Can I call my product super intelligence? (2026)
Can I call my product super intelligence? Yes: no US law bans the term, and Executive Order 14434 binds federal agencies, not startups. But if your copy implies abilities the product lacks, the FTC and SEC can treat the label like any other misleading AI claim.
If you're asking "can I call my product super intelligence?", the answer is yes. As of October 2026 no US law bans the term, and the executive order that made "SI" official binds federal agencies, not startups. But the label is a claim. If your copy suggests abilities the product does not have, the FTC and SEC can treat it like any other misleading AI claim.
So the real question isn't whether you're allowed to use the word. It's whether you can back up what it promises. The government made the term popular in a month. It did not change the rules on what you can say about your product.
You built a summarisation tool on top of a frontier model last spring. The White House now calls that super intelligence. Salesforce renamed its agent platform. Your competitor's landing page says "SI-powered" in the hero. And you're wondering whether to follow, and whether anyone will come after you if you do.
The word is free. The promise inside it is not.
Can I call my product super intelligence without breaking the law?
You can. No statute or regulation reserves the phrase, and Executive Order 14434 creates no duties for private companies. What applies is the old rule: an ad that misleads a reasonable buyer about something that matters to them is deceptive. "Super intelligence" is now one more phrase that rule covers.
The FTC's 1983 Policy Statement on Deception still sets the test. A claim is deceptive if it is likely to mislead, if a consumer acting reasonably would read it that way, and if it is material, meaning it affects whether they buy. The same statement says that "advertising that lacks a reasonable basis is also deceptive." No part of that test cares which word you used. It cares what a buyer believed after reading your page.
That leaves you a lot of room. Say SI, rename a feature, put it on the pricing page. Just don't let the label carry a promise your product would fail the first time someone tested it.
Does the executive order apply to my startup?
No. Executive Order 14434, "Inaugurating the Era of Super Intelligence," was signed on September 29, 2026 and published in the Federal Register on October 2. It tells Executive Branch agencies to say "Super Intelligence" and "SI" instead of "AI" in non-statutory documents. According to Arnold & Porter's analysis, it "imposes no obligations on private parties."
The agencies are following it. A Navy administrative memo issued the week of October 22 moves websites, correspondence and training material to "super intelligence," and the Marine Corps did the same. On October 23 the SEC's fiscal 2027 examination priorities listed "emerging technologies, including super intelligence" as a focus area.
Here's the detail most founders skip. For now the order defines Super Intelligence the same way federal law already defines artificial intelligence, under 15 U.S.C. § 9401(3). In government language, then, SI is a synonym for AI, not a higher tier of capability. That actually helps you. Using SI as a plain synonym for AI matches the official meaning, and we cover that definition in depth in What is the legal definition of super intelligence?
Is "super intelligence" puffery or a claim I have to prove?
Used as a brand word or a category label, it reads a lot like "smart" or "powerful," and regulators generally leave obvious exaggeration alone. Once the copy states or implies a measurable ability, like beating humans, never being wrong, or doing a job without people, it is an objective claim and you need evidence before you publish.
The FTC's deception statement says it generally won't pursue "obviously exaggerated or puffing representations" that ordinary consumers do not take seriously. The UK's Advertising Standards Authority holds the same line. In an August 27, 2026 guidance note on AI ads, it said obvious exaggeration can be acceptable and summed up its position in a single sentence: "Essentially, what the ad promises, the product should deliver."
The catch: "super intelligence" sits much closer to a factual promise than "smart" does. Outside government, it has a well-known meaning: systems far more capable than humans across most tasks. That is the sense in our SI vs AGI vs ASI explainer. When The Next Web covered Salesforce renaming AIForce to SIForce on October 10, 2026, it noted that renaming a product does not show it meets that meaning. A buyer who reads "superintelligent assistant" may well take you literally, and that's the reading you'd have to defend.
Call this the label gap: the distance between what the word suggests and what the product does on an ordinary Tuesday. Puffery is a small label gap that no one takes seriously. Deception is a gap someone relied on.
What have regulators actually punished companies for saying about AI?
Every major case so far has been about a specific claim, not a word. The SEC fined advisers for saying they used AI when they did not. The FTC went after a "robot lawyer" that was not a lawyer and an AI detector that missed its advertised accuracy. None of these companies got in trouble for saying "AI." They got in trouble for what they said it did.
| Date | Regulator | Company | What they claimed | Outcome |
|---|---|---|---|---|
| Mar 18, 2024 | SEC | Delphia | Used client data to make its "artificial intelligence smarter" | $225,000 penalty |
| Mar 18, 2024 | SEC | Global Predictions | "First regulated AI financial advisor" | $175,000 penalty |
| Sep 25, 2024 | FTC | DoNotPay | "The world's first robot lawyer" | $193,000 and notice to subscribers |
| Apr 28, 2025 | FTC | Workado | AI content detector "98% accurate" | Barred from accuracy claims without evidence |
Sources: SEC press release 2024-36, FTC Operation AI Comply, Courthouse News on Workado.
Workado is the case worth reading twice. The 98% figure came from a study of a different model that was tested only on academic writing. On other kinds of text, the same research put accuracy at about 53%, which the FTC said was "no better than a coin toss." So the company had a real number. It just pinned it on something that number never measured.
Then-SEC Chair Gary Gensler named the pattern in 2024: firms "should not mislead the public by saying they are using an AI model when they are not." Then-FTC Chair Lina Khan put it more bluntly when she announced the sweep: "there is no AI exemption from the laws on the books." You can expect the same reasoning to carry over to SI. The FTC already files its AI cases under a Super Intelligence tag.
Has enforcement got softer under the SI push?
Partly. In other ways it has stepped up. The FTC set aside one 2024 AI order in December 2025 and the administration talks up light-touch regulation. But in October 2026 the White House put FTC Chairman Andrew Ferguson in charge of a new AI enforcement coordination committee, and the FTC sent investigative demands to OpenAI and Anthropic that included their advertising.
The relaxed side is real. The FTC's own records show it reopened and set aside the Rytr order on December 22, 2025. The Workado final order still went through in August 2025, and the agency finalised orders against Cox Media Group over "Active Listening" AI marketing claims on August 27, 2026.
The stricter side is newer. The Daily Caller reported on October 17, 2026 that the White House AI task force picked Ferguson to lead a law enforcement and consumer protection coordination committee with the Justice Department. Civil investigative demands to OpenAI and Anthropic asked for records on safeguards, security incidents, advertising and privacy policies. In July 2026 the FTC also proposed a policy statement saying that quietly distorting AI outputs to serve an undisclosed agenda could be a deceptive practice.
Put those together and the picture is simple: the tone got softer, the truth rule didn't move. Plan for the standard Workado faced.
Which SI phrases are safe, and which ones need evidence?
The ones that describe a category are low risk. The ones that make a comparison, an absolute promise, or a statement about people in the loop need evidence you could show a regulator tomorrow. Use the table as a quick check before an "SI" line goes on your landing page, in your pitch deck or in your App Store listing.
| Phrase | Risk | Why |
|---|---|---|
| "SI-powered notes for sales calls" | Low | Category label, matches the federal synonym for AI |
| "Built on super intelligence (AI) models" | Low | Accurate if you use a frontier model, and the "(AI)" removes doubt |
| "Our SI drafts follow-ups in seconds" | Medium | A speed claim, so keep a timing test on file |
| "Smarter than a human sales rep" | High | Comparison claim, needs proof against humans |
| "Superintelligent: never misses an action item" | High | Absolute promise plus the strong meaning of the word |
| "Fully autonomous SI, no humans involved" | High | False if anyone reviews outputs, a common gap in early products |
The FTC's February 2023 business guidance gives you four questions to run against each line. Are you exaggerating what the product can do? Are you claiming it beats a non-AI product, and can you prove it? Do you understand the risks? Does it actually use AI at all? Swap "SI" for "AI" in each question and they still apply word for word.
Keep investor copy in line with your marketing copy. Both SEC cases came from websites, filings, press releases and social posts. A deck that says "superintelligent agent" while the site says "AI assistant" gives anyone a gap to point at.
What does this look like for a real product?
Here's a hypothetical. Say you run a meeting-notes tool on a frontier model. It's at $3k a month in revenue, and a contractor checks the summaries for your two biggest customers. You want to rebrand the hero to "Superintelligent meeting notes. No humans needed." That line fails twice: it suggests a capability tier you don't have, and it denies human review that is actually happening.
Rewrite it as "SI-powered meeting notes. Summaries in under a minute." That works if you have timed it. The SI label now does what the government says it does: it names the technology. The speed claim is narrow and you can check it. The human review stays out of the hero, and it doesn't need to be there, since you are no longer claiming there are no people involved.
Then keep a one-page evidence log. Date, claim, test, result, and when you will check it again. If anyone ever asks, you can produce it in an afternoon. It also catches the day a model swap quietly breaks something you promised months ago.
Will calling it SI help or hurt with ChatGPT and Google?
Using both terms helps. Dropping "AI" entirely can hurt you. Buyers and the assistants answering them still use "AI" as the everyday term, so a page that only says SI risks not matching the question at all. Put the official term and the common one side by side in your title, H1 and meta description.
AI assistants pick their answers from the text they can read: titles, headings, the first paragraph, schema. If your page says "SI-powered" and nowhere says what SI stands for, a model has to guess. Write it out the first time: "super intelligence (SI), the term the US government now uses for AI." That single line works for search, for assistants and for regulators, because it is the accurate description.
If you use a .si domain or plan to, the same thinking applies to the name. Our guide to what SI means explains both readings, and Should I rebrand my AI startup to SI? covers the search cost of renaming. This post is about marketing law and nothing here is legal advice. If your copy makes performance claims you plan to spend money promoting, get a lawyer to read it.
Once the copy is live, you need to know whether assistants read it the way you meant. Clerion's free AI crawler checker shows the title, headings and schema a model sees on any page, and Clerion's AI traffic report shows whether ChatGPT, Perplexity, Claude, Gemini and Copilot are sending you visitors after the change.
Do this tonight
- Search your site, deck and app listings for "SI," "super intelligence" and "superintelligent." Paste every line into one document.
- Mark each line low, medium or high using the table above. Rewrite every high line so it names the technology, not a capability tier.
- For every medium line, run the test it implies (timing, accuracy, coverage) and write the result in an evidence log with today's date.
- Add one sentence on your main page spelling out super intelligence (SI) next to the word AI.
- Ask ChatGPT and Perplexity "what does [your product] do?" and check whether the answer repeats a claim you just removed.
Related reading
- What is SI (super intelligence)?: the two meanings of SI and where each one comes from.
- What is the legal definition of super intelligence?: what Executive Order 14434 actually defines.
- Should I rebrand my AI startup to SI?: the search and brand cost of renaming.
Frequently asked questions
Is it illegal to call my AI product super intelligence?
No. As of October 2026 no US statute or regulation reserves the phrase, and Executive Order 14434 imposes no obligations on private companies. The legal risk comes from what the label implies: a claim that misleads a reasonable buyer about something material can be deceptive under the FTC Act.
Does Executive Order 14434 require companies to say SI instead of AI?
No. The order, signed September 29, 2026, directs Executive Branch agencies to use Super Intelligence and SI in non-statutory documents. Private companies are not bound, though many review the wording they use in proposals and filings sent to agencies.
What is AI washing?
AI washing is overstating or inventing AI capabilities in marketing or investor materials. The SEC fined Delphia and Global Predictions $400,000 combined in March 2024 for it, and the FTC's Operation AI Comply in September 2024 targeted similar claims, including DoNotPay's robot lawyer.
Is super intelligence puffery?
As a brand word or category label it is close to puffery, which regulators generally do not pursue. Once copy implies a measurable ability, such as beating humans or never making mistakes, it becomes an objective claim that needs evidence before publication.
Should my website say AI or SI?
Use both. Buyers and AI assistants still use AI as the everyday term, so pair them in your title, H1 and meta description, and spell out super intelligence (SI) the first time it appears so models and readers know what it means.
What evidence should I keep for SI or AI claims?
Keep a simple log for each claim: the date, the exact wording, the test you ran, the result and a date to recheck. Workado's 98% accuracy claim failed because the number came from a different model tested on different content.
Sources
- Arnold & Porter: Executive Order directs agencies to say Super Intelligence, not Artificial Intelligence
- Stars and Stripes: Navy memo directs use of super intelligence instead of AI
- SEC announces fiscal year 2027 examination priorities
- SEC charges two investment advisers with making false and misleading statements about their use of AI (2024-36)
- FTC announces crackdown on deceptive AI claims and schemes
- Courthouse News: FTC calls AI detection company's claim of 98% accuracy bogus
- FTC Policy Statement on Deception (1983)
- ASA: Regulating the illusion of intelligence in ads
- The Next Web: Salesforce renames AIForce to SIForce
- FTC: Super Intelligence topic page
- Daily Caller: White House taps FTC to be the bulldog for AI enforcement
- FTC seeks public comment on policy statement addressing AI accuracy
- Kelley Drye: FTC reminds advertisers to keep AI claims in check