What is Super Intelligence?

What happened in September 2026?

Executive Order 14434, signed September 29, 2026 and published in the Federal Register on October 2, directs federal executive agencies to use “Super Intelligence” and “SI” instead of “Artificial Intelligence” and “AI” in certain official communications. The important starting point is the scope: this is a federal executive-branch terminology policy, not a worldwide change in scientific vocabulary.

The White House executive order is the primary source for the directive. The Federal Register publication provides the official published text. Read the document itself for the precise requirements, qualifications, and context rather than treating a headline as the full policy. For the reasons the order itself gives, see why the order renamed AI to SI.

Who does the instruction apply to?

The instruction concerns federal executive agencies and certain official communications, including websites, reports, correspondence, and non-statutory policy documents, to the maximum extent permitted by law. It should not be read as a requirement that every person, business, researcher, school, or government outside that branch stop saying AI. Section 2 does not require alteration of previously issued regulations, Presidential actions, contracts, grants, or other historical documents.

For a beginner, the practical question is “Who is using this term?” If you encounter SI in federal communications, the policy meaning may be relevant. If you encounter superintelligence in a research discussion, the speaker may mean a proposed capability level instead. Those contexts are not interchangeable.

Does the order define a new technology?

Section 3 currently defines SI by reference to the existing federal AI definition in 15 U.S.C. 9401(3), except where otherwise provided by law. That connection matters: the new term points back to a category already recognized in federal law. It is not, on its own, a test demonstrating that a system has acquired greater intelligence or can outperform people across broad domains.

The order also requests proposed legislative language within 60 days. A request for a proposal is not an enacted new definition. As of this guide’s October 9, 2026 editorial check, the order’s reference to the existing statutory definition is the relevant starting point.

A change in wording does not itself change how a model is trained, what data it can access, how accurate its answers are, or whether it can act independently. Those are questions about the particular system. They require evidence about its actual behavior, not an inference from the letters on a page.

What does superintelligence mean in research?

Technical artificial superintelligence, often abbreviated ASI, is generally discussed as a hypothetical system exceeding human capability across broad domains. It is more specific than “a helpful chatbot” or “software that is good at one task.” Researchers may debate definitions and evaluation methods; a terminology directive does not settle those debates.

You can therefore understand the policy without assuming that today’s tools are technically superintelligent. When someone claims a system has reached that level, look for a clear definition, supporting evidence, and the limitations of the evaluation.

What should you take away?

Read SI in context. Separate an official naming choice from a demonstrated technical achievement. You can continue learning about AI using ordinary terms while recognizing the federal policy’s vocabulary. Our SI vs AI comparison puts the three meanings side by side; PolitiFact’s explainer offers a secondary account to compare with the official documents.