Anthropic's IPO filing now calls AI "super intelligence", some of the time

Anthropic's IPO prospectus has learned a new word. Eight days after its first amendment, the company's second amended filing introduces "super intelligence", the term the White House now uses for AI. It uses the new name in some places and keeps the old one in most.
What changed
On September 29, President Trump signed Executive Order 14434, "Inaugurating the Era of Super Intelligence". It tells federal departments and agencies to use "Super Intelligence" and "SI" in place of "Artificial Intelligence" and "AI" in official correspondence, websites, reports and other non-statutory documents. It does not require anyone outside the executive branch to follow suit, and it defines the new term by pointing to the existing legal definition of artificial intelligence in the US Code.
Anthropic's Amendment No. 2 to its Form S-1, filed with the Securities and Exchange Commission on Tuesday, November 10, adopts it anyway. The list of defined terms now has an entry for "super intelligence" or "SI", worded exactly like the existing entry for "artificial intelligence" or "AI". It then adds: "For purposes of this prospectus, to avoid confusion, we use the terms 'super intelligence,' 'SI,' 'artificial intelligence,' and 'AI' interchangeably," and cites the executive order by number.
None of this was in Amendment No. 1, filed on November 2. Yonhap News counted the terms and found "SI" 42 times and "AI" about 600 times in the new filing. Our own count of the SEC text matches it: 42 for "SI", 27 for "super intelligence", and 579 for "AI", down from about 780 in the earlier amendment. Some of that fall is the switch; some is ordinary redrafting of a 250,000-word document.

Where the new word landed
The swap is patchy, which is what makes it entertaining. The opening of the business overview used to say, "We believe this abundance of AI will rapidly transform every sector of the global economy." It now reads "this abundance of super intelligence". The company that "was founded in 2021 by a team of researchers responsible for many of the breakthroughs that define AI" now defines SI instead.
Elsewhere, the old term survives. The filing still mentions "AI models" 79 times against 14 for "SI models", and the risk factors, financial statements and governance sections are still mostly AI. The prospectus also carries two definitions of the same thing, one under each name, which is perhaps not the clearest way "to avoid confusion".
Why a company would bother
Anthropic is not obliged to use federal terminology. Yonhap describes the company as at odds with the US government, and Anthropic has historically argued for stricter AI regulation than the administration favours. Yonhap called the adoption unusual for exactly that reason.
But a prospectus is a selling document, and the buyers of this one include federal customers, regulators and an administration that has opinions about vocabulary. Others have already moved. According to Yonhap, Nvidia's Jensen Huang, Microsoft's Satya Nadella and Intel's Lip-Bu Tan have all used "SI", and Elon Musk has said SpaceX's AI division will be renamed SpaceXSI, The Korea Times reported. South Korea's science ministry, by contrast, said in October it had no plans to follow.
There is also a technical wrinkle. Researchers have long used "superintelligence" to mean AI that exceeds human ability in every domain, the kind of system AI-safety researchers worry about most. Calling today's chatbots "super intelligence" in an SEC filing blurs a line the company has spent years drawing.
Our take
This is a search-and-replace, and not a thorough one. A company that prides itself on precise language has put two identical definitions under two names in a legal document and then used both, in roughly a one-to-fourteen ratio.
That does not make it trivial. Securities filings are read by lawyers looking for inconsistencies, and "super intelligence" is a loaded term in AI safety. If the product is officially super intelligence, a plaintiff can ask what the risk factors about advanced AI were really describing. Anthropic has hedged by saying the terms are interchangeable, which is the right legal answer and a slightly awkward commercial one.
The bigger signal is political. A company asking public investors for up to $64bn, with Washington as both customer and regulator, decided a week before pricing that matching the president's vocabulary was worth the edit. Expect other AI filers to make the same call. Expect them to do it more carefully.
Sources
- Anthropic, PBC, Amendment No. 2 to Form S-1 registration statement (November 10, 2026) and Amendment No. 1 (November 2, 2026), US Securities and Exchange Commission
- The White House, Executive Order 14434, Inaugurating the Era of Super Intelligence, Federal Register, 91 FR 63129 (signed September 29, 2026)
- Yonhap News, Anthropic also adopts Trump's "superintelligence" terminology (November 11, 2026)
- TechCrunch, Anthropic sets IPO price range of $70 to $78 per share (November 10, 2026)
- The Korea Times, Korea has no plans to follow US push to replace term AI with "super intelligence" (October 8, 2026)
- Word counts: The Graze, from the SEC text of both amendments (whole-word matches, case-sensitive)







