Felix Ortiz: Rough week for anyone who works in federal communications, I'll tell you that much — how was yours?
Tess Hollis: Fine, but now I need to know what happened to federal communications people.
Felix Ortiz: So — September 29th, 2026, Trump signs an order called 'Inaugurating the Era of Super Intelligence.' And what it does, mechanically, is instruct all federal agencies to retire the terms 'artificial intelligence' and 'AI' from official use. Replace them everywhere with 'super intelligence' and 'SI.' Every document, every agency.
Tess Hollis: Wait — is that enforceable? Like, who checks?
Felix Ortiz: That question gets more pointed when you know that by October 8th, Trump was publicly calling holdouts — people or outlets who wouldn't adopt the new term — 'THE ENEMY.' His capitalization, not mine.
Tess Hollis: Okay — but 'superintelligence' already means something. Nick Bostrom defined it in 2014 as a hypothetical agent exceeding human cognitive performance across virtually all domains. That's not a vague term, that's a technical bar. Has any system cleared it?
Felix Ortiz: No independent scientific body has said yes — not even close, as far as we know. Which means the government didn't just rename the technology. It declared — by executive order — that we're already in an era defined by something that hasn't been built yet, scientifically speaking.
Tess Hollis: That's a reality-declaration, not a label update. And someone chose that word deliberately — so what does the word choice actually do?
Felix Ortiz: Yeah — and here's the click, I think. When a company renames a product, they're doing marketing. When a government issues an executive order and calls holdouts the enemy, the name has legal force. It's not — it's not describing a thing anymore. It's commanding what the thing is allowed to be called inside every agency that writes rules about it.
Tess Hollis: The analogy is like the government declaring a prototype car is already a finished product, and requiring mechanics, insurers, and regulators to call it finished. The name shapes what questions you're allowed to ask.
Felix Ortiz: Right — because if it's 'finished,' the regulator can't ask 'is this safe yet.'
Tess Hollis: And CNET actually made that call explicitly — they rejected the label and kept using 'AI' per AP Stylebook guidance. Which is a newsroom saying: we won't let an executive order override editorial judgment about what's scientifically accurate.
Felix Ortiz: Okay — but don't companies rebrand stuff all the time? Like, is this actually different in kind, or just — wait, actually I think it is different, because Naomi Susan Baron called it 'Pollyannish and dangerous,' and that phrase does real work. 'Pollyannish' means it's sunny about capabilities that don't exist. 'Dangerous' means the sunny framing actively blocks the hard questions.
Tess Hollis: And the dangerous part lands harder when you look at who can't agree on what the word even means. Mark Zuckerberg calls superintelligence a personal AI that expands individual capability. Mustafa Suleyman calls it 'humanist superintelligence' — powerful but bounded. Those are contradictory definitions. And the government just picked one triumphalist label and made it mandatory over both of them.
Felix Ortiz: Oh — that's the part that actually gets me. The experts can't agree, and the government didn't resolve the disagreement. It just — steamrolled past it.
Tess Hollis: That's not clarification. That's consolidation of narrative power. And the AI Now Institute would say that's the whole mechanism — the label makes technologists look like the only people qualified to speak, even on questions that are really about labor, privacy, accountability.
Felix Ortiz: And that authority-concentration thing — okay, that's what makes the 'artificial' deletion the real move, not the 'super' addition. Because 'artificial' is doing quiet accountability work every time it appears in a document.
Tess Hollis: That's the thing nobody is saying out loud. 'Artificial' contains 'made.' Made things can be audited. Made things have an author. The moment you swap it for 'super,' the technology appears to transcend that logic entirely.
Felix Ortiz: Oh — wait. So the accountability question doesn't just get harder. It gets — yeah, it gets linguistically erased before anyone even files a complaint.
Tess Hollis: Picture a compliance officer at a mid-size logistics firm. She gets an email from legal: does the route-optimization tool get relabeled 'super intelligence' in the federal grant documentation? The tool hasn't changed. Not one line of code. Only the political stance attached to it has.
Felix Ortiz: And she can't push back by pointing to the science, because the government's SI label already covers that tool — it applies to the same systems federal statutes were calling AI a month ago. The label jumped ahead of any capability change.
Tess Hollis: Right — and that framing shift is measurable. Research on linguistic framing effects shows the specific words used to describe a technology change whose expertise gets treated as authoritative and what policy responses even seem possible.
Felix Ortiz: Okay, and then — this is the part that made me actually laugh — 'SI' is Slovenia's country-code internet domain. Dot-S-I. And the mandate triggered domain registration speculation almost immediately. People buying up SI-adjacent domains because a federal order accidentally made those letters valuable. I want to be honest, the sourcing on the speculation side is thin, but the collision itself is real.
Tess Hollis: So the top-down mandate produced material chaos its authors definitely did not anticipate.
Felix Ortiz: Yeah — and what's wild is this isn't even the first time AI vocabulary has cycled through a politically loaded rebrand. 'Strong AI,' 'AGI,' now 'SI' — what makes this one different in force and speed is something we should actually get into.
Tess Hollis: But that lineage — 'strong AI,' then 'AGI,' now 'SI' — every one of those resets the clock on what was promised and what accountability was owed. AGI was literally a 2000s rebranding of 'strong AI' after narrow systems failed to deliver general reasoning. Nobody got fired for that.
Felix Ortiz: Right, and — okay, here's what I keep tripping on. Bostrom's 2014 'Superintelligence' spent years migrating from fringe philosophy into actual policy discourse. Years. That's the normal speed of how a loaded term captures a field. And this executive order did the same migration overnight. With teeth.
Tess Hollis: So what's actually different this time isn't the content of the rename — it's the mechanism?
Felix Ortiz: Yeah — velocity plus enforcement. Those two things together are new. And the AI Now Institute's point fits exactly here — because every rename concentrates authority, but this one concentrates it in the executive, not just in technologists. It's not 'trust the engineers to define this.' It's 'the president defines this, and calling it something else makes you the enemy.'
Tess Hollis: And Bostrom was actually careful. His framing was explicitly about hypothetical future scenarios, recursive self-improvement, exponential capability growth — hedged throughout. The order just borrowed his grammar and stripped every hedge.
Felix Ortiz: Oh — that's actually — wait, that's the part that's been bothering me and I couldn't name it. The grammar got imported, the caution didn't. So now narrow AI systems — tools optimized for specific tasks — get the 'super' label that Bostrom reserved for something that doesn't exist yet.
Tess Hollis: That collapse is doing real work. A dermatology image classifier is narrow AI. It does one thing. Under 'SI,' it's categorically the same as a hypothetical general reasoning system. That distinction — narrow versus general — is gone.
Felix Ortiz: And the regulator writing safety rules for that classifier now has to use vocabulary that implies it's already beyond human-level cognition. She can't write 'this system has known failure modes in darker skin tones' and also call it superintelligence without the words fighting each other.
Tess Hollis: That's the accountability clock resetting. Same as it did with AGI, same as with strong AI — except this time there's $280 billion in private AI spending in 2026 sitting on top of whatever label the government picks. The stakes of the vocabulary choice are just categorically larger.
Felix Ortiz: And that's — yeah, that's where it lands for me. Politico put U.S. private AI spending at roughly $280 billion in 2026. Seventy percent of S&P 500 companies already running these systems. The vocabulary that wraps all of that determines which risks even show up on a regulator's radar. Safety, labor, privacy, national security — they all get filtered through whatever word the government mandated.
Tess Hollis: And Mustafa Suleyman's 'humanist superintelligence' framing is just — sitting there. Powerful systems, deliberately bounded, designed to stay controllable. That vocabulary exists. It didn't get picked.
Felix Ortiz: Right — which means this wasn't a forced choice between 'AI' and one mandatory triumphalist label. There were other options. Someone chose the one that doesn't contain any brakes.
Tess Hollis: If the public can't contest the official name — because contesting it now makes you 'THE ENEMY' by executive decree — they lose the lever. That's the part that doesn't resolve cleanly.
Felix Ortiz: Yeah. And we started this whole thing with a rough week for federal communications people — that felt almost funny at the open. It's a little less funny sitting here.
Tess Hollis: No, it really is. Good one to sit with.
