US renames AI ‘Super Intelligence’ — with policy consequences

What this is about

US President Donald Trump signed an executive order on September 29, 2026, directing federal agencies to use the term “Super Intelligence” instead of “Artificial Intelligence” in non-statutory documents. The change applies to official correspondence, websites, reports, policies, and other executive-branch publications. Within 60 days, the president’s science and technology adviser must also propose a statutory definition.

On the same day, leaders from Google, Anthropic, Meta, OpenAI, xAI, and Nvidia signed a safety accord with the government for powerful AI models. The pairing is unusual: the government is changing the political language first, while the most important providers promise voluntary controls.

What the order and accord actually do

The order does not amend existing laws, contracts, or historical documents. Until Congress acts otherwise, the order says “Super Intelligence” covers the same systems that US law already defines as artificial intelligence. In practical terms, it initially creates a mandatory language rule for the executive branch.

The legislative proposal due after 60 days must assess whether the new definition should modify, expand, or replace the existing AI definition. According to Nextgov, the parallel industry accord names four layers of control: internal monitoring of capabilities and alignment, an internal oversight team, independent external auditors, and an independent committee overseeing those reports.

Why it matters

Words shape policy. If agencies start calling a broad range of current systems “Super Intelligence,” expectations and public debate may shift even though the technology did not become more capable overnight. A later statutory definition could also determine which providers, models, and applications face special obligations.

The accord matters because its signatories control a large share of leading models and compute infrastructure. Yet it relies on self-governance and external audits rather than a new regulator. Vice President JD Vance rejected an AI regulator modeled on aviation or drug authorities, according to Nextgov, and pointed to existing powers held by the FTC and Department of Justice.

In plain language

Imagine a city deciding to call every bicycle a “high-speed vehicle.” The bicycles do not become faster, but forms, signs, and political arguments suddenly sound more urgent. At the same time, manufacturers promise to inspect their own brakes and let outside inspectors in. The key questions remain who writes the test and what happens when a product fails.

A practical example

A US agency publishes guidance in November 2026 for a system that sorts applications. Under the order, the document calls it “Super Intelligence,” even though it only classifies documents. A provider of a more capable model sets internal safety limits, asks an external auditor to run 100 attack scenarios, and reports the results to an independent committee. The order controls the language; the accord describes the review process. Neither automatically guarantees that a failing system will be withdrawn.

Scope and limits

• The order is initially a language and definition policy. It does not prove new technical capabilities or immediately create enforceable private rights.
• The industry accord is voluntary. Public details do not yet show how strict the test standards are, who selects auditors, or what sanctions follow a breach.
• The statutory definition does not exist yet. Only the promised proposal and any later act of Congress will show whether legal duties actually change.

SEO & GEO keywords

Super Intelligence, artificial intelligence, White House, executive order, Donald Trump, US AI regulation, AI Safety Accord, OpenAI, Anthropic, Google, Nvidia, Meta