Kamala Harris has re-entered the public conversation with a statement on artificial intelligence, and the mockery arrived on schedule. Some of it earned. This is the woman who explained in 2023 that AI is a good thing and, helpfully, that it’s two letters. This is also the former border czar and the Biden administration’s designated AI lead, two portfolios that produced roughly identical results. The joke practically assembles itself: a warning about machine-generated text that reads like machine-generated text.
KAMALA: "AI is kind of a fancy thing…it's two letters! It means artificial intelligence!" pic.twitter.com/6qm58BoST8
— Townhall.com (@townhallcom) July 12, 2023
But the joke is the least interesting thing here, and conservatives who stop at the dunk are going to miss what was actually proposed. Strip away the framing and Harris asked for three concrete things. Slow down frontier development. Create a new federal entity to oversee and independently test the technology (yikes!). And have the president negotiate a treaty with China and others to cap the speed of AI development and set global testing standards. Those are not vague gestures. That is an agenda, and it is the same agenda Barack Obama was reportedly pitching behind closed doors to House Democrats days earlier. When the entire senior wing of a party converges on identical language eight weeks before a midterm, the correct response isn’t laughter. It’s attention.
The frontier of artificial intelligence is advancing at an alarming speed. Over the weekend, industry leaders voiced support for an idea that the American people and many AI researchers in their own labs have been calling for: We must responsibly slow the pace of frontier AI…
— Kamala Harris (@KamalaHarris) September 14, 2026
Take the treaty first, because it’s the easiest to dispose of. Arms control works, when it works at all, because warheads are countable and missile silos are visible from orbit. Model training is none of that. It happens inside buildings, on hardware that has a hundred legitimate uses, by researchers who can be moved and funded quietly. A treaty in which the People’s Republic of China solemnly agrees to develop AI more slowly is a treaty that binds exactly one signatory: us. Beijing has spent two decades demonstrating what its commitments are worth on trade, on fentanyl precursors, on Hong Kong. Anyone proposing to hand it a veto over American technological development in exchange for a signature should be asked to name the verification regime first. They never can. China is in an AI race with us; no wonder they’ve already said ‘no way’ to this asinine proposal.
The new federal entity is the bigger prize, and it’s the part that should worry people regardless of how they feel about AI. A standing agency with authority to license, test, and gate what gets built does not stay neutral. It gets captured — generally by whichever firms are large enough to staff the compliance department and lobby the rulemaking. The last thing we need is another lobbyist-driven federal rulemaking entity, especially with something as vital as AI.
Safety regulation written with the incumbents at the table is a moat with a nice name. And the constitutional problem is the one conservatives have been fighting for a decade — Congress writing a two-page mandate and delegating the actual lawmaking to unelected administrators who will decide, with no vote and no accountability, which software an American may build. No, thanks.
None of which means there’s nothing to address. Deepfakes in elections, synthetic fraud aimed at the elderly, chatbots doing real harm to kids: those are concrete, they’re already happening, and they can be handled with laws that name the harm and apply to everyone, passed by people who face voters. That’s the conservative answer — legislate the abuse, not the technology. It’s harder work than announcing a new agency, which is precisely why nobody announces it.
So go ahead and enjoy the clip. Then read the statement again and notice that beneath the syntax is a request to hand permanent regulatory control of the most important technology of the century to the administrative state and to negotiate its pace with Beijing. Harris may not be the most persuasive messenger her party could have sent. That doesn’t make the message harmless. It makes it easy to underestimate.
