Brian Schatz

Madam President, you know, this is about a couple of very simple questions. The first question is, Humans in control of technology or not? That is the first question.

Even the mechanisms that we are accustomed to, for instance, in aviation, where you say ``Before this plane can take off, we have to certify to its safety,'' don't really get at the fundamental problem because what we have is technology that is equivalent to a plane being able to take off on its own while it is still being built. So the question is first, Humans in control or not?

The second question is, If we agree that humans should be in control, should this be voluntary or not? Do we trust--not any individual CEO or board of directors, but do we trust that companies that are rolling toward the biggest IPOs in human history will voluntarily regulate themselves as opposed to maximize the value of that IPO, maximize shareholder profit?

We have traditionally tried to strike the right balance, and Democrats and Republicans don't always see the sort of level of regulation necessary on an eye-to-eye basis, but I think at some level, there are 100 of us who believe that when it comes to serious risk to public safety, that those should not be voluntary safety regimes. They should be mandatory. They should be Federal. It shouldn't be every legislature. It shouldn't be some announcement from a bunch of tech CEOs. It has to be in Federal statutory law.

Now, the third question is, Should we do this comprehensively or not?

The answer, in my view, is not--not right now. Here is why. We don't have the capacity to pass a very complicated law to deal with all of the societal and economic and human impacts of the AI revolution. We are not going to do that

with the House out of town. We are not going to do that with the Speaker of the House being sort of so skeptical of any policymaking in this space that he basically left town to avoid dealing with this. So what we have is the most modest but most serious proposal to take down the acute risk that we are being told by the labs and by AI safety experts and others really does exist.

We could have an argument about ``Is it a 10-percent chance of a terrible thing happening or a 1-percent chance of a terrible thing happening? How much time do we have?'' but I think nobody is disagreeing that the risk is nonzero of something terrible happening if we don't have a basic architecture of these labs having to not just test before they commercially upload something but to have a safety regime from the very beginning because, as we now have learned, this is a unique kind of product that could become dangerous even before it is released to the public.

So the question is, Comprehensive versus not--and as much as Senator Warner and I talk about our pet rocks--I have a bunch of bills in the AI space that are important. Of course, I think mine are the best. I am sure Senator Kim and Senator Warner think theirs are the best. But the point is, we are not going to come to a resolution on this. This bill is trying to thread that needle of being nonpartisan, on the level, not a setup, not designed as an electoral bludgeon but simply to try to take down the acute risk that is right in front of us.

So that is why we are trying to be modest enough to actually be able to look the Commerce Committee chairman in the eye and say: There are no tricks or traps here. This is what we have come to conclude is the safest way forward in the very short run. It also doesn't preclude Senator Cruz and Leader Thune and Senator Klobuchar and others from legislating in this space because there is lots more to do, but right now, we are facing something rather precipitous.

The other question is, Are we going to do something after something terrible happens or not?

I think that we have an opportunity to tee up legislation that takes down--that doesn't reduce to zero--but that takes down the risk of something really horrifying happening to people in the United States of America. So this is about humans in control, this is about keeping politics out of it, and this is about taking down the short-term risk of something really terrible happening.

To the extent that it is possible in this polarized time and in this election season, we tried like heck to display enough discipline on our own to not introduce a bunch of stuff that was going to freak a bunch of other people out. I actually haven't heard of any substantive disagreements with what we are proposing--I am sure they exist--but in terms of Member interaction, in terms of the expert community, in terms of my Republican colleagues, people aren't saying: I hate this bill. I have an ideological objection to it. It won't work.

They are not saying that.

They are just saying: We are doing another process and give that time.

And my judgment is that we don't have the luxury of time.

You know, the Senate is supposed to be the cooling saucer that allows the hot liquid from the House to cool off, and then we can deliberate, right? Well, the House is out of town, and they did not send us anything to drink. So that is a lovely way to understand this body, but we are all that is left, and we have to work on a nonpartisan--forget bipartisan--on a nonpartisan basis to make sure that humans remain in control of the technology.

With that and with the permission of the Senator from Virginia, I yield to my friend the junior Senator from New Jersey who knows more about national security and who is, I think, appropriately alarmed at the potential impacts of inaction in the U.S. Senate.

Madam President, I am mildly encouraged. I, obviously, would rather he not object, but I am encouraged by the potential for common ground here.

Composition of Boards, what the mechanism is for ensuring safety is something that I personally don't have religion on. The question at hand is: Should we have some sort of mandatory safety and cyber security regime, and how do we stand up something that is quick enough to deal with the hastening pace of this technology and the

abundant evidence now that some of these AIs are escaping human control?

Look, I am always ready to modify something. If the Secretary of the Commerce Department shouldn't be the person in charge of determining the AI safety standards, that this should take 2 weeks rather than 6 months, you know, I am very openminded. But the threshold question is: Should this be voluntary or mandatory? And if you are saying you are open to some sort of mandatory regime with the caveats that you have expressed, I think that is progress.

I will make one final point. I think we get a little hung up on anthropomorphizing the tech itself, including this idea that the values are going to be imbued in the AI.

The AI gets given a task. It doesn't have ideas of its own. It doesn't have thoughts of its own. It doesn't have consciousness. It doesn't have feelings. And so this idea that we are going to have American, patriotic, liberty-loving AI and they are going to have the other kind I think is a misunderstanding of what we are actually talking about right now, which is we actually just want normal cyber security controls for this incredibly powerful tech. And the cyber security controls are going to have to be more robust than ever, but, like, that does not seem at all inconsistent with American values.

And if you want us to win the AI race--the quickest way to lose it is to allow a terrible incident to happen without any Federal guardrails, and then I think, figuratively speaking, the pitchforks will be coming.

I ask for the yeas and nays.

Madam President, I want to thank the Senator from Virginia and ranking member of the Intelligence Committee and also thank Senator Kim, the Senator from New Jersey, both of whom have real expertise in this area and as it intersects with public policy.

This is really about one simple thing, and that is, humans in control of the code and not the code in control of the humans.

We are at a seminal moment in world history, not just in American history, where it is possible for this code to escape human control. That is not an apocalyptic statement from a person who wants to make a point in a political context; that is a fact that is new to all of us. This code is capable of escaping our control, and it is potentially capable of something they call recursive self-improvement, which means the code can write its own code and that humans would find the code that was written by AI so inscrutable as to not be able to control it at all.

So what we are proposing--and I want to reiterate what Senator Warner said. When we were in private conversation, he said: Nobody can bring their pet rock to this conversation.

What do I mean by ``pet rock''?

Everybody has got a bill on AI. By the way, they are really important things--energy use, the safety of children, the future of work, the future of intellectual property--but we are not going to solve that in a 10-day period. So what we sat down on a bipartisan basis to do with AI safety groups, with companies that are in the business--with experts--is to say: How do we reduce the acute risk now, not the societal risk over the long run, not to balance the equities between whatever it may be 2, 3, 4 years from now. How do we take down what some people are calling civilizational risk? Some people find that too apocalyptical. Fine. Let's not call it risk of extinction. Let's not call it civilizational risk. Let's just call it a high risk of something very terrible happening, and that should be enough for us to be on the floor to say: Well, why don't we try to take down that risk?

There is a very straightforward pathway. First of all, it is predeployment testing, but one of the things, I think, we have all learned is that, by the time a company is ready to deploy what they call a frontier model, it has been through some testing. So it is important to test it before it is commercially available, but the truth is that the Hugging Face incident, in particular--that thing jumped the fence way, way earlier than when you would do the predeployment testing. So we need third-party review from the jump.

And the thing about a sandbox--and I was taught this term of art about 10 years ago, and I remember talking to my tech expert in my office: What is a sandbox?

And they said: You should think of it like an actual sandbox, sort of where you experiment, where there are no wrong answers, where it may get a little messy.

So the configuration of these sandboxes has been like that--for experimentation, almost casual, very entrepreneurial.

But the problem is, these things can escape control at the sandbox level so it is not enough for predeployment testing. We also need third-party review of the process at every stage because we don't know at which point an agent--an AI--or a swarm of agents could escape human control. Then, at the end of it, we need postdeployment breach notification, analysis, transparency so that, if there are big, dangerous events, we know how to prevent them from happening in the future.

So that is all this bill does. It says: Let's get third-party review all the way through the process and then all the way up to predeployment testing. Then, if something goes badly, we have transparency on the back end. We threw out all of our ``pet rocks''-- nothing partisan, nothing like my longtime priority on AI, but rather the very narrow question, which is: Don't we have an obligation, if there is any risk of a power grid going down, if there is any risk of a financial system buckling, if there is any risk at all, don't we have an obligation to take down that risk

and fight about the rest of the stuff at some other time?

The kind of ironic thing about this Congress right now is that the private sector is begging to enshrine what they are starting to promise to do in Federal statutory law because they are about to say: Here are all of the safety measures we are doing. But there is a reason you don't regulate toxic chemicals or biomedical research or aviation by voluntary agreement. You need a Federal statute, and then the private sector can work toward compliance on that Federal statutory standard.

But you don't want 90 percent of the AI companies--the frontier models, the hyperscalers--complying and then the other 10 percent saying: Wait, I don't have to do that, do I? I don't have to do that, do I?

So all we are asking is that this bipartisan proposal--and it is bipartisan. We are working on it in very, very good faith to try to narrow the scope of disagreement but not narrow it to the point that we pass nothing but, rather, narrow it to the point where we pass something that takes down the immediate risk and allows us to look our constituents in the eye and say: We did something on this very, very serious and urgent matter.

With that, with the permission of the ranking member, I want to bring in the Senator from New Jersey, who has a background in tech and has a background in national security that I think adds a lot to this conversation.