Anthropic Hits the Brakes, ShinyHunters Breaks Trust and China Builds AI

Hey everyone, happy Wednesday. Welcome to Techstrong gang. Seeing the blue and orange, it reminds me every day of my Knicks.

And I guess the big question is, do they return to form? Oh, Mike's got the lucky shirt on. There we go.

And we got no orange bananer in the house, so there's that, too. But will they return to form or will they get pushed around by the bullies from San Antonio? We'll find out- ...

soon, tomorrow. But for today, we got Techstrong gang to talk about. It looks like we've still got Elton out in Las Vegas there.

Yeah. It's sweltering, but I'm leaving pretty soon. All right.

Yeah, it gets hot. When I was out there for Cisco Live, it was like 104 or something. Yeah, it's going to be 106 today.

Yeah. Oh, geez. It's a dry heat, though, John.

Okay, that's true. It's a dry heat. You just feel like a piece of barbecue.

Anyway, John Schwartz is with us. Of course, Terry Robinson, our friend Fred Wilmot, back in the Pacific Northwest, and our resident Knick fan, Mike Vizard. Excuse me.

All right. Here we go. Go New York today.

But Mike, beyond the Knicks, there's big news yesterday, was Anthropic launched what they're calling Fable 5, which is Mythos with security guardrails built in. Basically, you can't use it for its security functions. But from everything I'm reading, it is heads and shoulders above any model that's come out before it.

But interestingly, they did put in, you want to call it security brakes, safety brakes? I think splitting the baby in half maybe a little bit, but what do we got? Well, I guess we're wondering how good are the guardrails, because all previous guardrails seem to have been not all that particularly useful.

But Anthropic is making some claims, and then it gets more interesting. They're going to keep your data for a little while just to make sure you're not evil. So John, walk us through what's going on.

Yeah. That's a good distinction to make. So Anthropic, it was yesterday, Tuesday, Claude Fable 5.

First time the company's making Mythos available to the public, although it comes with safety guardrails, as you mentioned, and to controversial data retention policies. The company says, and the company's going public soon, so this is interesting in timing as well, they've implemented what they call this hard brake pedal. 8.

Now, according to Anthropic, they claim the safety triggers are conservative, but rare, but 95% of user sessions running entirely on Fable 5. So for the most part, you can do what you want to do, but with rare occasions, you can't. I get a ton of responses from folks, especially people in cybersecurity, which I can't wait to hear particularly what Fred, Terry, Alan, Mike think about this.

There are concerns that the primary mitigation appears to be routing sensitive cybersecurity and biological queries to lower capability models. While this would reduce the risk, they mention that it introduces this new attack surface, which is the routing layer itself. Somebody else I talked to referred to this as opening a Pandora's box.

They're predicting within the next six months, we're going to see open source models that can do the same thing with similar levels of efficacy, and that'll be able to run on commodity hardware. So they fear that any sufficiently motivated attacker can point these tools at a software stack or targeted organization and expect success. So there's a lot to unpack, but it is really interesting, and the level of interest in this story was sky high compared to anything we've experienced in the last few months.

So I've been dying to ask Fred this question because I went to this concert last night, and I'll probably explain what that was all about, but I'll just put this little souvenir I got from the concert. Okay. And so Fred, my question to you, Fred, is all this leading to the de-evolution of all the things that have gone on before and are we all going to hell in a handbasket?

Is Devo right? We hit a new low on Techstrong gang today, my friends. Wow.

That's going to break the internet, Mike. Yeah. That's going to break the internet.

I didn't know we were bringing props. I'm going to have to- Yeah. Well, now the floodgates are open.

But guys- That's- There's a lot of really interesting, I would say, moments of contention that goes between this press release and some of the conversations about what does it mean to downgrade to a model? What is the, you mentioned it, John, the connective tissue about what is transported from what place to where. And there are some fundamental things here.

Whether or not you spend 1,000 hours to evaluate the breakage capabilities of being able to operate a model at that scale or that capacity, there are open source models that we've talked about before where obliteration becomes a thing in a relatively short period of time. So who's to say smaller models with more capability that are open can't take advantage of some of the adjacencies here, especially in agent harnesses now where you can use multiple models to pass the same information, to work across models, not just through a model. But there's a transparency issue you mentioned, the data retention versus identity is an interesting conversation.

Whether or not that I want to downgrade you based on trust and the way that I ask my questions, right? I can tell you today that this particular model does infer that you might have a malicious intent just from about 15 different questions I asked. And I said, "Hey, by the way, you don't know me well enough to infer my potential interests here.

" And the answer is fair point, right? The question mark is how fast does someone through normal practices, as a cybersecurity professional, asking questions around the remnants of some of the issues that are obviously going to surface here, become something that this faster, stronger, cheaper model quickly relegates to something that can then be taken advantage of in some capacity? So I think there's a lot of issues here, especially that this now opens the door to folks that don't have what I would consider to be the earned experience on what the right types of questions are to ask.

And it might sort of quickly spoon into, it's not very expensive for you to figure out how to build an exploit harness. Not saying you can do that today. I wasn't able to in the first 15 questions I asked, but that's not to say that's not available to you.

So I think, Mike, to your undoing the fabric of the internet, it's possible, right? Evo might not be wrong. It might not be today, might be in a week.

Mm-hmm. " The bad guys who want to do bad things are going to find a way around this. 100%.

Whether it's attacking a new layer, finding a back door, something. They're going to find a way around it, and if they don't find a way around it within Anthropic, as Fred is saying and as John mentioned, it's only a matter of time, right? When the US exploded the first nuclear bombs over Japan, it was a given that there were going to be other nations that were going to join the nuclear club.

We've done a good job over the last 75 years of keeping it to a relatively small group, though there are rogues like North Korea, and Iran want to be. But this kind of technology cannot be bottled up. It cannot.

You're not going to put a do not enter sign and people are going to stay out. Mm-hmm. I think the smarter thing to do is say not a question of if, but when.

And we need to start coming up with answers for what we do when, not if. Exactly. And that's where we've got to put our effort.

But to that- To your point, though, it's going to happen because a lot of those secrets from the Manhattan Project were stolen. There was espionage, and you can bet your bottom dollar there's more espionage going on here still. And so some of this technology is going to magically show up in some sort of model somewhere.

Might be open source, it might be from some other country, some other where, but it's almost impossible to keep a lid on this thing. So, eventually we're going to see multiple models with this capability. Yeah.

So we shouldn't pretend we can keep a lid on it then. Right. And I think, Alan, you're right.

I think it's a persistent problem with security that- Can you say that again, Terry? Wait one second. Did you say that again?

You're right. Okay. I don't get enough of that.

I'm your girl. Then I will tell you, you are right. But I think, it's a persistent problem with security, that it's almost always on its back foot and reactive, and we've talked about that a number of times.

And if you don't sort of do that, not if, but when, I think you're going to perpetually be on the back foot. And this is probably, for this kind of technology, a dangerous place to be, more so than in any other time. Mike, you're right, Mike, about what you just said, because I was just interviewing the chief security officer at Zscaler.

So I'm out here in Las Vegas at this conference, and he said his greatest concern was these open source models are going to start surfacing in the next several months, within a year. He's talking about half a dozen examples maybe emanating from China. So it's going to happen, so why not do what Anthropic's doing now?

Mm-hmm. There's a few interesting problems here that I think are just, I would say, really would've loved a little bit more collective thought on this and weaving something together here. But access gating historically is a really hard thing to do because you have to verify who I am to expect what I'm going to be able to get access to.

Transformers are notoriously brittle, so it's kind of not a great combination of putting those two pieces together. So if I decide that I have a classifier, excuse me, not transformers. If I decide I have a classifier model-based approach to determine whether or not I'm doing the right things, that inherently is going to require a significant amount of tuning.

So as you look at that and you think about what does it mean to responsibly disclose something? What does it mean to you keep and analyze logs that me and every other security person, all the normal QBs look at and ask questions about over 30 days. How does that get back to the community?

If you say there's parity between what's happening from a vuln research perspective and what's meant to happen, that these breaks are supposed to support the defensive side, I don't see defensive side symmetry here that suggests that, hey, I would probably like to know some of those advanced threats that are happening in a way that I should be shared. And maybe I don't have the credentialed access as a security researcher to Mythos, but I am paying money for this particular access to this model. I'd like the value or benefits if we do have defender side symmetry that can be utilized.

I agree with you, Fred. And again, it's part of why we can't have nice things on the internet. You may say you're a security researcher, and you wear a white hat when you come see me, and then later on that night, I see you down at the club, well, not in one of those orange cone hats probably- ...

but maybe in a black hat, right? And today's white hat is tomorrow's black hat and vice versa. " Right?

That is the crux of the issue. But there's another point here to be made, and I'm not making excuses for Anthropic. I don't necessarily think they're worse or better than any of the others.

But the pressure to put out that next model. Mm-hmm. The pressure.

I'm going IPO. There are already people questioning whether my model is profitable, whether it'll ever be profitable, whether it's going to justify a trillion dollar plus valuation. That kind of pressure, both on OpenAI and Anthropic, and the rest of the stack for that matter, leaves them little choice but to put security into its usual seat at the train station while the train's pulling out.

Mm-hmm. And this is the issue, right? And unless we could do something to alleviate that pressure, some sort of steam valve, you can't blame them for doing what they're doing.

Terry- They gave it their best shot, right? Or a good college try, good school try. Terry, won't they just go discover who's been granted access among the "security researcher elite" that Fred was describing, and just steal their credentials like they do everybody else's, and log in like a regular user?

And who's going to know? Well, right. Why would we think that wouldn't happen?

That's crazy. I'm just thinking over the course of my career, how many stories I've written about credentials being swiped. I don't think we can count on the fact that we're going to have this sort of exclusive group, well-vetted, whatever, because I think the credentials will get stolen.

Yeah, but even if they don't get stolen, it's a matter of time till the open source models give you that functionality anyway. Well, I think- You could have your cake, and you could eat it, too. Mm-hmm.

Either way, it's going to happen. Yeah. I don't think it won't.

And so the argument to be made is, getting real Marxist on you, what's better for the greater good here? The greater good is to get as many security researchers as I can access to this, because the quicker I do that, the quicker we can get to a secure code base. Right.

Okay, and it goes back to what you said, it's not if, it's when, so. Yeah. So back to the original question here, and I'll throw it to Alan, what should we be doing about all this as people, right?

Should we be, I don't know, digging bunkers in the backyard, or we just kind of sit here and watch and see what happens? I think... Well, what should we do as people?

It's hard to answer that when I see you on my screen here with this hat on. But- But seriously, what we should do is be vigilant, don't leave our common sense at the door. Know that this is a possibility, and it's going to be things, quite frankly, that are beyond our control, but we can't stop living.

No. Put one foot in front of the other and keep doing what you're doing. Carry on.

As Devo says, man, are we not men? So- We are ... hopefully it comes back to us.

On that note, we've got to hop to our next segment. But Mike, you don't have to take the hat off. How many of you want to vote to make Mike wear the hat at least for all of today?

What? Wear it the whole episode, Mike. Come on, you can't take it off.

We're going to leave this topic because it's going to keep coming through anyway. All right. Let's get to our next topic here now.

You're wearing that stupid color anyway. All right, Mr. Devo, what have we got next?

All right. Well, Terry has this story. So here we are talking about the pending apocalypse, and yet, when we look at security, there's this group out there called the Shiny Hunters, and basically, they are running around and making a mockery of our defenses because they keep breaking into things and stealing things, and grabbing it, usually just by simply grabbing a bunch of credentials.

But Terry has a story talking about how they accomplished all this on Security Boulevard that I highly recommend. But Terry, why are we unable to thwart this very, what seems, simple attack? I'm beginning to see them a little bit like Iran here.

They're able to like- ... get their stuff no matter what- ... measures are up.

Well, I tell you what has basically happened. We write about Shiny Hunters all the time, right? They have these high-profile targets.

And more than that. So what they do is they play on trust. They found it's not just a vulnerability here or there, it's like something that's more structural.

It's trust. And that's how they sort of worm their way... They break the trust, but this is how they worm their way in.

And I guess it's really more clear in their attack on, I'm going to probably pronounce this wrong. Is it Anodot? Is that how you say it?

Anyway, that's- Anodot ... a data analytics firm, and they basically just sort of banked on federated trust to move through that environment and into the environments of third parties associated with them. And most of the security tools that we have now, or the way that we approach security, doesn't really detect that.

And that's how they've been so successful. They're just like a thorn in the side of organizations out there trying to have some sort of steady and strong security posture. They're just kind of proving that they haven't done it yet.

But yeah, it was interesting. It was interesting to talk to people about this, and that maybe one of the things, and I guess we can get into it, is we should start concentrating more on behavior, because that's maybe the way to recognize where these guys are and that they're there. So some of the victims are probably asking themselves this question, which is, we spent millions, in many cases, on cybersecurity, and yet we still fall prey to this thing.

So I don't know. Fred, is there something amiss with the way we think about investments in cybersecurity and these, shall we say, unfortunate outcomes? The short answer is yes, but this is a really interesting conundrum here.

Because a lot of what Shiny Hunters has been able to exploit is not what we would normally associate with first-party risk. Right. It's third party.

I have Salesloft and Drift tokens, and all of the relationships that those valid credentials that are used regularly by automated systems provide. So as a CISO, I have N-party risk concerns, third-party risk concerns. I have people that are processors downstream and upstream that also we've got to think about the risk and have to operate with that.

That's all residual risk that I can't control. And so when you look at some of the vehicles here, if you couple that with valid credentials and normal processes and automation used for these valid credentials, and you couple that with the most predominantly effective way of getting humans to do things in social engineering context, and you harden that with OAuth tokens and multi-factor auth and smishing and phishing, and these terrific ways to sort of sidetrack normal, good security principles that most people try to follow. It's a nightmare.

It's super difficult. It's hard to figure out whether or not, to your point, Mike, whether or not the money you spent or invested in your cybersecurity program is actually going to mitigate this. This cuts against the grain, specifically on the technology stack, and then again, in the opposite direction with human interactions and social engineering.

It's clever, and they've been able to take a number of very sophisticated organizations down. Sales- And something everybody should pay attention to. Mm-hmm.

And they just keep doing it. I can only imagine the number of organizations, and we know some of the ones that haven't announced. Not everybody's going to tell you that they had this sort of thing happen here.

But there's a lot of news coverage on researchers who might share something that's happened as a function of their knowledge of seeing data on dark web interactivity as to who maybe has been popped in some category or another. But fundamentally, the challenge is once you've hardened some of the exploitation here in valid credentials by normal users, the behavior part, Terry's right, is how do you characterize what that behavior looks like? This is an area that most people don't do a great job at.

Normally, we rely on technology and tools to do that. And Shimi, here's where you go, "Okay, was my investment in that SIM a thing that does me good? Should an IDP, if I have delegated auth or whatever, should they be doing this work on my behalf?

" So maybe we should just have less integration because there's too many things to be exploited here. Why stop there, Mike? Just unplug from the internet.

Well, okay. Write letters again. But it sure does feel like we are not vetting enough of our integrations here with these things, and that's what these guys are exploiting.

Well, that's right. So there's that. Yes and no.

Here's the thing. We've known this in the security world for a long time. 80-plus percent of incidents are exploiting well-known vulnerabilities that have been around for a long time.

For all the talk about zero days and exotic things, garden variety bread-and-butter security incidents take advantage of well-knownvulnerabilities or ingress, egress. We know that the weakest link very often is the person behind the keyboard, security training, security awareness. We know these things are not bulletproof.

The problem is, again, like we were talking about Anthropic's pressure to release the next model and stay on top of the game, it's the pressure to do business. It's the pressure to get things done. It's the pressure to do those integrations because they bring such business value.

And security concerns are something that supposedly get measured into the risk of doing them or not, and you make your decision. It's not the security person who makes that decision. No.

And this is the plight of the security person, of the security admin. We know there are things out there that we'd like to do better, that we should do better, that we can have maybe less integrations, go slower, whatever. But at the end of the day, security is about risk management, and if we deem the risk versus the reward suitable, it is what it is.

I just don't think we're having that conversation enough. But Terry, let me ask you something else about this. In real life, every time there's a crime, there's usually a copycat not too far behind.

So am I going to start seeing multiple ShinyHunters out there, or is this a one-off because of some unique capability these guys have? That's a good question. You would think that others would follow suit on this because these guys have been extremely successful and their formula, if you want to call it that, is kind of known by now, right?

But I don't know. I haven't seen it yet. Have you guys?

I think that these guys have just been really bold, and they've just kind of figured out the structural weakness. But I don't see why we wouldn't have copycats eventually, unless they just make the calculation that it's not worth it now that everybody's onto ShinyHunters, but I don't know that they can. How do you know we already haven't?

Well, that's what I'm saying. Has anybody heard of any? I haven't heard of them, but that doesn't mean that they're not there.

And these guys remained undetected, right? They remain undetected, so there could be all sorts of little copycats floating around. And my analogy at the beginning of this, it's like West Nile, right?

And delivered by the lowly mosquito, and people don't know that they've been infected until they start showing signs of disease sort of way down the pike. So yeah, they could already be there. There's 30-plus vendors that have been, in essence, affected by just the drift breach in and of itself.

Mm-hmm. Thirty-plus vendors. I don't know the exact number of how many of those are air quotes zero trust vendors, but I can guarantee you at least one that starts with Z is in that list.

And when we look at what that actually means, when identities get compromised at the edge, that is a significant risk for every organization. Significant and difficult to protect against. So when we look at the arbitrage of this, a certain amount, a very large, a Fortune 500 company has to look at the number of vendors that they may be affected by.

If there's 30, I would be demanding transparency, right? And then also rotating all the things, but the way that these entitlements work between these, they're not meant for rotation every so frequently. And as much as we hem and haw about what does it look like to be passwordless or zero trust or whatever, at some level, what's a CISO's job?

To have the business operate business as usual. It is not to make it so difficult for you to operate that people don't do business with you. So inherently, Shimmy's point stands, which is- ...

you have to look at this as to whether or not the cyber- What a day I'm having. Yes, I know. Too right.

Just get one of these broke. One. Huh?

Otherwise, it's going to go to all of our heads or Mike's- Well, no, but even the broken clock's right twice a day. But go ahead. Here's for two.

And that's really where I want to see this accountability stretch is we look at, as a CISO, my cyber risk looks like this, my cyber insurance looks like this. I have to do business with hundreds of other businesses, SaaS and otherwise, and I want to know, and I want to be able to operate on that risk in collaboration with my trusted partners. If folks have had some level of knowledge of this for more than six months and it hasn't been remedied or cured, why are we still doing business?

These are the kinds of questions that we look at- Right ... pull back into the first story we had. How much of that accountability for software vulnerabilities and third-party risk does a CISO have to eat with these new models coming out, whether or not these vendors decide that they're going to cut their margins to make their products more secure, or at least more aware of behaviors that are unfitting my CISO's policies?

I get it. Mm-hmm. I get it and I agree.

Mike, unfortunately, I think we're at the edge of this segment. We've got to hop. Yeah.

So John's got to hop to the airport, so I'm going to let John run. Yeah, I'm looking at the clock. So Elton- Go do your thing, John.

All right. Talk to you guys. I hope you have a clearing pre.

Have a good show. Safe travels. All right.

Safe travels. Thanks, Jim. I'll see you.

Okay. Bye-bye. But that does bring us to our next block, which kind of relates to some other conversations we've been having earlier this week and into late last week.

But China is making some significant investments in AI and especially in the area of networking. They're tossing 295 billion in this investment. Now, that may not seem like a big number when we're talking trillions in valuations, but it signals that they are starting to think about AI as a utility.

And what they're trying to do is create something that feels like a national service, and maybe they think that their startups and software companies will leverage that infrastructure to go interesting things and add value. But I guess I'm going to toss this to Alan, but is this something that we in the United States should be considering rather than the United States government buying a share in all these AI companies? But maybe we should start thinking about this as a utility, and after all, the value is in the software and what we do with it, and not necessarily in the models themselves.

Mike, I've got an article coming out tomorrow, and I'll be doing a "Shimmy Says" on something adjacent to this, and it deals with China actually as well. Here's the thing. The thread running through everything we've been talking about over the last few weeks is that AI as a category is bigger than just the GPU semiconductor question, bigger than just the what model are we using question, even bigger than is this going to break our security?

The real issue we're dealing with AI is that this is fundamentally threatening to morph, stretch the fabric of civilization, right? The whole AI build-out, the data center build-out, where Google made a conscious decision that it's cheaper to pay a billion dollars a month to SpaceX and let them worry about building the data center. Where Elon released the first plans for the data centers in space about the wingspan of a 747.

China just announced their underwater data center, 10 meters under with a wind turbine farm on top of it, generating the energy. The amount of build, the $8 trillion in build-out of data center infrastructure, it's tearing at the fabric of what we do, what we are, who we are, what we do. It's bigger than any one of these companies, and quite frankly, it's too big to concentrate in the hands of four or five tech bro billionaire egomaniacs who are going to determine the future of humanity or some dictatorial thing.

We need government intervention, whether it's government ownership, government oversight, like a public utility or something else. And I'm not a big government guy, and I'm not, you know how I feel about the present administration. But this is one of these things where you need government to step in here and make sure we do what's right for humankind, not what's right for a few good men.

I don't know how good they are. Is our government capable of that, though, at this point? I mean, is that just trying to read the tea leaves?

Who else you got? Well- What do you want, the CNCF to step in? I mean, who else do we have here?

It's bigger than any one government. Maybe you need some sort of worldwide consortium. I don't know.

Well, that's an important point, right? So let's look at what China's been up to elsewhere. So they're clearly building roads and all kinds of infrastructure for other countries around the world, including Africa.

They'll do the same thing with this AI utility. They'll extend it out to those other countries that can't necessarily afford to make those investments themselves. So I kind of feel like ultimately a hand in the United States is going to get forced because it's going to become a much more strategic conversation that goes to Alan's point beyond the tech.

What do you think, Fred? Completely agree. 0, all of the above.

And I agree, the jurisprudence of this is going to require government involvement, maybe not complete government sponsorship. There's a lot of reasons why China wants to own networks, too, and this is a very clear statement on they want to be reliant on their own infrastructure, their own manufacturers. Okay, cool.

All of that makes a lot of sense if you view the United States as the ultimate conflict for you. Similarly, we have to be able to look at this problem with the same lens, and I agree on the requirements for not having five tech bros in whatever capacity. Five humans rarely make the best-informed decisions for the greater good, in a sense that allows them and allows us the opportunity to fully maximize what it is that's in our best interest, and it is a long game that we're playing here.

So the challenge that we have is historically, the United States thinks in 10 years and 20 years at a max. China thinks in 50 and 100 years. And so when we think about what measurement, what moves, and what sort of futurist things that we need to be collaborating on here, I think more time spent on what that should look like nowwill have much greater returns.

Also, oversight, transparency, clarity, collaboration, and the knock-on effects are maybe we aren't going to destroy all of the environments, all the waters. I'm channeling my inner Gina right now, by the way. And data centers that exploit everything, and so on.

And I think that kind of collectivism just in that construct, without the bounds of what everybody else will say immediately following that, is really important, and we should seek to embrace that as fast as possible. Right. So Alan, are we having the right conversations?

Because it seems like we're obsessed with IPOs and what's going to come out of Wall Street, but do you think that governments and policy leaders are even aware of what the conversation should be? So again, you know me, I'm not a fan of this administration generally. But I do think that there are people in this administration who realize the stakes.

There are some who are going to want to do it for their own personal enrichment, greed, and I get that. But I do think that people in this administration know just how important this AI thing is, and because as I said before, it transcends just the technology. Look what this data center build-out thing is doing to our electric grid, to our water conservation.

I've written a bunch of articles, parts of Louisiana that were dead since cotton are now latching on to building a data center there, giving them 80% property tax- Rayville ... rebate. That's exactly right.

I passed right through there ... for a couple of hundred jobs at most. And that's just in the US, where we're relatively advanced.

What's going to happen in Africa, to Fred and Terry's point? What's happening in Europe? What's happening around the world?

This is a big issue, and that's only level one. Level two is now this stuff starts working, and we're starting to displace workers. And Mike, as you've wrote up earlier this week on the gang, do we need to go to some basic universal income?

So people hang out. Because what people are going to do for jobs? If you're a woman pregnant or a husband and your wife's expecting a baby, what are the prospects for that child growing up?

Are they going to be a security person? Are they going to be a programmer, a coder? What are they going to be that AI's not?

Right? These are big questions. These other people have had their 15 minutes in the sun.

It's time to get serious now about what is this really going to wrought, and how are we really going to deal with it? We can't keep kicking that can down the road. No.

Mm-hmm. Well, why not? We kick everything else down the road.

Well, we do, and then we lament over why it didn't turn out so good. This is true. Well, I think the next generation is starting to ask some questions about what the hell we've been up to, so maybe they'll pay more attention.

I don't want to be pessimistic. Here's the good news, is there's still time for us. There was a time when we'd had the brightest and the best working at NASA and other kind of government institutions.

There's still time for us to get our act together and decide what's the best way to deal with this. How do we make this lemonade here? How do we really leverage this amazing technology?

And I do think at some levels of the US administration, that is a consideration. Mm-hmm. How do we do that then?

Because all those people are working for companies that are going to do massive IPOs, and they all want to be billionaires, and they're not going to do it for the nation, for public service purposes. You know what? Yeah, that never seems to .

There's always been, in our history, people who rise up who do what's right for the country and for the people. We got to find those people. Mm-hmm.

There's an awful lot of money involved. There's crazy amounts of money involved. It's nuts.

And look, I look at communities like here in Louisiana and that Richland Parish Meta- Mm-hmm ... you know, data center, that's a thing, and they're promising 300 to 500 jobs now there. I think that's- By 2035.

Yeah. But I don't really think that's going to materialize. And just this week, they announced that because of the tax windfall from all of this, that qualified teachers are going to get up to $50,000 bonus for the school year in Richland Parish.

Now, they won't all get that, but what I guess my point is they're kind of seducing the people in those areas who are going to have a lot more lofty ideas about what's going to be done there. And I just, all of it, the money, the sort of shading things, really, and then a bunch of people in government at all levels, even at the state level here and local, that they not so interested in the long term, and they're not so interested in harnessing all of this and- Well, let me give you another analogy. When we were building out the highway system, the Eisenhower Highway System.

If your town was at an exit on the highway, you were goldenYou could have gas stations, restaurants, hotel, motels, shopping, and everything else. If you weren't at an exit on the highway, you became a ghost town. And I get that, that you kind of need to be an exit.

You need to be at an exit. But here's the thing, what's the equivalent of that? Because they're not huge ecosystems that build up or- No ...

economies around data centers. They're sort of isolated. But like I wrote with that Richland Parish, if you're a starving man in the desert, any morsel of food looks pretty good.

100%. And listen, I never put down the fact that they're getting some jobs right now in terms of constructing these things- No ... even if they're temporary.

I know that parish very well. I have friends who are business people there, and I know they're hurting, and they can use whatever they can get. I'm just saying I feel like the dream sort of overshoots the reality when it comes to the business.

Doesn't it always? It always. Doesn't it always?

Hey, we're about out of time, though, guys. We got to end this. Mike, I appreciate you being a good sport and wearing your Devo hat for the rest of the show.

It was great. Here's the one piece of advice for all of this. Whip it.

Whip it good. " But- ... all right.

I was going to say economy of control, and it... Yeah, never mind. All right.

Let's go. Let's end it right here. Fred, Terry, Mike, thank you for joining.

Thank you for watching. We hope you've enjoyed. tv, on our OTT app, on YouTube.

Anywhere you've got a screen, we're probably available there. Stay tuned for more Techstrong Gang coming, or more Techstrong TV coming, in addition to the Gang. And we'll be back tomorrow with even more stuff.

Until then, this is Alan Schimmel. Thanks for watching.