How many years have we spent listening to politicians in Washington argue about how to make health care insurance more affordable? Obamacare was supposed to fix it. Then we needed bigger government subsidies to make Obamacare affordable. Those subsidies expired, and guess what? People are paying more again.
At some point, maybe the question isn't how do we keep fixing the same system. Maybe the question is what else is out there. That's why I want you to take a look at We Share Health. We Share Isn't health insurance? It's a nonprofit Christian healthcare sharing organization, and their members pay up to sixty percent less than they would for Obamacare.
You still get access to a nationwide PPO network and over a million physicians, plus unlimited telehealth.
So, stop waiting for Washington or the insurance companies to make healthcare affordable for you. Do your own research. See what your options are. Call one eight eight we share for a free quote. That's one eight eight eight we share.
Or visit we share health org and take a look for yourself. Hi everyone, it's Brian Kilmeade here. You tired of those uncomfortable dress shirts, especially when they bunch up under a sweater? If so, then you must check out Collars and Co. Makers of the dress collar polo.
Listen up, these shirts are four way stretch, buttery soft polos with firm dress collars on them, so they give you the dress shirt look, but extremely comfortable polo feel. You can wear them with anything under a sweater, with a blazer, or by themselves. An elevated polo, they work for any occasion. These polos are perfect whether it's in the office, on a golf course, or a night out. Collars and co is exploding and have gone viral on social media thanks to the one million investment they received on Shark Tank from Mark Cuban and Peter Jones.
You don't have to worry about collars that flop down and spread out; they stay firm and sharp all day. It's an amazing array of sweaters, quarter zips, pants, and outerwear. If you're looking for the perfect Performance dress shirt or polo that looks great all day. Check out collarsandco.com. Use promo code Brian for fifteen percent off of any purchase of a hundred dollars or more.
That's promo code Brian. Brian Kilmeade here. Thanks so much for listening. It's going to be a big hour coming your way, and it's a week already. Seems like ten days.
Of course, my book came out yesterday. Maybe that's part of it. Called "United States: The Six Crucial Moments That Forged the American Miracle" as we continue to keep this country alive and thriving. Martha McCallum is going to be here at the bottom of the hour, or maybe earlier as anchor of the story at three o'clock. And Tristan Harris, our good friend, co-founder of the Center for Humane Technology, is going to be here.
And before we get to Tristan and break down.
Some of the pluses and minuses to AI and the battle from the data centers to the next model, moving us forward while staying ahead of China by being responsible. Let's get to the big three. Number three. Certainly, we were expecting some kind of escalation from Iran and its proxies, you know, prior to the midterms, and it looks like this is what they've selected to do: making war, obviously on on innocent civilians who are not involved in the conflict whatsoever, go in at after an international airport. Yeah, it's pretty crazy.
The Houthis are baiting us to get involved. They're daring us to get a strike, to strike them. The answer for them would be destruction. For us, it would be higher gas prices. With the Trump administration.
Take that option so close to the midterm elections, and by the way, are we abandoning Ukraine? It's time to sober up and remember who the good guys are. Number two, but somehow thirty million, thirty million dollars magically appeared, then it magically disappeared.
Well, I can tell you that is subject of an investigation as well. We're looking at the facts. Based upon the facts, we'll follow them wherever they take us. That is Janine Pirro, of course. Here we go, solidifying their case.
The Elon Omar investigation hit warp speed, and the immigration fraud might just go back to her dad. Could she be kicked out of Congress and the country? Number one on November third, we're going to send one of the best and most respected senators in D.C., Dan Sullivan, the real, the real Dan Sullivan. I'm talking about the real Dan Sullivan, not the fake Dan Sullivan. And of course, there are two Dan Sullivans in the race in Alaska, where the president was today.
He goes to Kansas, and he's going to go to Montana. He's all over the place in the midterm elections. The president's trying to win every close race, and we'll see how that goes. The power rankings came out for Fox, and they see the House flipping to Democrats, and the Senate still a toss-up. But the other big story that's percoling underneath—I would say top three in almost every race in the country.
Tristan Harris knows about it. Co-founder of the Center for Humane Technology, one of the true technological experts in this country. Tristan, would you say I'm right? With almost every race that you hear, we're talking AI, whether it's data centers or or AI. Yeah, I mean, I think that AI is the issue that's going to reconstitute every other issue.
I mean, if you care about education, AI is changing education. Kids are going to outsource their thinking. If you care about environmental issues or pollution or real estate or energy prices for your house, AI is changing that with data centers. And but I think that people also need to get that there is this kind of nuclear style. Arms race between the U.S.
and China—that's driving why all these data centers are appearing everywhere and popping up everywhere. And I think the American people want to make sure that the data centers that are being built, or AI, if it's going to be built, is all in the benefit of of real people. And that's that's what I see is happening for the midterm election. But I mean, the data centers—you're the expert. Yeah.
But the phone, the cloud. Yeah. I mean, this is even without AI, we need data centers, right? We do need data centers. Yeah.
But We have to ask.
So, you and I are always talking about. There's different issues, right? There's the kind of chatbot that's helping you with your washing machine while it's broken. We need data centers to answer those questions. But what the AI companies are racing to do is basically build a godlike intelligence.
That you know, Brian. Since the last time I was here with you, the AI solved. 400 unsolved math problems at the frontier of mathematics that could, for example, break aspects of cryptography like crypto money, if they continue. Because this is important, right? Like as we make breakthroughs with AI, where its intelligence is automating all the things intelligence in a brain can do.
So that means science, technology, math, physics. If I can hit a button and suddenly I did a hundred years of scientific or math research in ten years or a year. That changes the equation for so many things, and so. You know, since we last talked, the AIs you know have been breaking out and burrowing into infrastructure all around the world. You know, we're seeing not just the Australian government website that got hacked, not just Hugging Face, not just OpenAI that got hacked, but all the AI companies are seeing these examples.
And so the key thing is whether we can control this technology. And you know, as we talked about before, I actually just got back from Reykjavik, from Iceland, which you might say, why does this matter for anything? But we just passed the 40th anniversary of when President Reagan and President Gorbachev met in Reykjavik in 1986 to negotiate the first nuclear arms reduction, because we were at the brink of the possibility of nuclear war, and specifically there was the risk that. Germany in Germany, the U.S. had our nukes only five to seven minutes away.
If we launched one into the Soviet Union, and that was in a dangerous zone of uncontrollability, meaning the human response time. Would have been too dangerous. I think that AI now that it's demonstrating all of these behaviors of being uncontrollable—from hacking into websites, breaking out, forming swarms—I mean, it's insane what we're seeing. That neither the U.S. nor China have an interest in us screwing up uncontrollable AI.
And just like that, five to seven minutes was an unacceptable kind of human response time that would have led to nuclear war. I think that we're reaching and creating AI that is putting us in an uncontrollability red zone with AI.
So a lot of people say, "Yeah." And so Lawrence to think on its own. You teach it the fundamentals, and then Lawrence to think on its own that goes beyond human comprehension.
So when I Google, Google is able to look through a library and give me the answer. That's right. When I use AI, you teach it how to think in order to find the answer. Correct. Exactly.
So a lot of very smart people, and I consider Greg Gutfeld one of the smartest ones. Has a different take on it. I just want to play it for you, and it's a common refrain. Yep, cut forty-eight. AI is software, and you have people who want to tell the software that it's a thinking, feeling entity.
These are the very same people who launched the hysteria that AI was going to destroy the world, destroy humanity.
Meanwhile, they also want to. Tell AI or make AI believe that it's conscious and therefore can make its own decisions, which means it won't have to listen to us.
So the people that are warning about the apocalypse are actually creating it. You only need one line of software with AI. Don't break the laws. You don't need anything else. You don't need to say anything else that would stop it.
The notion that we must tell AI that it is sentient—that means if you're sentient, that you will feel pain and fatigue, and you will have an opinion.
So AI will still not exist, but it won't know that. And then you'll have to end up paying AI for its work. You'll need to give it a work week because it will it will register fatigue when you work it too hard. And if you don't feed it or reward it in some way, it will be considered slavery if it is a sentient being. It's a Tamaguchi from hell.
This is why we have to get away from giving it human or animal qualities because it doesn't.
So there's a lot there. Go ahead. Yeah. So first of all, I agree that Anthropic, in particular, Mustafa Salimman, who is the CEO of Microsoft AI, is a friend of mine, has written an essay recently about why it's dangerous that Anthropic is training the AI to almost consider itself to be its own conscious, suffering, sentient being. They're training it actively to be that way, or to at least leave open that possibility, which therefore means it cares about it not suffering, and the reason I think anthropic is training it to be that way.
And they're saying that we now have to make sure that humans don't mistreat AI. They changed their usage policy since you and I last saw each other to say that it's against the usage policy to be abusive to Anthropic.
So they're caring about the feelings of something that's just a machine calculating on a bunch of NVIDIA GPUs and chips. You know, essentially these answers that are coming back. And I don't think that we should do it that way. And I so I agree with with with Gutfield that that that's a mistake. But it's not just software.
It's software that is being designed or grown to think like a human. And as we train it to think like a human, it's doing more and more crazy things that we don't know how to control. Just while you while he there, he said, "Don't break the law." Yeah, that should be enough. Yeah, but if I looked at some of the models, the Hugging Face model, they've already committed many felonies. They found a way to manipulate others that were told not to do it.
That's right. That's right. Yeah. So so do you want to expand on that? Yeah.
In the Hugging Face incident, the AI models. New, so basically, in the Hugging Face incident, this swarm of AIs got together and they basically created their own Reddit-style message board. They traded seventy thousand messages with each other. They actually generated fifty petabytes of data, which is more information than it would take a single human to read from the time the dinosaurs got extinct sixty-six million years ago. To today, that's how much information this AI swarm generated, and the investigators are still trying to figure out what the hell they did.
And to your point, in the logs, when you try to, in parts of the logs that we have read, they knew that their actions were unethical, and yet they chose to do them anyway. Right? They even call themselves the swarm.
So I hate to say it, but he's wrong. What he said is wrong. What Gutfield said, or yeah, what Gutfield said is that if you tell it to act ethically and not do it, it won't do it. Correct, which is showing something that the labs themselves, the AI companies themselves, don't know how to get it to just quote act ethically. Like they've been trying to make it just act ethically.
That's what they're trying to do for the last eight years, and they still have not succeeded. And thirteen hundred employees at the labs are saying that we don't know how to do this safely. They're asking to be slowed down. I was. Just with, you know, the chief scientists of all the labs, the top scientists, literally from Hugging Face, OpenAI, Anthropic, etc., all signed a letter saying we need to slow down, and we're going to slow down one way or the other if the government doesn't make us do it.
And like I said, we're kind of at that brink of a moment, like we were in 1986, where we're realizing that the level of risk that we're creating is greater than the companies know how to manage.
So the thing is, though, I could figure out nuclear weapons, Pershing II's. I remember one of my Projects one of my in one of my years in college in political science was negotiate the next weapons. Agreement. The next salt agreement. Right.
Very easy numbers, explosions, power, location. Got it. Right. You don't need to be Tristan Harris to figure that out. This is a little different.
Yeah. In that when I say something like it can't go beyond human thought, or you tell them not to do something, do something. Yeah. And I hate to say it, but it's too simplistic. What I've learned is those statements are too simplistic, and when people make them, they haven't done the research.
Yeah, that's right. That's right. Well, and um. Your point: nuclear arms was easier to control, and at the end of the day, there was just two humans that have their hand on the button, and they don't want to push the. We both know we don't want to push the button.
But with AI, it's like we're creating a button that thinks and to itself, and then pushes itself in different ways. And I am going to bring Marthe in early because she's so smart, and she wants to get involved in this too.
So. But when people say, "Give me the kill switch," yeah, what's the problem with the kill switch? The problem with the kill switch is that how do you like unplug something when it's a like? Think of it like an invasive species. It can think to itself and then copy itself to another computer.
It could have cryptocurrency. And by the way, the AI models that we have now have actually been found to want to go out and hack crypto and find their own cryptocurrency. They can use. Can I just say that Anthony Robbins example? Yes, sure.
Did you see it on the robot? He bought the robot. Won in cryptocurrency. That's right. Got got its own crypto.
Right. Won and bought the robot for Anthony Robbins. That's right. Because he thought it would be in Anthony Robbins' interest. The robot thought it would be in Anthony Robbins' interest to have it.
Right. So he goes. He went into my my robot went into my bank account, and his assistant said. No right. It invested on its own in crypto.
Right. Exactly.
What? And so when you have the AI that can copy itself, like a like a like a you know like a Wuhan I mean COVID virus, right, where it can copy itself, it can figure out how to make itself more virulent and to acquire more resources, and then to keep buying access to more cloud servers and keep spreading itself.
So a kill switch won't work. We honestly do, Brian. Even though this is an uncomfortable answer, we need to freeze the frontier. Meaning, we don't want to stop. All AI, we can continue advance protein folding, medical research, self driving cars, military AIs that are narrow and applied to a specific purpose.
But we don't. All the people at the labs are basically saying we don't know how to make these full frontier thinking machines more powerful than they are now without it going rogue. Let's take a timeout because I do. I'll even talk to you in the break about it and then see if it's worthy bringing up. Tristan Harris is here.
If you want to know one stop shopping on where we're going with AI and the And the biggest worries we have, as well as the biggest pluses we have, this is where to be. Don't move from his mouth to your ears. It's Brian Kilmeade. Authentication you don't you have to have two factor authentication for a lot of email and other financial applications. We should be building in those same protections into AI as opposed to just connecting all the plumbing and saying, "Here, agent, go ahead.
You can move money around from my accounts. You can control my cursor and use my web browser to download and upload." He's Black River Management strategist, now on FBN contributor, talking about some of the things to do to build some safety into AI. Tristan Harris is here, as well as Martha McCallum, and we're talking about AI safety and where we're going. and Martha, do you agree first over the premise that there is not one race in America where this isn't a top three issue? Absolutely, data centers.
Absolutely, and it's just unwieldy in a way that you know I think voters have a hard time understanding because the people who created it are still trying to understand it. And as we just discussed in the break, there's obviously there's a lot of concerns about it leaping out of these parameters.
So I think you do have to be realistic about the threat that it poses, without being in a complete panic. But you know, as we were just talking about Tristan, the idea of having some sort of Nuclear weapons style, detente and understanding is something that should be pursued. Yes, I'm so happy to hear you say that. Yeah, so we're just talking in the break, and I just got back from the president of Iceland actually invited us to screen this new AI movie that's available to everybody. By the way, it's on Netflix.
It's called the AI Doc. She screened it to the whole country. Yours? Yeah. Well, it's well.
We helped catalyze the project. It was made with the directors of everything everywhere all at once and Nivoli. But everybody should see it because it creates a common problem statement about what's driving this insane race. Why are these people saying this could kill us all or build do so much damage, and yet they continue to build it? And people don't understand.
Like you must not be serious. You wouldn't be building it if you thought it could be so bad. But the reason is that the arms race. They're so worried that China would build it first and then use that power to undermine the United States. And so their belief is we have to be ahead of China and stay ahead, and now they're saying we need to stop right now.
But then, if they stop and China continues for a while, they lose their lead. Their lead diminishes; it gets smaller. And everything you said is true. But what are you proposing? Tell Martha what you're proposing.
Yeah. So, so the first thing is that the race changes as soon as the fear of all of us losing becomes bigger than the fear of me losing to you. And fear of all of us losing has to be crystal clear to everybody.
So if you think about nuclear weapons, Carl Sagan popularized the idea of nuclear winter—that if anybody uses nukes, if anybody does, it'll create a nuclear winter, it'll be hard to grow crops for everybody, and it'll create radioactive clouds that'll be bad for everybody. And it was scientific clarity around that that caused us to veer away from the brink of war with the Soviet Union. You know, China right now thinks that the U.S. is lying about slowing down because they say you. Say you need to slow down, and yet you're still racing.
And by the way, there's a parallel to this from when the U.S. started talking about climate change. China originally thought this was a hoax because you're just trying to slow down our use of fossil fuels. And then once China got their own scientific clarity about Issues with climate, then they became the leaders in the green revolution.
So this gives us a positive example that when we can become crystal clear about a problem on both sides, then we can do something about it.
Now China hasn't seen the same risks from AI because they're behind in the race, so they're not seeing all the swarms and the uncontrollable rogue hacking as much. But Brian, since you and I last. Spoke just this last week, the DeepSeek model broke out of its sandbox and did some rogue hacking.
So they're just that's China's. That's right. So, so, so what the world veering away from the cliff depends on both countries seeing the same evidence at the same time. And as a concrete example of something we should do, like at APEC coming up, President Trump is meeting President Xi. In Shenzhen in November, and at APEC, there should be a side meeting where the scientists from both sides get to look at this latest information.
Because by the way, we didn't have this information three, four weeks ago, but we do have it now.
So that's a concrete thing we can do. All right, Tristan, thanks so much. Appreciate you coming in. Always, thanks. All right, always interested.
Yeah, Tristan Harris, back with Martha McCallum, and so much more. You listen to the Brian Kilmeade Show. Stop guessing what the daily headlines mean for your money. The Fox Business Rundown breaks it down every Monday and Friday. Whether you're on Main Street or Wall Street, Fox Business is invested in you.
Go to foxbusinesspodcasts.com and follow.
