Today, the White House published Executive Order 14434, Inaugurating the Era of Super Intelligence. The order asks federal agencies to stop saying “Artificial Intelligence” and start saying “Super Intelligence”. AI becomes SI. Everywhere: official correspondence, websites, reports, policy documents. So far, so… meh.
My instant reaction, and the motivation to write this post, is that it’s the first step in a process that could lead to a new legal category of SI.
What the order actually says
Right now, EO 14434 changes nothing serious, really. Section 3(a) says it explicitly:
“Super Intelligence” and “SI” mean the technologies and systems encompassed by the term “artificial intelligence” as defined in section 9401(3) of title 15, United States Code.
Same same, but different name. But then, Section 3(b) orders the Assistant to the President for Science and Technology to propose, within 60 days, a new legislation to establish the federal definition of SI. The proposal must assess whether SI should modify, expand, or supersede the existing statutory definition of AI.
That’s the part that caught my attention.
What could happen next
Here’s one possible trajectory:
- Label: The executive branch starts calling everything SI.
- Definition: Congress or the White House defines SI with specific criteria (capability, autonomy, safety, auditability…).
- Certification: Systems used in critical sectors or federal contracts must meet SI requirements.
- Procurement: Federal agencies can only acquire SI-certified systems.
- Subsidies: Companies receiving federal funding must use SI-certified providers.
- Standards: Industry adopts SI compliance as a market norm.
Suddenly, nobody is talking about AI as a generic thing. Now is about SI.
Why It Smells Bad To Me
Imagine two models a couple of years from now:
| Qwen-X (hypothetical) | US Frontier-X (hypothetical) | |
|---|---|---|
| Benchmark performance | 10/10 | 8.5/10 |
| Cost-efficiency | 10/10 | 3/10 |
| SI-certified | No | Yes |
The US model doesn’t even need to be better anymore. It needs to be SI-certified. And, somehow, only US decides which models are SI-certified.
You Are Exaggerating
Yeah… maybe, but US has a well-documented history of using so-called “neutral” mechanisms to shape competitive outcomes without explicitly banning foreign products. Take the “Buy American Act (1933)” as an example, where Federal procurement preferred domestic goods. Nobody said “Japanese steel is banned”. The government said “we prefer American products” and applied evaluation adjustments that made foreign suppliers compete at a structural disadvantage.
The pattern is consistent: you just need to control enough of the infrastructure, standards, and procurement rules that your product becomes the default in the markets that generate scale. Scale generates revenue. Revenue funds the next generation of models. The cycle reinforces itself.
Banning A Product VS Defining Who Can Use It
This is the core of the mechanism, and it’s worth stating clearly.
Policy A: “Models from Alibaba are prohibited”
Visible. Politically costly. Raises immediate questions about trade law, discrimination, reciprocity, and market freedom.
Policy B: “Entities operating in regulated sectors must use SI-certified systems”
Looks like a technical standard. Doesn’t mention any country. Doesn’t prohibit any product. But if SI certification structurally favors closed, US-based, centrally operated models, the effect can be identical.
You haven’t banned anyone’s product. You’ve redefined what counts as admissible. And that’s a much more durable tool than a ban.
That would be especially relevant if Chinese open-weight models demonstrate that pure technical competition can no longer be won on engineering alone. The battlefield shifts from engineering to engineering plus regulation plus procurement plus geopolitics. And that shift has precedent in every strategic technology the US has ever cared about.