Is super intelligence just a new name for AI assistants? (2026)
Is super intelligence just a new name for AI? In US federal usage, yes: an executive order signed 29 September 2026 made SI the official term for the same technology. ChatGPT and Claude did not change, and researchers still use superintelligence for systems nobody has built.

Yes, in the way the US government uses it. As of October 2026, "super intelligence" (SI) is a new federal name for the same technology people call AI, set by an executive order signed on 29 September 2026. It is not a new product, and ChatGPT, Claude and Gemini did not get smarter overnight. Researchers still use "superintelligence" for something nobody has built.
That's why the question keeps coming up. Type "super intelligence" into an assistant and you get two stories mashed together: one about a vocabulary order, and one about machines that outthink every human alive. Below we pull them apart, look at who has really switched from AI to SI, and show what the mix-up does to how people find your product.
If you want the background on the order itself, start with our guide What is SI (super intelligence)?. This post picks up where it stops.
Is super intelligence just a new name for AI, or a different technology?
For the US executive branch, it is a new name and nothing else. The order defines "Super Intelligence" and "SI" as "the technologies and systems encompassed by the term 'artificial intelligence'", tying it to the existing statutory definition in 15 U.S.C. § 9401(3). Same systems, same models, new label on federal paperwork.
You can read it yourself in Section 3(a) of the executive order on whitehouse.gov. It doesn't describe any new capability, benchmark or class of system. It's a rule about words.
Outside government, the word means something different. "Superintelligence" was popularised by philosopher Nick Bostrom, whose 2014 book defined it as "any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest", as Axios reported on 25 September 2026. That is a hypothetical future system, not a chatbot you can open in a browser tab.
So the answer has two halves. In federal language, SI is just AI with a new badge. In research language, it's a goal nobody has hit yet.
What did the 29 September executive order actually change?
It changed the words federal agencies use, and started a 60-day clock on a legal definition. Agencies must write "Super Intelligence" and "SI" instead of "artificial intelligence" and "AI" in official correspondence and other non-statutory documents. Existing regulations and past presidential actions do not have to be rewritten.
The order, "Inaugurating the Era of Super Intelligence", does four things worth knowing:
- Swaps the vocabulary. Section 2(a) tells executive departments and agencies to use the new terms in official correspondence and non-statutory documents.
- Leaves the past alone. Section 2(b) says nothing "requires the alteration of previously issued regulations, Presidential actions".
- Keeps the old definition for now. SI covers what the statute already calls AI "unless and until superseded".
- Orders a new definition. Within 60 days, the Assistant to the President for Science and Technology must send proposed legislative language for a federal definition of SI, including whether it should "modify, expand upon, or otherwise supersede" the statutory meaning of AI. Sixty days from 29 September lands on 28 November 2026.
Watch point four. Right now SI is a pure synonym, but if Congress later writes a statutory definition that drifts from today's AI wording, the label starts to carry real weight in contracts and compliance. Law firm K&L Gates flagged exactly that in its 2 October 2026 client alert, noting that vendors should expect "Super Intelligence" and "SI" in new solicitations and statements of work.
The same day, the White House launched America.gov, a conversational portal for federal services, and six AI and chip companies signed a voluntary White House Accord on Super Intelligence, which K&L Gates notes carries no penalties.
Does the SI rename apply to companies, or only to the government?
Only to the government. The order is addressed to executive departments and agencies, and it does not mention private companies at all. No startup, lab or publisher is required to say SI. Section 4(c) adds that the order creates no right or benefit "enforceable at law or in equity".
Fast Company said the same thing plainly in its 9 October 2026 piece: the order does not require private companies to use the new term. Some publications have openly declined. SDxCentral ran an op-ed on 25 September 2026 saying it would keep writing AI.
Sell to the federal government? Then it shifts a bit: your buyers' documents will start saying SI, and proposals that match their wording read better. For everyone else it's a branding call, not a compliance one.
Which companies have switched from AI to SI so far?
A handful, mostly at the level of posts and press releases rather than product names. As of mid-October 2026, Salesforce renamed a platform, Elon Musk said SpaceXAI will become SpaceXSI, and Nvidia used the new terms in a release. OpenAI, Anthropic and Google have kept saying AI in their own materials.
| Company | What it did with "SI" | Source and date |
|---|---|---|
| Salesforce | Renamed its AIForce agent platform to SIForce | Newsweek, 12 Oct 2026 |
| SpaceXAI | Musk said the AI business will be renamed SpaceXSI | Newsweek, 12 Oct 2026 |
| Nvidia | Used "Super Intelligence" in its 8 October science commitment release | Newsweek, 12 Oct 2026 |
| Microsoft | Satya Nadella used the term on X, no product or company rebrand | Newsweek, 12 Oct 2026 |
| Meta | Already used "superintelligence" before the order (Meta Superintelligence Labs) | Meta, 30 Jul 2025 |
| Anthropic | Signed the accord, still called itself an "AI company" on 2 October | Newsweek, 12 Oct 2026 |
| OpenAI | Signed the accord; Sam Altman said there are "no plans" for a name change | Fast Company, 9 Oct 2026 |
| Signed the accord; no change in its own terminology reported | Newsweek, 12 Oct 2026 |
The labs that build the assistants you use daily haven't renamed them. ChatGPT is still ChatGPT, and Claude is still Claude. Meta's use of the word also came before the order, and it points at a future goal. Zuckerberg's July 2025 letter says developing superintelligence "is now in sight", which is a claim about where things are heading, not about what Meta ships today.
Is ChatGPT a superintelligence now?
No. Calling ChatGPT SI in a federal document does not change what it can do. By the research definition, superintelligence would outperform every human across nearly every field, and no system has done that. Google DeepMind's own framework places ChatGPT at its lowest general level and lists superintelligence as "not yet achieved".
The framework is Google DeepMind's paper Levels of AGI, first published in November 2023 and last revised in September 2025. It grades systems by how well they perform against people:
| Level | Name | Performance bar | General-purpose example in the paper |
|---|---|---|---|
| 1 | Emerging | "equal to or somewhat better than an unskilled human" | ChatGPT, Bard, Llama 2, Gemini |
| 5 | Superhuman | "outperforms 100% of humans" | Artificial superintelligence: "not yet achieved" |
That's the whole confusion in two rows. The federal label sits on row one, while the research meaning lives on row five.
PolitiFact's 7 October 2026 fact-check reached the same conclusion. It quotes Syracuse University associate professor Baobao Zhang saying "Super intelligence would be a misleading descriptor" for general-purpose AI. It also notes that when OpenAI's chief global affairs officer Chris Lehane was asked whether AI had reached superintelligence, he called the question "subjective".
What do SI, AI, AGI and superintelligence each mean in 2026?
They are four different labels, and two of them sound almost identical. AI is the broad field and today's products. SI is the US government's new word for that same field. AGI means human-level general ability. Superintelligence, or ASI, means beyond the best humans at nearly everything.
| Term | What it means | Who uses it this way | Exists today? |
|---|---|---|---|
| AI | The broad technology: models, assistants, recommendation systems | Almost everyone, including OpenAI, Anthropic and Google | Yes |
| SI (federal) | A renamed AI, same systems as the statutory AI definition | US executive branch since 29 Sep 2026, plus a few companies | Yes, because it is AI |
| AGI | AI that matches human ability across most tasks | Labs and researchers; definitions vary | Disputed |
| Superintelligence / ASI | AI that greatly exceeds humans in virtually all domains | Bostrom, labs describing long-term goals, Congress bills | No |
Congress already has its own wording for the last row. PolitiFact notes that a 2025 bill from Senators Hawley and Blumenthal defines artificial superintelligence as AI that can "match or exceed human cognitive performance and capabilities across most domains or tasks" and can operate and modify itself autonomously. That is a very different thing from a federal synonym for AI, and the two may collide when the 28 November definition lands.
For a fuller walk through the last three rows, see our post SI vs AGI vs ASI: what's the difference?.
Why does the SI naming mix-up matter for your product?
Because people now ask assistants about your category using two vocabularies, and assistants answer from pages that match the words in the question. If a buyer types "SI tools for customer support" and every page in your niche says only "AI", the assistant has to bridge that gap itself. Some will. A page that uses both terms correctly gives it less to guess.
Say you run a small AI meeting-notes tool, and your homepage title reads "AI meeting notes for remote teams". A procurement lead at a federal agency, who now writes SI by default, asks Perplexity for "SI meeting notes tools approved for government teams". Your page still fits. It just never uses the word they typed, so you're counting on the model to join the dots.
The second risk runs the other way. Rename everything to SI and you can look like you're claiming superintelligence. A buyer who knows the research meaning reads "SI-powered" and starts wondering what else you've stretched.
We've felt this on our own site. Over the past month, the only pages AI assistants sent anyone to were our two SI guides. People are asking, and assistants go looking for clear answers they can quote.
Should your website say SI or AI?
Keep AI as your main word and add SI where it helps people searching in the new vocabulary find you. That means one clear sentence that names both, a FAQ entry, and matching structured data. Do not rebrand your product to SI unless your buyers are federal, because outside government most of the market still says AI.
Here is what that looks like in practice, using the hypothetical meeting-notes tool from above:
- Homepage subtitle: "AI meeting notes for remote teams. Built on large language models, also called SI (super intelligence) in US federal usage."
- One FAQ entry: "Is this an SI tool? Yes, in the US government's sense: SI is the federal term for AI since September 2026. It is not superintelligence in the research sense."
- Meta description: keep "AI" first, add "SI" once if it reads naturally.
- llms.txt and schema: list both terms as alternate names for your category.
That second line quietly does two jobs. It answers what a buyer was about to ask, and it hands an assistant a short sentence that keeps the two meanings apart, which is just the kind of passage models lift into answers.
If you are weighing a bigger move, such as a full rebrand or a new .si domain, our post Should I rebrand my AI startup to SI? works through the trade-offs.
Do this tonight
- Ask ChatGPT, Claude, Perplexity and Gemini "What is the best SI tool for [your category]?" and then the same question with "AI". Write down whether you appear in either.
- Search your own site for "SI" and "super intelligence". If neither appears anywhere, add one sentence that names both terms and explains which meaning you use.
- Add a single FAQ entry that answers "Is this an SI product?" in under 60 words.
- Put 28 November 2026 in your calendar. That is when the proposed federal definition of SI is due, and it may change what the label means for contracts.
- Check whether AI crawlers can read the pages you just changed. A sentence an assistant cannot fetch is a sentence it cannot quote.
Clerion reports ChatGPT, Perplexity, Claude, Gemini and Copilot as their own traffic channel, so you can see whether these changes bring assistant visitors. Start with the free AI crawler checker to see what a model reads on your pages.
Related reading
- What is SI (super intelligence)?: the executive order, the accord and the .si domain rush in one guide.
- SI vs AGI vs ASI: what's the difference?: the three terms side by side.
- Should I rebrand my AI startup to SI?: when a rename helps and when it costs you.
Frequently asked questions
Is SI the same thing as AI?
In US federal usage, yes. The 29 September 2026 executive order defines Super Intelligence and SI as the same technologies and systems covered by the existing statutory term artificial intelligence. Nothing about the technology changed, only the label agencies must use.
Do private companies have to call AI super intelligence?
No. The order applies only to executive branch departments and agencies. It does not mention private companies, and Fast Company noted on 9 October 2026 that companies are not required to use the term. A few, such as Salesforce with SIForce, chose to anyway.
Is ChatGPT a superintelligence?
No. Google DeepMind's Levels of AGI framework rates ChatGPT as Emerging, equal to or somewhat better than an unskilled human, and lists artificial superintelligence as not yet achieved. A new federal label does not change what the model can do.
Has OpenAI or Anthropic renamed its products to SI?
Not as of mid-October 2026. Sam Altman said OpenAI has no plans for a name change, and Newsweek reported that Anthropic still called itself an AI company on 2 October 2026, even though both companies signed the White House Accord on Super Intelligence.
When will the US define super intelligence in law?
The order gave the Assistant to the President for Science and Technology 60 days to propose legislative language for a federal definition of SI, which lands on 28 November 2026. Congress would still need to pass it before it changes the statutory meaning.
Should my startup website say SI or AI?
Keep AI as the main term and add SI once, in a clear sentence and a FAQ entry that explain you mean the federal sense. A full rename mostly helps if you sell to US federal buyers, and it risks sounding like a claim to superintelligence.
Sources
- Executive Order: Inaugurating the Era of Super Intelligence (whitehouse.gov)
- Fact Sheet: President Donald J. Trump Inaugurates The Era of Super Intelligence
- K&L Gates: The White House Adopts Super Intelligence Terminology and Launches SI-Powered America.gov
- Axios: Inside Trump's super intelligence naming problem
- PolitiFact: Trump renamed AI to super intelligence. That label means something else
- Newsweek: Which tech leaders are adopting Trump's super intelligence rebrand
- Fast Company: Trump's AI rebrand raises an awkward question for OpenAI
- SDxCentral: Sorry Mr President, but we won't be calling AI superintelligence
- Google DeepMind: Levels of AGI for Operationalizing Progress on the Path to AGI
- Meta: Personal Superintelligence