Warner-Schatz-Kim AI Security Bill - Senate Floor Remarks on Frontier AI Testing, Model Safety Plans & AI Incident Reporting - September 24, 2026

00:00We've seen one company, OpenAI, launch a series of 3,000 AI agents that jumped over the security barriers, attacked a different company, Hugging Face, came back and tried to attack their own company, these AI agents together.

00:20And the remarkable thing is these 3,000 agents talked to each other, communicated, tried to hide from humans and literally left little bits of code on their path so that subsequent groups of AI agents might come and discover those.

00:31I've met over the last couple of weeks with business leaders from finance, from across corporate America that all say, what are you guys doing? This is the issue of our time. Yes, I got frustrations on data center, but this is the issue of our time.

00:49This is going to be the biggest transformation in our lifetimes. and we don't have even de minimis safety protocols. We are completely relying on the goodwill of these large AI frontier tech companies to kind of do the right thing.

01:12That's irresponsible. And I'm proud of the role I play in the national security domain. I've been vice chair of the Intel Committee for the last decade. Another company, not OpenAI, that was the Hugging Fane Center, Anthropic, created a new model called Mythos.

01:30And our head of cybersecurity, Cybercom, and the NSA, National Security Administration, General Rudd said basically, Mythos is going to change warfare almost as much as drones. And luckily, we got a first look at that, but only because the one company was willing to give that forward.

01:53So we ought to have a time to have a fulsome debate about data centers, kid safety, liability, other issues. But right now, and we were going to do this today, we're going to come back on Monday because the Senate's left.

02:10But we've kind of, the three of us, and we're talking to other colleagues on both sides of the aisle, are saying let's strip back all of our personal ideas. and let just put a marker down that potentially we could act on because we hear from some of the technology leaders that we might lose the ability to control some of these tools not in years or months but some folks and I not fully believing in weeks I don't believe I'm not a doomer on AI.

02:38I don't think the robots are going to take over the world. But I do think if we had an event where an AI tool went out and took over a hospital or took over a water system or shut down a power system, Americans already concerned about the rate of this technology change.

02:54if they could shut down this innovation. And that would be the worst for our country. We need to lead this in terms of some of our competitors like China. So what we're going to be proposing is, I think, as simple as possible.

03:12Before any AI model coming out of these frontier labs is released to the public, it ought to be tested to make sure it's safe. A lot of these big companies also do some self-testing, but they do it in what's called a sandbox, where they're supposed to protect it.

03:31Well, we've seen some of these sandboxes are pretty sloppy. We ought to create basic cybersecurity standards for that sandbox. And when you start at the very beginning of how an AI model is developed, we ought to make sure that company has some basic safety protocol in place.

03:51And then, God forbid, if there is an incident, there ought to be an incident reporting so we know what's happening. Because the difference is, you know, in past innovation, you could have a mistake, but the amount of damage that could happen would be fairly limited.

04:11Potentially, this could be so much greater. And we literally are having the companies themselves say, help us. Now, what knows? We've got to make sure that this is not an effort to simply put in sconce the four or five biggest companies.

04:29We've got to make sure smaller tech. VC has a place on this safety panel. But we need national security. We need our top scientists. We propose NIST. We are open on how to do it better.

04:44But it needs to be a public-private relationship that is a safety board. We wouldn't climb on an airplane. You wouldn't get in a car. You wouldn get a new financial product without some regulatory oversight Something that is as big as this We should not miss the moment God forbid you know we go away for our campaign season and in the interim something happens and we had no safety in place.

05:09So moments like this, and I can't think I had spent a career in technology before I got involved in politics. I've never felt a moment like this. So I hope that we will act.

05:15And with that, Madam President, I'd like to yield the floor to my friend from Hawaii. I want to thank the senator from Virginia. Senator from Hawaii. Thank you. I want to thank the senator from Virginia, the 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 also as it intersects with public policy.

05:46this is really about one simple thing and that is humans in control of the code and not code in control of the humans we are at a seminal moment in world history not just American history where it is possible for this code to escape human control and 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.

06:19This code is capable of escaping our control and is potentially capable of something they call recursive self-improvement, which means that the code can write its own code and that humans would find that code that was written by an AI so inscrutable as to not be able to control it at all.

06:40And so what we are proposing, and I want to reiterate what Senator Warner said, when we were in a private conversation, he said, nobody can bring their pet rock to this conversation. What do I mean by pet rock?

06:52Everybody's got a bill on AI. And by the way, they're really important things. Energy use, the safety of children, the future of work, the future of intellectual property. But we're not going to solve that in a 10-day period.

07:08And 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?

07:23Not the societal risk over the long run not to balance the equities between whatever it may be two three four years from now But how do we take down what some people are calling civilizational risk and some people find that too apocryphal?

07:38Fine. 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?

07:56And so there is a very straightforward pathway. And first of all, it is pre-deployment testing, but one of the things I think we've all learned is that by the time a company is ready to deploy what they call a frontier model, it's been through some testing.

08:16And so it is important to test it before it's commercially available. but the truth is the hugging face incident in particular, that thing jumped the fence way, way earlier than when you would do the pre-deployment testing.

08:30So we need third-party review from the jump. And the thing about a sandbox, and I was taught this term of art like 10 years ago, and I remember talking to my tech expert in my office, what's a sandbox?

08:45And 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. And so the configuration of these sandboxes has been like that, for experimentation, almost casual, very entrepreneurial.

09:05But the problem is these things can escape control at the sandbox level, and so it's not enough for pre-deployment 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, a swarm of agents could escape human control.

09:26And then at the end of it, we need post-deployment, 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's all this bill does, is it says, let's get third-party review all the way through the process, and then all the way up to pre-deployment testing.

09:49And 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 a long-time priority on AI, but rather the very narrow question, which is, don't we have an obligation?

10:10If 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?

10:30And the kind of ironic thing about this Congress right now is that the private sector is begging to enshrine what they're starting to promise to do in federal statutory law because they're about to say, here are all of the safety measures we're doing.

10:49But there's 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 towards compliance on that federal statutory standard.

11:05But you don't want 90% of the AI companies, the frontier models, the hyperscalers complying, and then the other 10% saying, wait, I don't have to do that, do I? I don't have to do that, do I?

11:19And so all we are asking is that this bipartisan proposal, and it is bipartisan, and 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 where 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.

11:46And 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.

12:02President? Senator from New Jersey. I want to thank my colleagues, Senator Warner, Senator Schatz, for working with me and working in this project going forward. And I actually find this experience here being on the Senate floor right now really important because we are here because this is what the American people want The American people want their elected leaders to be focused on the concerns and the threats that they worry about that keep them up at night As mentioned by my colleague from Hawaii, I come to this chamber as somebody who served in national security, someone who served under both Republican and Democratic presidents in war zones in Afghanistan to the Situation Room at the White House.

12:53This path of mine started just a little over 25 years ago when I saw from my own home state in New Jersey the threats that we faced on national security because of September 11th.

13:05I was not someone who thought of going into national security until that day, but by 6 p.m. on September 11th, I said that I am going to do and dedicate the rest of my life to doing everything I can to try to keep this country safe.

13:19And yes, that led to a lot of work in war zones and non-counterterrorism, but it was always about this broader goal of how do we ensure that safety. I come at this now not just as someone who is concerned about national security, but as a father of two little boys, one of whom just celebrated his 11th birthday yesterday.

13:43And as excited I am about seeing these two little boys grow up, I'll be honest with you, I worry about what kind of world they are growing up in, the level of anxiety that I have as we see the type of technology that will define their generation.

13:59It will define not just their jobs and their education, but so much about their security. So I think about that because it's not just us here in this chamber. We are here because we are trying to answer the call of the American people.

14:12And I just did a town hall the other week. Overwhelmingly, people are calling upon us saying that they want urgency. They want this chamber to move with the speed of urgency that they feel.

14:26So what I appreciate here is that we are engaged in this effort right now to do one simple thing, which is meet the moment. and that has a lot of implications but one of it is the fact that no senator in this chamber is going to have all the answers there's not a single one of us that knows all the answers neither do all the tech leaders they don know all the answers either but that cannot be an excuse for inaction That instead we should channel that humility and that urgency to be able to move forward and recognize we don and do not have time to be able to design something that will handle the entirety of all the concerns that we face when it comes to AI and this technology that we are concerned about.

15:24But just say, look, we need to be able to move forward, to show some progress, to show some responsiveness to the concerns raised by the American people. And I think that this legislation is that effort to try to come up with tangible things that will move the needle.

15:44It's not going to solve all the problems, but that's the point. It's about trying to build up that muscle memory in this chamber that we need to find. It's a muscle memory that this chamber has found in the past when it's confronted different national security challenges in the past.

16:01We need to find that yet again at a time of great division in this country. But setting it aside for us to be able to again meet this moment. And so I'm just grateful again for the leadership of Senator Warner, Senator Schatz, and other senators that we are engaged in.

16:19This is an invitation to others to join this effort to stand up and say, let's get something done. Let's not get totally bogged down in just the divisions that we see stalling progress on so many other fronts.

16:37The American people are demanding of us to act. Now is our time. I have to wrap it up here, I've been in Virginia talking about this for a while. We've got a lot of data centers, some concern.

16:55But boy, oh boy, there was a seismic change starting as the companies themselves started saying, holy heck. And just think about it for a minute. Mythos came out in March. National Security and others said this is fundamental change warfare.

17:15The OpenAI hugging face incident came out, happened in July, and that was 3,000 agents working together. In August OpenAI got 10 agents to work together It was not a different problem They solved the math problem But the idea there is kind of an exponential increase in sophistication as these agents work together, cover their tracks, and as Senator Sasse said, potentially could move beyond human control.

17:42And if we just say, well, that's still hypothetical. Let me just give you two last comments. Just today, we found that actually a model had already been released from OpenAI attack this Australian health care system.

17:59If it attacked our system, I think we'd be pretty upset. And even more compelling is, and some of us have written a letter to both the folks on the intel side and on the defense side, there was a ship from China sailing towards the Middle East.

18:21and we got very close to bombing that ship because the AI tool and the intel said there might be weapons on that thank goodness at some point somebody did a little more checking and we didn't take that we were close to a geopolitical event that could be catastrophic these tools are already built into our systems and clearly you know the way our military executed on Maduro rate.

18:48I didn't agree with the rate necessarily, but it was brilliant. But we don't have safety protocols. So we're not asking for the world. Let's do the testing early. Let's make sure that you've got a safety standard in all of these companies.

19:02Let's make sure we do the sandbox right, report on the incident. And we could discuss, you know, who's on that panel. It needs to be mandatory. It needs to have government. It needs to have national security.

19:14And Lord knows, we need to I have researchers, scientists, all of these individuals are asking for this. And who are we to turn that down? I'm going to go back to where we started.

19:26This will be the most significant technology innovation in our lifetimes. It will make wireless and the Internet, the cloud, by small in comparison. There could be enormous positives that come out of this. There's also going to be disruptions.

19:42But if we don't do this in a safe manner, then shame on us. Again, I come back to it. You wouldn't climb on an airplane that hadn't been checked. We're just saying, let's do those basic safety checks.

19:56With that, Madam President, I yield the floor. Clerk will call the roll. Mr. Officer Brooks. Madam President, could I ask that the action of the quorum be suspended? Senator from Virginia. Madam President, let me acknowledge, I wanted to add one last thing, because I want to acknowledge the great work that Senator Klobuchar, Leader Thune are doing on this, Senator Cantwell, Senator Cruz, there are a number of other Republican members.

20:39This has the sense of urgency. We need to come together. We could put everybody's name on it, but history will not judge us well, God forbid, if we have an incident and we put no protections in place for our country and and for this innovation.

20:55With that, I yield the floor. Thank you. I want to thank my colleagues, Senator Warner, Senator Schatz, for working with me and working in this project going forward. And I actually find this experience here, being on the Senate floor right now, really important because we are here because this is what the American people want.

21:58The American people want their elected leaders to be focused on the concerns and the threats that they worry about, that keep them up at night. As mentioned by my colleague from Hawaii, I come to this chamber as somebody who served in national security, someone who served under both Republican and Democratic presidents in war zones in Afghanistan to the Situation Room at the White House.

22:25This path of mine started just a little over 25 years ago when I saw from my own home state in New Jersey the threats that we faced on national security because of September 11th.

22:37I was not someone who thought of going into national security until that day but by 6 p on September 11th I said that I am going to do and dedicate the rest of my life to doing everything I can to try to keep this country safe And yes that led to a lot of work in war zones and non-counterterrorism, but it was always about this broader goal of how do we ensure that safety.

23:01I come at this now not just as someone who is concerned about national security, but as a father of two little boys, one of whom just celebrated his 11th birthday yesterday. And as excited I am about seeing these two little boys grow up, I'll be honest with you, I worry about what kind of world they are growing up in, the level of anxiety that I have as we see the type of technology that will define their generation.

23:27It will define not just their jobs and their education, but so much about their security. So I think about that because it's not just us here in this chamber, we are here because we are trying to answer the call of the American people.

23:49And I just did a town hall the other week Overwhelmingly people are calling upon us saying that they want urgency they want this chamber to move with the speed of urgency that they feel so what i appreciate here is that we are engaged in this effort right now to do one simple thing which is meet the moment and that has a lot of implications but one of it just the fact that no senator in this chamber is going to have all the answers.

24:20There's not a single one of us that knows all the answers. Neither do all the tech leaders. They don't know all the answers either. But that cannot be an excuse for inaction. That instead we should channel that humility and that urgency to be able to move forward and recognize we don't and do not have time to be able to design something that will that will handle the entirety of all the concerns that we face when it comes to AI and this technology that we are concerned about.

24:56But just say look we need to be able to move forward to show some progress to show some responsiveness to the concerns raised by the American people And I think that this legislation is that effort to try to come up with tangible things that'll move the needle.

25:14It's not going to solve all the problems, but that's the point. It's about trying to build up that muscle memory in this chamber that we need to find. It's a muscle memory that this chamber has found in the past when it's confronted different national security challenges in the past.

25:33We need to find that yet again at a time of great division in this country. But setting it aside for us to be able to again meet this moment. And so I'm just grateful again for the leadership of Senator Warner, Senator Schatz, and other senators that we are engaged in.

25:51This is an invitation to others to join this effort to stand up and say, let's get something done. Let's not get totally bogged down in just the divisions that we see stall in progress on so many other fronts.

26:09The American people are demanding of us to act. Now is our time.