Trump's "Super Intelligence" Renaming Order: Who Is Passing the Loyalty Test

Trump signed an executive order requiring government agencies to uniformly rename "Artificial Intelligence" as "Super Intelligence." On the surface, this is merely a terminological adjustment, but in essence it is a loyalty test aimed at the tech industry—who follows, who stays silent, and who opposes all become windows for observing corporate stances.

The Political Logic Behind a Single Order

According to The Algorithm, a newsletter under MIT Technology Review, this executive order signed in late September contains only one directive: requiring government agencies to rename what had always been called "Artificial Intelligence" as "Super Intelligence."

The order itself has virtually zero technical substance—no adjustments to algorithms, computing power, or regulatory frameworks; only the wording in documents changes. But precisely because it is so simple, it exposes a clear intent: this is a declaration test. Terminology choice here is not an academic issue but a political alignment issue.

When a term is redefined top-down by administrative power, its use or non-use naturally forms a dividing line. Those who follow gain political capital, those who hesitate are marked as观望 (wait-and-see), and those who oppose may face pressure. This logic is not unfamiliar in the tech industry, but applying it directly to a core technical term is still quite rare.

An executive order is a legally binding directive issued by the U.S. President directly to federal administrative agencies, bypassing the congressional legislative process. It is typically used to adjust agency operations or policy implementation details. Its effect is limited to within the executive branch; it cannot create new legal rights or obligations, and it can easily be revoked or modified by a successor president. For this reason, using an executive order to address terminology issues has extremely low cost and strong symbolic significance: no congressional wrangling is needed, yet a unified口径 (line) can immediately be formed in federal documents, agency websites, and external communications, thereby sending a signal to the entire industry. This also explains why an order that seemingly only changes wording is still seen as having political weight.

Who Is Following, Who Is Watching

The report's title directly points to the core question: "Who is lining up to pass Trump's 'Super Intelligence' loyalty test." From public reactions, responders can be roughly divided into two categories.

One category consists of agencies and contractors with direct business dealings with the federal government; their wording adjustments are almost instantaneous—in fields where government procurement and regulatory approvals are highly dependent on administrative will, the cost of following is low and the benefits are clear. The other category consists of companies hoping to gain a voice in policymaking; actively adopting the new term is equivalent to sending a signal of cooperation to decision-makers.

The situation of those watching is more delicate. For many tech companies, "Artificial Intelligence" is a term used for years in product documentation, academic papers, and industry standards; rashly replacing it would bring practical operational costs and might also trigger internal and external doubts. Silence thus becomes a compromise strategy: neither publicly opposing nor actively cooperating.

It should be noted that the report does not list specific companies one by one, and the degree of reaction among parties in public information also varies. As for which companies have explicitly stated their positions and which have only made technical adjustments, there is currently a lack of complete third-party verification.

The Industry Cost of Politicizing Terminology

Renaming "Artificial Intelligence" as "Super Intelligence" ostensibly elevates the规格 (status) of the concept, but in reality creates new confusion.

In the tech community, "Super Intelligence" usually refers to a form of intelligence far beyond human level, clearly distinct from the AI systems currently deployed. Using the same term in an executive order to cover two different levels of concepts will cause misalignment between policy discussions, regulatory texts, and public perception. Researchers need to additionally explain terminological differences, companies need to weigh wording in compliance documents, and international peers need to realign semantics when communicating.

The deeper impact lies in expectation management. When official discourse calls existing technology "Super Intelligence," public judgments about technological capabilities and risks are elevated, which may both foster unrealistic expectations and trigger excessive panic. Neither outcome is conducive to rational technology governance.

In AI research, capability levels are usually roughly divided into Narrow AI (ANI, focused on a single task), Artificial General Intelligence (AGI, reaching or surpassing human level across a wide range of tasks), and Superintelligence (significantly surpassing the highest human level in almost all domains). The three correspond to completely different levels of technological maturity, risk assessment, and governance frameworks. Broadly calling currently deployed narrow AI systems "Super Intelligence" essentially erases these level differences, depriving regulatory classification, risk grading, and responsibility attribution of an operable semantic basis, and making it difficult for the tech and policy communities to dialogue within the same conceptual coordinate system.

The Real Target of the Loyalty Test

The key to this renaming controversy is not whether the term "Super Intelligence" is accurate, but what it tests.

If the content of an order is merely changing the name, then cooperation or non-cooperation no longer reflects technical judgment but only political will. Companies' choices in terminology are endowed with meanings far beyond the terminology itself. Once this mechanism is established, when substantive issues such as data, computing power, and export controls arise later, the same declaration pressure may appear in stronger forms.

For the tech industry, the real risk is not writing one more word or one fewer word, but the transfer of the power to define technical discourse from the professional community to the administrative hierarchy. Terminology can be changed back, but the transfer of discursive power is often irreversible.

The Algorithm's report uses "who is lining up" as an entry point, but actually raises a question more worth asking: when loyalty becomes the measure, how much independent space remains for technical judgment. The answer will not be written in the executive order, but in each company's next public document.