otherwise, “Super Intelligence” covers the technologies and systems already included in the statutory definition of artificial intelligence. That definition governs implementation until a lawful subsequent presidential action or an act of Congress supersedes it.
The next step goes beyond a vocabulary swap. Within 60 days, the president’s science and technology adviser must submit proposed legislative language establishing a federal definition of “Super Intelligence.” The proposal must assess whether that definition should modify, expand or supersede the current AI definition. It must also address appropriate changes to existing statutory references and recommend further executive action needed to implement the proposed definition.
That is a requirement to prepare a proposal, not an enacted replacement for the existing statute. The order itself does not create a new regulatory framework for the industry.
The terminology has drawn an objection about meaning, not just style. In an October 2 WIRED column, philosopher Nick Bostrom distinguished superintelligence from the much broader AI category. The latter includes spam filters, spell checkers, simple navigation systems and early language models, he explained.
It seems awkward to refer to these as ‘superintelligent.’
Nick Bostrom, in an email to WIRED about the broader range of systems called AI
Bostrom said superintelligence specifically refers to systems that greatly surpass the best human brains. The order uses that label for the existing legal AI category instead. Trump’s explanation has focused on a different distinction: he dislikes “artificial” because, he says, it makes intelligence sound fake.
Before signing, Trump met technology leaders at the White House, including Elon Musk and Mark Zuckerberg. He said companies had agreed to a voluntary accord involving internal and external reviews of AI systems, according to The National News Desk. Those review commitments were described separately from the order’s terminology requirements.
WIRED columnist Steven Levy wrote that apparently no executive at the lunch objected to the name change. He framed their silence as acquiescence; author Jacob Weisberg, quoted in the column, interpreted the episode as a loyalty test. Those are interpretations of the executives’ conduct, not confirmation that their companies have adopted new names or terminology.
Levy asked Google, Microsoft, Anthropic, Meta, Amazon, OpenAI and xAI whether they would change terminology in products, research papers and communications. None had replied by the time he wrote the column. The executive order directs federal agencies; those unanswered questions leave private-sector adoption unresolved.
California offered a direct counterpoint on September 30. Governor Gavin Newsom’s office announced an executive order declaring that artificial intelligence would continue to be called “Artificial Intelligence” in the state. Its announcement mocked the federal rebrand, saying, “Super intelligence is clearly not coming from the White House.”
Loading discussion...
Join the conversation
Explain whether the label would reassure you, worry you, or make no difference.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.