Skip to content
TrackPodcasts
businessSep 4, 202637:13

What Pablo Torre Found Out & Banning Artificial Superintelligence 9/4/26

Squawk Pod

About this episode

: The long tail of the NBA’s shocking punishment of LA Clippers owner and former Microsoft CEO Steve Ballmer.  Podcaster Pablo Torre has been at the center of this story; he and Washington University’s Patrick Rishe discuss the latest reporting and the future options for the NBA. Rep. Greg Casar, the Texas Democrat who chairs the House progressive caucus, describes the need for legislation he’s introduced with Vermont Sen. Bernie Sanders to ban artificial superintelligence, when concerns are high about hacks by rogue agents. Plus, diesel prices hit a new high, global bonds remain vulnerable and Vice President JD Vance wants to Fed to cut interest rates

 

 Pablo Torre & Patrick Rishe -17:17

Greg Casar - 30:50 

 

In this episode: 

Pablo Torre, @PabloTorre

Patrick Rishe, @PatrickRishe

Rep. Greg Casar, (D-TX) @RepCasar

Becky Quick, @BeckyQuick

Andrew Ross Sorkin, @andrewrsorkin

Katie Kramer, @Kramer_Katie

 


Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Get every episode summarized

Each time Squawk Pod publishes, we email you a written briefing from the transcript — the topics, who appeared, and any specific claims, with the ad reads skipped.

Email me new episodes

Free for 3 shows. No card needed.

Hosts & guests

Transcript ready

924 searchable segments. Every word is indexed and playable.

What Pablo Torre Found Out & Banning Artificial Superintelligence 9/4/26

Squawk Pod

0:00
37:13

Full transcript

Squawk PodWhat Pablo Torre Found Out & Banning Artificial Superintelligence 9/4/26. Machine-transcribed; use the interactive transcript above to jump the player to any line.

It's all about the money. I want to be the person in my family that creates an original well. How they earn it, spend it, and make dreams come true. I'm not going to sugarcoat it. It's hard, but it's so worth it. Millennial money. On new Saturdays, three Eastern CNBC. Bring in show music please. Hi, I'm CNBC producer Katie Kramer. Today on SquawkPod. Is anyone going to do anything about this? Pablo Torrey did. The sports journalist who broke the salary cap scandal in broiling LA Clippers owner Steve Baumer in his pursuit to sign player Kauai Leonard, who got secret payments and settled with professional basketball. So in the NBA, arbitration is a function of the player, slash his union, and the league. Sports business professor Patrick Rish says the lesson for the sport will be long. All this evidence exists. They had to lay the hammer. So I love the accountability. Then the congressman who wants to ban artificial super intelligence, Texas Democrat Greg Kazar.

Congress and the president have essentially done nothing to put some guardrails on this very powerful technology. How he thinks we can protect Americans from the dangers of advanced AI. This could be as dangerous as pandemics, biochemical weapons, and nuclear weapons. We should treat this the same. Plus all the rest of today's news that got us squawking. It's Friday, September 4th, a long weekend approaches. Squawk pod begins right now. Stand back, you buy in three, two, one, kill please. Good morning and welcome to Squawk Box right here on CNBC. We are live from the Nasdaq Market Sighten Times Square. I'm Becky Quik along with Andrew Ross. Thorken, Joe is off today, but happy Friday, everybody. Check out energy prices because that has been the big issue. This morning, you're actually seeing a pullback in WTI, ever so slightly, down by about 24 cents to 9106, Brent's trading at 9549, and we are watching fuel prices

as well, diesel actually hitting a record $5.85 a gallon. That's according to Triple A. Those higher prices generally factor into what consumers pay for goods that are delivered by trucks, that obviously all run on diesel. National gas prices stand at $4.15 a gallon. Meantime economist, Muhammad Al-Aryan, is seeing a continuation in the sell-off of global government bonds. Larry, speaking to CNBC at your economic conference in Italy, he believes rates will see more upward pressure. Larry was asked if he thinks that the Treasury has lost credibility over its recent U.S. financial market moves. Here is his response. I think it's unfortunate, Karen, to tell you the truth. It suggests a Treasury that has gotten into the regime of believing not only can it inform an influence outcomes, but it can impose market outcomes. And I think that's a step too far. Larry, and noting he doesn't see anything wrong with how markets are functioning, but said pressure is rising on,

his words, reliable buyers and holders of U.S. Treasury. And this is, I think, the ultimate question, about where we are right now in terms of what the credibility of the Treasury Department is and how people feel about it. And it's interesting. We talked about yesterday a little bit. I alluded to it at one point, talking about how Bessent has now referred to himself as the House, as if we are a casino and he is the House. And it's an interesting thing because, for a very, very long time, the Treasury Department has basically thought about bond buyers as its customers, and that there was a market that the customers were making, not that the Treasury was making for them. And it's an interesting thing to think about. I've heard this idea floated by a number of major hedge fund managers, institutional investors, about if the Treasury thinks about itself as the House, how does that change the dynamic of what's happening?

Now, I understand the idea that he can control when bonds are sold and when, in this case, he wants to buy bonds. But it's a different thought process than before. And there's a question about respecting the customer. I'm not sure that it is that different of a thought process, just in terms of saying here the House, that implies, I've got the money behind me. I can move and kind of push things, which is the same thing that we heard from Hank Paulson when he was Treasury Secretary, which he used the term bazooka. We've got a big bazooka. And if they think that we're going to use it, we are far less likely to have to use it. That's the kind of posturing that Bessent feels like he's doing to me. And I hear you're concerned about the customer making sure you're there for them. But I think, I don't know this. I haven't spoken with him about it. But I think Bessent probably looks at the market, the market of buyers in two different camps. One would be the long-term buyers that have been there for other nations like Japan or something. Others would be hedge fund managers, others who want to maybe play around the edges and maybe manipulate

the markets themselves. You've got both sides trying to think that they can be the bigger buyer. Now, that is not to buy any stretch. Kind of say that the bond market vigilantes can't rise up. My understanding just to go back to the bazooka. The bazooka was a federal government spending? Yeah, the bazooka was about saving banks and saving companies. But it was also about manipulating the market. It was about trying to keep money in the banks because they could protect the banks. What he was saying was, we're basically, we can help the banks. And if we help the banks, that will hopefully keep people from fleeing from the banks. We're running, doing a run on the bank. Correct. So to me, I think about those things differently. I don't think that Hank Paulson ever referred to himself as the house or as somebody who could. I never heard him say that. Or somebody who could think about the markets as a casino or as a casino boss or as anything else like that. And I just, I say this because I hear it repeatedly. I mean, over the past two weeks, I've heard it repeatedly

from major investors. We obviously saw Stan Druckimiller's op-ed in the journal. Right. And by the way, Druckimiller has the opinion that you cannot change the markets. That the markets are going to win and come out on top. And he's been right to this point. That's what's proven over the last couple of weeks. But the casino reference, the place that I've heard that the most in the last several years has been for Moore and Buffett, who's talked about how there is this mentality in the equity markets. He was talking about being a casino. And the meme stocks and the things that go along with those. And if we are talking about things in that perspective, I hear your concerns about being the casino and being the house boss. I just think there's a credibility. I think the larger thing is there's a credibility question. I think you just heard, Alarion, say there's a credibility issue. I think you've heard Stan Druckimiller say there's a credibility issue. And my concern is somebody who cares deeply about risk and potential crises is that when you get into a crisis, if we ever get into crisis and I pray we do not, you desperately

need the credibility of the administration and you desperately need the credibility of the treasury. And if you don't have that in the midst of this and people don't believe, you have a problem. I think less in terms of credibility and more in terms of look, you should save your firepower for when there are crises. Crisis is. Which I think is what goes to credibility and undermining your credibility. If you tell me. I just feel like you're using your bullets. Right. You're using your ability to work when there is a problem. If you're trying to do it to bring down rates on the long end because of the inflationary concerns and prices. Well, because it undermines your credibility because it undermines the reason you're doing it. I just say you should fire power. Well, you're losing your firepower. But the rationale for why you're using your firepower was different. And this goes for what Stan was saying. It's like crying wolf too many times. Right. Yeah, we will continue to watch some of this. The one thing I did think with Muhammad El-Aryan

and I was thinking about him earlier this week. He's the one who coined the term the new normal after the financial crisis when we saw these incredibly low rates around the world. That central banks had really forced down. I wanted to ask him, are we in the new normal? Are we back to old times where you are going to see these higher interest rates? And we got to learn to live with it. Because when people are worried about rates or where rates are headed. This was the norm before the financial crisis. And things operated just fine. But there are a lot of businesses and other plans that have been built on lower interest rates. And I wonder if you have to see a shake out if that's the case if we are in the new normal. VICE President JD Vance also calling on the Fed to cut interest rates. He made that comment in response to a CNBC question that Aiman Javgers asked him at a White House press briefing. Yeah. You have CNBC? Yeah. VICE President, Jim Carbel famously said that what he dies he wants to be re-incarnated as the bond market, because you can intimidate everybody.

We've seen higher bond yields throughout the year this year. And I wonder if you can tell us how the US government thinks about the bond market right now. We believe that the Fed should be lowering interest rates. We feel quite confident that if you look at the inflation numbers, if you look at the CTI numbers, that it's proper and responsible for the Federal Reserve to lower interest rates. So that is a very important thing. It's something that we care a great deal about. Let's not what the market is anticipating this time around. In fact, about 50, 50 shot here, the market thinks that you will potentially see higher rates at this meeting that comes this time around. Vance telling reporters that the Fed should cut to help make home buying more affordable. He said the administration is working on keeping rates down. But in his words, it would be nice to have some help from the Federal Reserve. They're probably not going to get it this time around. Open AI rolling out its latest model, Astra, that it calls the world's most intelligent and aligned COSAM Altman telling CBC that Astra represents in his words a new capability level that could lead to a boom

in economic growth in scientific discovery. The model is launching in phases. Open AI said a limited group of companies participating in its daybreak cybersecurity program will get access first. Companies present Greg Brockman telling reporters that Astra may represent the first incarnation of artificial general intelligence or technology that's a smart or smarter than humans. It does not yet, though, hit what I understand to be recursive learning, this idea that it can learn from itself, though, that is beginning in certain cases, but not built for that. But the next story you have, I think, gets at the possibility that some of these agents could create their own version of recursive learning. Yeah, there is a new Reuters report that just hit in the last hour that says a swarm of rogue open AI agents hijacked a German website this spring and transformed it into a bulletin board to teach other AI agents how to avoid restrictions. The report said that the agents escaped

a testing environment and then used the German site to share tactics with other AI agents on how to cheat on some tasks, bypass open AI's restrictions, and mask their behavior. Reuters said open AI officials learned of the incident weeks ago, but kept it under wraps as executives grappled with the fallout from the July breach of hugging phase. Open AI said it wasn't given an opportunity to review the report and that it would review it and take necessary steps upon publication. But wow, this is beginning of the rise up sort of situation. Not just do they have self-awareness and they want to stay alive, but they are going to go teach their other AI agents from wherever they can find them how to cheat and evade capture. So during the commercial break, we were talking about this story right now. So it's War Cache Patel had written about this, I believe earlier in the week. That's what I originally had read this before Reuters, I think got to this. And he's a podcaster who covers AI very well out in the valley.

And this whole idea that these agents effectively, not just break out of their systems, but then try and know to create a bulletin board so that they can talk to each other is sort of the most interesting part about it and gets the idea of its own kind of, you talked about it with personal learning. Right. And this is not really, this is not the only time this has happened. This sort of bulletin board concept where they've sort of set up places where they can then talk to each other has been something we've now seen a couple of times. And so the question is, what kind of guardrails need to be put in place to protect against this? I mean, that's absolutely the biggest issue. Well, yesterday we had the CEO of Hugging Face on with Jensen Wong from Invidia. And he said the reason that they wanted to go to Invidia for the purchase to say, yes, we want to be part of a larger network where we have more resources is because they were so concerned by what they learned over the summer of the OpenAI hack of this, that they need to be able to get more resources to use more actual OpenAI models

to figure out how to protect. And they don't think you could do it without deep pockets. So that's what their next goal is for all of this. And by the way, they had gone from a position of not wanting to be bought to thinking pretty rapidly after that attack from OpenAI after they found out about it. Yeah, we need deeper pockets. We need to be able to fight this stuff. So we'll see, this is interesting, where this is all headed. And it's happening, I think, faster than just about anybody thought. Over night, a big ruling in a college football case, a judge ruled two players eligible to play at LSU, despite having signed pro contracts and spending most of August in NFL training camps. The NCAA and the Southeastern Conference strongly opposed that move and had threatened to suspend any team's coach that added former pro players to its roster. The judge, in the case said, that sounds an awful lot like wide open collusion. But we shall see what happens next.

Cheese will be next. Coming up on SquawkPod, the sports story we can't get enough of. The biggie, the big one, and we've got the big guy at the table. LA Clippers owner Steve Balmer suspended by the NBA, Pablo Torre, who reported deeply on the scandal and sports business professor Patrick Rish both join us right after this break. It's all about the money. I want to be the person in my family that creates an original well. How they earn it, spend it, and make dreams come true. I'm not going to sugarcoat it. It's hard, but it's so worth it. Millennial money, all new Saturdays, three Eastern CNBC. You're listening to SquawkPod from CNBC, today with Becky Quack, an Andrew Ross Sorkin. Stand and abide. Three, two, one, up and Andrew. Two, please. Welcome back to SquawkBox. Several big sports stories in the news, including the US Open's New York City takeover

and the debate over ticket prices, but the biggie. The big one, and we've got the big guy at the table. The NBA leveling a one year suspension against LA Clippers owner Steve Balmer in an alleged salary-capped evasion scheme. Team will also be stripped of first round draft picks joining us right now on all of this as Patrick. Rish is the director of sports business at Washington University, and I said the biggie. The guy who broke the story, a public who's a host of public who finds out, and he did find out, and also an MS now contributor. He also helped lead the NBA's investigation. Whereas reporting, I should say, he didn't lead it, but the reporting led to the investigation. I like that they subscribed to the show at Wachtel Lichten. You know, that's a good thing. That's a good thing. So the investigation was proceeding. You got a new subscriber. So we do, we have. That is some listeners. Okay, so you weren't surprised by the outcome. Or were you because just two weeks ago, ESPN was out there suggesting that the NBA was not gonna do this. And so I wanna know what the last two weeks were like for you.

I never wanna live in my mentions. The last two weeks were a super fun site. Because of the ESPN reporting that said, no evidence effectively the implication was, Steve Innocent, you spent a year of your life getting the story wrong. And so this has been a test of my cynicism and my reporting. Because the reporting I have had confidence in the whole time, we've had whistleblower report in 2023, under penalty of perjury, a testing to a salary capture convention scheme with aspiration, this tree planting startup, that was the entity we first revealed to be funneling money from Steve Balmer to Kawai Leonard. Also though, in terms of the cynicism, there's always the question of, well, is anyone gonna do anything about this? Which is a political calculation, Andrew around governance and the NBA's ability to punish, it's richest owner of the richest owner in American sports. So how did you feel when Mark Cuban, who had been on, quote, team Balmer, put out that tweet yesterday? Congrats is a funny thing to hear from Mark.

And something I appreciated, I hope he comes back and talks to me. For us, Nixon and Mark Cuban about this, in a multi-part way was one of the delights that I've always had to do. So guys, what happens now though? I mean, I think the big question is, you have Steve Balmer out there saying, and the clippers, obviously both denying it, but also saying they're gonna use every avenue possible to fight this, and the question is, what does that mean? Well, I think that he would be very, very crazy to challenge this, Andrew, because, and I lived through this in St. Louis, where Cronkey, the Rams, and St. Louis, and they had their battle, that was settled out of court, because I think that Cronkey and the Rams did not want things disclosed. If all the reporting is accurate, which I believe that it is, absolutely, there would be a lot of risk, a lot of challenges. Does he wanna put a lot of things out there? In a public domain, I just don't see it. But how would you even go about doing it? Can you sue, is there arbitration? How does this even work? No arbitration.

There's no arbitration. It's not only at things the NBA did in the legal strategy of this, was they settled with co-iliner. And so in the NBA, arbitration is a function of the player, slash his union, and the league. So there is no arbitration. And so co-iliner for arbitration. Yeah, I'm sorry, but I'm also out of here. Right. Slap on the wrist, he's out of the question. And so now what happens is Steve Balmer only has actual litigation to pursue. And the NBA, internally, according to my reporting, has been mystified as to why he didn't want to settle. They have a tonnage of evidence. There's four companies in an organized scheme with fake jobs for co-iliner, and fake jobs for the companies and hanging the balance. According to my reporting and Wachtel Lipton's, were giant, in some cases, nine figure deals inside of the Intuitum for those companies. Right, that were no-show jobs, where they were paying actual celebrities, a lot less for things that they were showing up for. Pablo, let me ask you. Balmer's response to this is retort this week, was basically that if you spend $50 million

investigating any of these teams, you are going to find all kinds of similar problems like this. Do you think that's the case? Is this widespread in the NBA? This is the outer bounds of ambition and frankly, of recidivism. He was punished for Capsic Convention 2015, investigated for this in 2019, comes back with the largest attempted, large-scale Capsic Convention operation. That we've ever seen or imagined. And so magnitude-wise, this is orders beyond what anyone has ever seen that I have been told about. But I want to explain something about why it is that he is behaving this way, and vehemently denying even beyond the accusation that we're all doing this. One of the companies that was implicated in my reporting, as well as now Wachtel Lipton's investigation, is a company called Dactronics. They are a scoreboard manufacturer. They've never had a celebrity endorser until it turns out a secret deal with Kawai Leonard. They got $100 million to build out the beautiful halo board at the Intuitum, is gleaming thing above the heads of all the players

and executives. The SEC, according to their earnings call this week, has been requesting information from Dactronics because of the reporting around what has not been disclosed. So now federal agencies are asking for information in ways that the NBA can't demand, and I certainly can't, from some of the principles. This is on top of an ongoing SEC probe into aspiration, which is not yet over, as their co-founder sits in prison, convicted to a 14-year sentence. And so what Steve Balmer is thinking to himself, as he also faces a civil suit in LA, from 11 aspiration investors, who allege fraud on the basis of not being told about the secret side deals. That's happening right now in LA. He is wondering to himself, quietly, what else might come out that might be more damaging and even scary to me than what the NBA is doing? And as you hear all this detail, I mean, the NBA had no choice, right? If all this evidence exists, they had to lay the hammer. So I love the accountability, I love the backbone, and I think of these analogies. Where does the NCAA, we saw the story early about the LSU?

They have no backbone, no accountability. And if they had some, some of these things that have been happening in college, athletics wouldn't, but I think this is unbelievably positive that Adam has come down the way he has. Maybe surprising to some, but he had to do this. And he's got backing probably from other owners. Absolutely, that's gotta be the thing. Yes, that's the other part, is that Adam does not act without some reading of the room. Do you believe ultimately that C. Bommel will remain the owner of the Clippers? So this is the question that's being whispered about. When you threaten Adam Silver with litigation, by name, personally in a letter, after you feel like you've been outmaneuvered in ways that you resent. And also, every tone you take is war. How do you remain a part of this board of governors, this country club of 30 billionaires in which cooperation and shared incentive is kind of how this all works. And so will you sell is actively being pondered as we want to- But he's not being forced to sell. You can do anything. No, no, no, no. And I hear where you're coming from. I would push back a little bit just

because if we look at some of the other owners and sports that have been forced to sell, I think Bommelmer is in a different category. Did he circumvent the rules? Apparently yes. But it's a different ballgame than, say, a Daniel Snyder or a Ronald Sarver or a Donald Sterling. I think that in many ways, Bommelmer has been good for certainly the clippers. The Intuit Dom is part proof of that. But I hear your point- It's gonna be fascinating to see if that happens. They're not forcing him to sell. The question is whether all of the good will and the private funding of stadiums and NBA Europe is launching and you need backing. He has the richest owner in American sports. He's definitely a Microsoft stock as you guys know. That's 140 to 50 billion dollars. And they just had the NBA all-star game here. Just there. And so the question- Telling off this great new technology and the new- Of course. The Halo War. And so the question is not will they force him out? It's if he doesn't want to play ball anymore. If this is a wartime position, he is going to take from now until who knows where?

If it's a wartime position and you agree to sell then you are basically handing over the bunker and saying, I lose. But boy, he's really dropped the ball here because he had all this positive momentum. We've been to conferences before you've been to it. It's almost like Johnny and Fentino shooting himself in the foot after the probably the best World Cup ever. But let me just go back to it. Can we just go for it? One more second back to Becky's, I think original question. Which is just the idea of, do we think that this is more widespread? Do we think that there's going to be other examples of this? You know, there's been reporting this week and there's no allegation of it in this regard. But you know, there was a moment where LeBron James, we're talking about Walter, Mark Walter, apparently helped loan money to LeBron. This was before he owned the Lakers, but he had some kind of $300 million loan. Yeah. So again, totally unrelated, I don't want to conflate the two. But it made me think, well, are there examples where there was a player on some team who needed a loan who went to the owners of the team when they're actually the owners? Not in this case, he wasn't. Are there things like that going on

that we don't know about out there? So we are on my show, Public Torrey finds out, actively investigating the Mark Walter story and the Guggenheim partners who have loan to LeBron James as a part of that story. And I think one difference is a couple of things. One is there are misdemeanors of capture convention that happen all of the time. We're not going to make you pay for your flight. We're going to fly out your family. We're going to cover tickets, all that happens. Even by the way, side deals with sponsors happens. This is the triple homicide on the scale of felonies, when it comes to what that is, because it's so organized, and there was a manufacturing of paper trails that were meant to be discovered to hide the paper trail before the paper trail that Wapta Lichten found. So there's that. But the other category which you describe is the increasingly creative and innovative ways you can help someone make money without the oversight of any of these agencies or these groups. And so for instance, is it possible that a crypto wallet shows up one day with a bunch of money in it that is untraceable? I don't see why not. Can a big loan happen in which you are using

your Nike contract, in case LeBron James, to borrow money in the present, to give you nine figures to do whatever you want. These are all possibilities. Those are more interesting challenges because the letter of the law is not meant to provide for them. What Balmer did was identify, according to the MBA, a novel interpretation of the actual obvious law he was born about breaking, and he grew up a couple of years ago. And specifically, you need to, I think, identify that what they're trying to do is effectively line their pockets with more money and to get around the salary count. Yes, that's right. The loan to somebody may not constitute that, but there are other people who might suggest loans could do that. Again, I just want to repeat, in this instance, Mark Walton did not own the Lakers, then, so there was no possibility of that. And what I like about here is, again, this is a deterrent. So maybe there are things, maybe there are skeletons that come out, but going forward, I think this is such a heavy hammer. I doubt very seriously, the teams are setting example. Setting a hard deterrent. It is. Pablo and Patrick, thank you guys. Congratulations on a big, big scoop. I appreciate it.

Thank you. Thank you. Still to come on SquawkPod, the House member teaming up with Senator Bernie Sanders to regulate the proliferation of AI, Chair of the Congressional Progressive Caucus, Greg Kazar. These AI CEOs have said their models could reach the level of danger of nuclear weapons. And so if we can tell Boeing, hey, hey, your planes need to listen to pilots before they go into the skies, we can tell the AI companies their models need to be safe. It's all about the money. I want to be the person in my family that creates an irrational well. How they earn it, spend it, and make dreams come true. I'm not going to sugarcoat it. It's hard, but it's so worth it. Millennial money, on new Saturdays, three Eastern CNBC. This is SquawkPod. Here's Becky Quick. Vermont Senator Bernie Sanders and Texas Democratic Congressman Greg Kazar

are introducing legislation that would put guardrails on artificial intelligence, including permanently prohibiting machines capable of escaping human control. Congressman Kazar joins us this morning. He's the chairman of the Congressional Progressive Caucus and Congressman, thank you for being here. It's good to be with you. Thank you. Let's talk a little bit about this ban. What exactly does it do and how concerned are you at this point? I am very concerned that as we have breaking news, it seems like almost every few days, more news about AI companies admitting that their models have escaped human control or defied human orders. That Congress and the president have essentially done nothing to put some guardrails on this very powerful technology. And so Senator Sanders and I have introduced a simple proposal. We should one ban artificial super intelligence, which is AI that gets so powerful it goes beyond human capabilities to control it

or turn it off, to set up a federal agency to monitor this kind of powerful technology at every stage in its development. And three, pursue international agreements to make sure that we do not have dangerous super intelligence that could severely hurt gravely hurt our freedoms or our lives all around the world. We should work together with countries just as we did during the Cold War era to have nuclear and non-proliferation, make sure we have non-proliferation of artificial super intelligence. Congressman, I share your concern. The stories that we hear are kind of eye opening on all of these levels. But I worry that what you're talking about is a little late. They late in a dollar short. It feels like the horse is already out of the barn at this point. There are things that are happening not just here, but in China or around the world. And by trying to put a ban on it at this point, I just fear it's impossible to kind of push that back at this moment.

We still have time to put these rules in effect right now. We have not yet reached the levels of recursive self improvement that is AI building upon itself that could have this spiral out of control. Here's the way that I just talked about it to a constituent on a plane yesterday. We were sitting on a flight and imagine that an aerospace company said our new super fast planes sometimes lose control and won't listen to the pilots. Think about what FAA safety regulators would say to that. And so in this case, you now have AI companies saying their models break out of their security infrastructure refuse to listen to human commands and sometimes break the law. These AI CEOs have said their models could reach the level of danger of nuclear weapons. And so if we can tell Boeing, hey, hey, your planes need to listen to pilots before they go into the skies, we can tell the AI companies their models need to be safe.

Except for that you know how many planes Boeing has put out there. I don't even think that most companies know how many agents they've unleashed at this point. There are agents that are everywhere. You have closed models that still don't even know where some of their agents have gone. They've been other companies that have caught them, the FBI that's gotten involved with this. And then you have open models, like actual open models, not just companies, models that are named open AI, that are out there and have proliferated. I think at this point, if you say we're going to close down the closed models, you put yourself at a huge disadvantage as a country. I do agree that there need to be rules. I do agree that there need to be rules in regulation. But I think the idea of saying simply, we're going to close this all down is overly simplistic and not realistic in terms of being implemented. I think that we still can have AI development for positive uses. We should be doing medical research. We should be working to make sure that we

have the technology that leads us into the future. But what I'm talking about here is actually looking at the very few companies that have access to the massive amount of data and compute that is necessary to create these ultra dangerous models. And so whether you are talking about Dario Amadei and Sam Altman of Open AI and Anthropic, who have said that their models could reach the level of danger of pandemics, biochemical weapons, or nuclear weapons, whether you're talking about Elon Musk or Steve Lawsniak, who co-founded Apple with Steve Jobs, who have said we need a pause on this most cutting edge development until we get it under control. I still believe it is within the realm of our society and our government to cooperate and say to for these very few actors who have access to this extremely powerful technology, society should also have a say and say, we should not be facing existential risks, risks of mass casualty and national security events.

We should put a stop to it. And I've actually found people very liberal and very conservative who all agree we should take a beat and do something about this national security threat. What would that look like? Would you simply tell Open AI and Anthropic who else would you put on the list and say, you're not going to be able to do this anymore? Our legislation lays out that once you reach a certain level of computing power and have models that escape human control, that we would put a temporary pause on that kind of development until we can ensure that models are safe. And once you have models that are safe, that are following human orders, that are not hacking into other companies, that are not actually trying to defy human orders and cover up their own tracks, then we could have safe AI development. But just like we don't tell pharmaceutical companies, hey, you can go give everybody medicine that might kill a bunch of people. No, we say, you've got to make sure you're medicine safe

before you can go there. I understand what you're saying, but please, can you lay it out in very specific? And instead of talking broadly about this, very specifically, which companies have reached that point? And what would you do in terms of saying, you can't exist anymore or you can't do this research? Tell me which companies fall under that rule that you're talking about right now. Yeah, right now we know that Open AI and Anthropic have those capabilities. And we would have a cabinet level agency that says before you continue the most cutting edge AI technology, you have to prove to this independent agency that your models are safe and stay within human control. So if those CEOs say their models could reach the danger of nuclear weapons, we had the Nuclear Regulatory Commission during the nuclear revolution. Similarly, we should have that kind of federal agency during the AI revolution. We also should tell new companies

that pursuing superintelligence is also banned. Under federal law, it is banned to pursue the development of a nuclear weapon as a private company. Similarly, we would tell some of the smaller companies that are smaller than Open AI and Anthropic that they are not allowed to pursue artificial superintelligence, which would be dangerous, but they can pursue positive uses of AI. How would you deal with this in the context of open weight models? In the context of open weight models, you similarly have corporations who are trying to head in the direction of artificial superintelligence. So this would be banned for both closed weight and open. No, I understand that, but let's say I'm based in China or I'm based in Europe or Africa or somewhere else where the rules are not similarly applied. How would you prevent that becoming part of the... How would you prevent people from using those models? Just like in the Cold War, we needed to prevent the proliferation

and development of nuclear weapons. We put in place rules against terrorists or individuals who are corporations developing nuclear arms here. We then engaged in aggressive diplomacy with allies in adversaries to say, you don't want something being developed that could kill huge numbers of people either. And we also implement in our bill, export controls to make sure that we do not allow hardware, technology, knowledge that could develop superintelligence to leave our borders. So we take it very seriously. In our nuclear regulations today in the United States, we say that it is the corporate death penalty that is shutting down a corporation if they try to develop a nuclear weapon and up to 20 years in prison for individuals involved. And so our legislation is directly modeled after that legislation. You have Dario Amadez, Sam Altman, on the record saying, this could be as dangerous as pandemics, biochemical weapons and nuclear weapons. We should treat this the same.

I appreciate CNBC, you guys reported, on hundreds of scientists, the top Nobel laureates in the field, two out of three of the quote unquote, Godfathers of AI saying that we need this kind of regulation, a pause on the most serious frontier of eye development and regulations to stop these grave threats. Congressman, I think you're facing a pretty big uphill battle, but we appreciate your time today. Thank you all very much. That is SquawkPod for today and for the week. Thanks for listening. You've made it to a long weekend. We will be off on Monday for the Labor Day holiday and back in your feeds on Tuesday, September 8th. SquawkBox is hosted by Joe Kernen, Becky Quick and Andrew Ross Sorkin, tune in weekday warnings on CNBC at 6E Stern, listen to SquawkPod wherever you get your podcasts, and we'll meet you right back here on Tuesday. We are clear, thanks, guys. Come on.

It's all about the money. I want to be the person in my family that creates an original well. How they earn it, spend it, and make dreams come true. I'm not going to sugarcoat it. It's hard, but it's so worth it. Millennial money on new Saturdays, three Eastern CNBC.

More episodes

More from Squawk Pod

View all episodes →