Windows 10 Concession, Anthropic Armistice and AI Crime Prediction

Hey everyone. Good day. It's Monday.

Man, we are at the end of June already. It's officially summertime. Well, if you're in Europe, you know it's summertime.

I was talking to some friends there this morning outside of Stuttgart. Other friends of mine just got home from Paris and Amsterdam. It's melting there.

Mm-hmm. My experience in Europe is they don't have a lot of central air conditioning built into a lot of the buildings, and they have these units per room, and it's tough when it's hot like that. Anyway, we're pretty hot here at the Techstrong gang, too.

We've got all kinds of stuff going on. We've got a great gang lined up for this beautiful Monday. Let me introduce you to them.

First of all, it looks like he's home in Columbus after his Zabumbafu adventures, the one and only Stephen Foskett. Out west, also home, though we hear he might be heading east soon. Go east, young man.

John Schwartz. And of course, the man in the high castle who's going up to Lake Placid to practice his bobsled runs in the middle of summer. Mike Vizard.

Mike Vizard. Good to see you again, Mike. Gentlemen, welcome to Techstrong gang today, and welcome to you out here for watching us.

So, Mike, I find it funny when we call it a concession that a vendor, out of the goodness of their heart, is going to extend support to their product that we paid hard-earned dollars for. Yeah. But is that what we got here?

Well, I think it might be even better than that, but I'm going to toss it to Stephen in a minute. But basically, Microsoft's been making this case and saying we're going to turn up support, free support anyway, for Windows 10 security updates, and they were trying to push people either to-- Well, they wanted you to go to Windows 11, but they would also make available various paid programs that you could agree to and get some updates from. And I guess, Stephen, though, there are folks out there who always thought that this was never really going to be feasible because nobody was moving off of Windows 10, and other people who just said this was a bit of a scam to drive people to Windows 11 that nobody seems to really want.

But Stephen, I don't know, what's your take here? Yeah, it's funny. The uptake of Windows 11 has been a little slow.

I think that has a lot to do with the fact that it's a major refresh to the user experience, but also the fact that it requires specific hardware. 0, without Secure Boot. It won't run, well, I know this sounds like ancient history, but anything that can't run 64-bit.

So, there's a number of PCs out there that are still stuck on Windows 10 for legitimate hardware reasons. Now, I guess you could ask yourself, should those still be in operation? Microsoft is a victim of their own success here, right?

Because Windows 10 was a massive success. It's running everywhere. And now they're finding themselves stuck with an operating system that is well over 10 years old, a replacement that is over five years old, that people haven't thoroughly embraced, running on systems that could be as much as 20 years old.

It's kind of amazing to think that people could still have these things in production, but they do. Frankly, I think that Microsoft is doing the right thing by extending essential security support here by not necessarily requiring payments, though there are some hoops to jump through. And I would guess that they may push this out by another year, kick the ball down the road by another year, depending on how uptake is going forward.

But that being said, eventually, they have to stop supporting Windows 10, not just because they would love for people to use Windows 11, but theoretically, there will be a Windows 12, and I'm sure that they would love for people to use that. Also, because a lot of these systems are probably getting pretty old and cobwebby and full of bugs and dirt and lint and probably going to stop working pretty soon. So, I applaud them for extending support, but come on, folks, it's time.

I don't know. Here's the part I never quite fully grasped. So, I get that there's TPM and that makes things more secure, but most of the code that sits in Windows 11 is related to the code that sits in Windows 10.

So, what makes you think that this is really going to be a better outcome from a security perspective, especially in this age where we discover vulnerabilities using AI tools? And I have to wonder, is Windows 11 really the more s-- I imagine it is a little more secure, but I'm not sure it's fundamentally secure enough to make a difference. Alan, what do you think?

I think there are two things here that you're not factoring in. Number one, I think as Stephen mentioned, if we can force people, and maybe force isn't the word, but it is forcing people to move off of old, and I do mean old, hardware. That's a good thing because the vulnerabilities are not necessarily in the software there.

They're hardwired into some of the hardware. And so that in and of itself is a good thing. Secondly, at the end of the day, Microsoft has a business to run, and supporting two code bases, even if it's just for security purposes ...

on a major rollout like that is huge. And so are you cutting the baby in half, giving half the resources to support Windows 10 security while I need to support Windows 11? And I need to put all my wood behind that one arrow and make sure Windows 11, if that's the latest and greatest, is the most secure it is.

But one other thing I'll add, if this was Apple- Yeah ... no one would be thinking twice about it. Apple does this all the time.

Yeah. You're running it, macOS on an Intel chip? You're SOL.

You're SOL, right? You could still run it, but we're not doing anything with it, right? And there's a company out there, as the sizes on my screen keeps changing, so it's very confusing for me.

But most companies that I see, even open source tools that get orphaned and stuff, there's a company called HeroDev, I don't know, HeroDevs, I don't know if you've ever heard of them. Great company. They've built a tremendous business just supporting orphaned open source projects.

So this idea of, well, I bought something back in 2014 or whenever it was that Windows 10 came out, right? And I still expect you to be supporting it and giving me updates, rubbish. I can't even- Well, here's the part that I never understood, and I'll toss this to John, but I'm like, okay, so if this was Apple and they were really in charge of this thing, they would've launched Windows 12 by now and said, "This is- Yeah ...

the only option for everybody," and moved along from there. But, for whatever reason, Microsoft doesn't want to seem to admit that maybe Windows 11 isn't quite what everybody's expectations were initially, so maybe it's time to move to an odd number. They're luckier, or an even number.

I was stopped in my tracks by what Steven mentioned about the longevity and popularity of Windows 10, and I'm trying to think back. I remember going to a Windows launch in New York. It was a long time ago.

It was the security-focused version of Windows, and the name escapes me, and it landed like a thud. It was a complete no-show, and they kind of did pivot away from it faster than this. And I was thinking of two things.

I'm glad you mentioned, both Alan and Mike mentioned Apple. Yeah, they never would've stood for this. They would've moved on by now, and I'm just wondering, is the adoption rate of 11, it's been around forever.

Is it that bad? Put this on mute. It's tough.

It's one of those things where- So ... it's been around for five years now. Yeah.

Microsoft still hasn't introduced a new Windows 12. I think people were hoping for it at Build. As you mentioned, Apple has moved very strongly into a release cadence now.

Every year there's a new operating system. Some of them are major updates, some of them are minor updates. Every year they deprecate older hardware.

Now, some people are pretty annoyed that the new macOS 27 is not going to support any of the Intel Macs anymore because some of those are not all that old. But at the very least, Apple is moving people forward. But Apple and Microsoft is not the same situation.

Apple has a much smaller install base. It's much more of a client install base, much fewer embedded devices and IoT devices. And as you mentioned, if you want longevity, you really have to go toward open source.

You really have to go toward Linux-based operating systems. I'm running a lot of outdated equipment, including old Apple machines and old unsupported Windows machines on Ubuntu and Debian and things like that, and they run great. But Steven, I'm glad you brought that up.

Red Hat's dealt with this. I believe Canonical Ubuntu has dealt with it too. They have their long-term releases that you know get updated- Oh, yeah ...

every 18 months or whatever it is, and then there's short-term releases that get updated quarterly or semi-annually. That being said, though, even there they tell you when they're going to deprecate end of life support for a version of Linux, like Red Hat Enterprise Linux or something like that. Yeah.

So look, the open source guys are doing it. Are we holding Microsoft to their feet, to their fire, unreasonably here? Yeah.

But Microsoft seems to be afraid, to Steven's point, to force the issue entirely, right? They keep saying- Because that's always been their MO, right? When Bill Gates and those guys, and Ballmer, and the rest of them were there, that was a big distinction they always had, that they don't orphan their software.

It's always backwards compatible. Mm. Where Apple has a long, proud history of not being backwards compatible.

Right. But does Microsoft have more to lose here? And to Steven's point, because it's like, so yeah, their share of the market is much bigger and a much larger percentage of those people on Windows 10, if forced, will move to something other than Windows 11, so that- Oh, you think so?

Where are they going? Is this the year of Linux desktop? Is that what you're trying to tell me?

Maybe. Maybe they'll go to a MacBook Neo. Come on.

" And then when Apple tells them, "We're not supporting you going forward," they'll go back to Windows. They'll go back to Microsoft. Yeah.

Come on. Apple's not the play there. Well, guys- No, I think- Can I ask a...

I'm going to ask a naive question, I admit it, but that hasn't stopped me before. So this consumer ESU program's been extended, I saw this, until late 2027. Is that in a weird way, is Microsoft acknowledging that this massive footprint of PCs cannot upgrade to Windows 11 because of strict hardware requirements?

Or am I reading too much into that? At the core of it, you're not reading too much into it. There's a lot of nuance in that statement.

But yeah, basically, they're saying that the Windows 10 machines don't support that TPM module that Steven was talking about, and then that's what forces the upgrade issue and everything that goes with it. But a lot of people, heck, there's stores around town here where I still see point-of-sale terminals running Windows XP. So I'm like .

That is the point right there, Mike. Yeah. Steven, the TPM modules, those are out more than 10 years now, no?

Yeah, they're pretty widespread and it's completely true. Basically, if you're running something that old, you've got bigger problems, I think, than just the security patches, because frankly- Yeah ... it's probably not patched at all.

It's probably- Right, you probably stopped patching. Go ahead ... running on the original stem from way back.

But I think Alan- Go ahead ... points out something really smart, which is that the companies that are doing this whole thing correctly are companies like Canonical and Red Hat, which have very specific release schedules. As you mentioned, they have long-term support.

The LTS versions, for example, of Ubuntu, are 10 years they will support that, and you can get legacy support up to 15 years. They have a very specific release cadence. They're very clear and transparent about it.

They support a lot of old hardware. I feel like that's doing it right, and what Microsoft's doing by just continually kicking the can down the road is frankly doing it wrong. I know.

And I wish that they would give their buyers more visibility. I'll go another step further about the wrongness of this. The wrongness is Windows 11.

So it's not enough to sit there and tell me that I'm going to be a more secure environment if I upgrade from Windows 10. People are not laying out money to upgrade for security. I know that's going to cut Alan to the quick, but the truth of the matter is, unless there's a compelling user experience and feature capability that I'm going to see in Windows 11, I'm just going to sit on Windows 10 till somebody shows me something that's worth upgrading for that has nothing to do with security.

It has to do with- This is why I'm so excited about Golden Gate, because it's going to fix those transparent glass things that Steven loves so much. Yeah. I remember that from last year.

Man, I still haven't upgraded to Tahoe. Yeah, but here's the deal. Haters are going to hate.

Mm-hmm. The only thing Microsoft is guilty here of is supporting these old versions for too long. Right?

They've got to get their user base used to more frequent updates, and maybe the way to do it, as Steven says, is do it the way the open source, the Linux guys do it. Yeah. Where it's very clearly laid out.

You buy this today, you got 10 years. That's it, and you're done. Now, but there is the issue of embedded software, and Microsoft made a huge play to become the embedded, the headless OS, headless Windows, and all that stuff.

And when you get into those markets, that's a different ballgame, right? You're going to have legacy equipment at the bottom of the ocean and the top of the sea that's not very easy to replace- I think- ... from a hardware perspective.

I think Microsoft should, A, do something more interesting on the OS level than 11, and then on the other side of it, I think they should open up some Windows 10 assisted living facilities, right? If you want to go live in there, you can. You can relax, you can be with your buddies and hang out and that's where you're going to be.

But you know- You probably don't care about security anyway. Go play in front of your own house. All right.

Well, that was a good opening for today's show there, Mike. Looking here at what's next, have they let Anthropic out of jail yet, or is Dario under investigation? What's happening?

I wouldn't go so far as to say that, but apparently, I guess Anthropic sent somebody else to talk to Trump and the White House that they could understand or deal with, so they cut this deal that- Was it the foreign story? Yes. That Anthropic, a limited number of people will have access to the latest Anthropic models in much the same way that we're now getting access to the OpenAI models.

But John, you were covering this over the weekend. What's the nuance here? Yeah, there is a nuance.

It's good and bad. So, we talk about Friday being like the news dump day, and this is what happens. In a sense, the administration partially reversed its export control directive over Anthropic, and what they did was, and the individual, I guess, who hammered this out with Letnick, I believe is the co-founder and the chief compute officer, Tom Brown at Anthropic.

There was a two-week old directive from the administration. They rolled back this two-week suspension of Claude Mythos 5. So now Mythos 5 is available to more than 100 vetted US organization, which essentially means that ultimately, critical infrastructure defenders, Fortune 500 enterprises are going to leverage AI tools against digital threats, perhaps.

However, Anthropic's mass consumer model, which is Fable 5, remains offline while the sides between Anthropic and the government negotiate. And this brings up the criticism that's voiced specifically by Sam Altman, of course, that the federal government is wielding excessive non-transparent authority over commercial AI distribution. And a lot of- You got a little more time, though Oh, I know.

He has nothing to gain from this outcome. No. He might as well- He's dumping what they're doing to Anthropic's wrong.

Let it stay at Anthropic. So yeah. My takeaway was this decision, the latest decision before they changed their minds again, it kind of underscores this emerging regulatory framework where unsafeguarded frontier capabilities are restricted to a highly controlled elite tier of trusted partners decided by the government.

So take that for what you have, for what you will. But I kind of agree with Sam, one of the few times, with his overall attitude about this. But yeah, it was kind of a mixed blessing until there's another blow up and then we change the rules all over again.

It's- It's good ... pretty wild that we've got Howard Lutnick of all people being the one who decides what AI models we're allowed to have. I checked his credentials in this, and I don't see that he actually knows anything.

Yes. So I'm glad to have it be him who decides. I think the Trump administration- But he is-- Wait a second.

He is a member of Mar-a-Lago. Yes. All right.

Well, David Sacks is out of the picture, evidently. So now I guess he's the designated AI cyber czar or whatever we want, AI czar. But yes, Steven, you're right.

This guy has absolutely no business. No, we've got him running that, Pulte running national intelligence. The whole- Kushner's our diplomatic lead, so we're covered.

Only the best. It's Camelot all over again. So there was a development, though, that came out.

There's a story in The Wall Street Journal today talking about, I don't really know this company out of China, but they have a model called Z dot AI, and they're, I guess I'll butcher this, Zhu AI or something like that. But they're claiming that they've already developed the ability to discover vulnerabilities on a par with what Anthropic and OpenAI have done. Their reasoning models may not be as strong elsewhere, but they were claiming that they can find those vulnerabilities just as quickly, which brings me to- Is that a Chinese company, you say?

Yes, a Chinese company. So I don't know if it's been verified or not, but I will say this. Hmm.

At some point, is this whole conversation about to be rendered moot because while these other AI models are going to be out there, and this is just standard capability. Alan, what do you think? I've written about this.

The tighter you hold it in your hand, the more the sand comes out between your fingers. But let me do something totally going to surprise you guys. I actually agree with the government.

I think we are coming to a point with AI models where each release is X better than the release before it, and then the next release is better than that, and the next release is better than that. And the capabilities of these models, while they offer all kinds of promise, and we've spoken about it extensively now for two years here. We are getting to the point where these models need some directions and some warnings on them, and maybe some regulation.

Is the way this administration doing it fair, impartial, balanced? No. But we as a humanity need to decide these things are the potential for using it wrong and the potential for damage it could cause.

We need some regulation. Now, the problem is you hit it, Mike, the so-called, we call them the open source models. That really is code word for the Chinese models.

Because they're the only ones who are, right now anyway, in this game with us. But we need a planet-wide sort of regulations here. Otherwise, this stuff falling in the wrong hands.

I remember, you guys are all old enough to remember the idea of countries can develop the bomb. They could develop nuclear weapons because it took a certain amount of technological sophistication and making centrifuges and all these things. How long will it be till OpenAI can give you the recipe for yellowcake?

Since you opened this can of worms, are we going to wind up bombing some country someday because they have OpenAI research that can be used for AI because it's a type of munition? And for that matter, is AI models, should they be subject to the Defense Production Act and make sure that they're- You can't regulate this on only one border because you've got the open source models. And I think it's only a matter of time, though.

China's not looking to kill themselves either. They will take control of their models for their own benefit as well. It only makes sense.

Then you've got to look for underground open source models to keep up. Kind of like we're getting into red pill, blue pill Matrix kind of stuff here. But- Actually, and I agree with you, Alan, to my surprise, too.

I was thinking, in a sense, there is a bit of nuance. I can't believe I'm saying this. It's the Trump administration's decision because in a sense, they're prioritizing national security infrastructure over blanket restrictions.

So they make a distinction. And again, the debate, and I'm listening to a podcast yesterday about AI, and the debate, there's this constant comparison to this arms race Nuclear bombs are being equated to AI open source models at certain times. And it's not just China, but you think about other countries like Iran or Russia, what have you, with capabilities.

So I actually do think it was actually a wise decision and a nuanced decision. Yeah. I'm not going to go on record and say there shouldn't be government involvement in this.

In fact, I agree with you as well that there should be government involvement in this. I think it should be transparent. I think it should be built on expertise.

I think it should be broader than just the opinion of some guy who watches Fox News. I think that it should be an informed decision-making body. Now, that being said, I also think that we are drastically overestimating the capabilities of these models even today.

I think they sound real good and they look real good, but they actually don't produce the kind of results that people wish they did. And I don't think that Mythos is a munition. I think Mythos is just an improvement over the last one, which is an improvement over the last one, and I don't think it's going to give us an atomic bomb-making recipe anytime soon.

I think it might give us one that looks good enough to get ourselves blown up, but I don't think it's going to be good enough to give us one that's actually going to work. " Mm-hmm. I mean, if they keep improving at this rate, it's only a matter of time.

Is it? And it's more of a chorus of regulation or some form of national regulation because where this is headed and the fears over all this, and again, it's within the midterm timeframe, you're going to hear a lot of this saber-rattling and AI is dangerous for all of us, it could be the end of us. I mean, you're going to hear a lot of BS.

But I actually think absent, we're not going to have any national legislation on this. I'm just convinced. But there needs to be some sort of framework so that then- But here's my point.

I don't think national does it. This is something that was made for UN. For so many years, the UN Nuclear Regulatory Agency actually served a very useful purpose in this world, right?

You're going to need, because, and it's always been the problem with the internet, the internet knows no borders. You can't bod. Even ask China about their great firewall.

You just can't do it. And we're going to need worldwide type of... Because Dario Amodei said this, right, when he asked for a pause maybe last month.

Anthropic themselves can't afford to pause because if they did, OpenAI will bypass them and leave them in the dust. If Anthropic and OpenAI both did it, the open source models will leave them in the dust. If the US pauses, China will have an advantage.

If US and China pause, who knows who else? Only the shadow knows. It can't be a piecemeal thing.

It's got to be all in or it doesn't work. I don't know, man. One good guy with an AI model.

That's all I got to say. Could be. Could be.

Anyway, I'm glad to see that Mythos is getting out to more people because I do think it's... I did a panel on this at Platform Con last week. I had my friend John Willis with me.

We had a guy named Drew Gutstein, former Google Cloud director of security, and then CISO down on Wall Street. Mandy, Mandy's last name, I think, is Hall from PagerDuty. And my friend Ian Amit, who's also a multiple-time CISO, company called Gambock.

The thing about this Mythos thing is, do you know how many CVEs have been registered as a result of Mythos finding them? Anybody know? Not as many as people are saying.

No, I'll tell you. Zero. Yes.

Zero. Not one. But if you look at the patch releases that have come out since companies got their hands on Mythos, record-setting.

No one's saying it's because I found them in Mythos, but all of a sudden, Microsoft released 164 vulnerabilities in one Patch Tuesday. Google put out 40 or 60, I forgot what the number is. Apple put out a huge amount.

We're seeing huge amount of patches coming out, but not corresponding CVEs being registered for them. So it's already kind of changed the whole apple cart here of how we're doing vulnerabilities. On the plus side, we can address a lot of vulnerabilities in one patch motion.

So, maybe we'll get to some level of even, who knows? We didn't keep up with patching before Mythos, what makes you think we're going to do it after? But anyway, let's jump to our next report, or to our next segment.

And it is- Right ... " Yeah. There's- It is one of Tom Cruise's undervalued movies, by the way.

Right. I really enjoyed it. I think it'll be coming back to theater near soon, because this conversation's going to keep coming up.

So there's a report by a bunch of researchers out of India. Now, on the face of it, you can go, "Well, it's not necessarily ... what I would call rocket science.

So if you look into it, it's basically saying you can use AI to determine that it's more likely there'll be crime in a certain location than another one. Well, I could've told you that living here in New York, that it's more likely that you might be mugged at midnight in the South Bronx than you would be walking around Staten Island. But, at some point though, we're going to have enough data to look into this and say, all right, well, maybe we can get more granular and do I need to put two cops on the corner of randomly picking 184th Street and Jerome Avenue because I know that's where the action's going to be?

Or, does that become too much where we're now allocating police forces on where we think there's going to be a crime, and won't the crime just move to somewhere else if we do that? So Allen, can AI fight crime or not? Can AI help us fight crime, I think is the answer, not can AI fight crime.

Can AI help us? And can it help us? Absolutely.

So the specific things here, like take body cam footage and create a police report from it. I think that's a good idea. " Right?

The body cam footage is right and the way you wrote this up is right. The problem is, and it's much like the problem with Mythos and this stuff, it depends whose hands it's in and do you trust them or not, right? " Right?

Mm-hmm. To be using AI to decide whether I was thinking about committing a crime or what that body cam footage showed or anything like that, right? And just because I happen to be...

We start getting into racial profiling or how I'm dressed or why I'm in this neighborhood at that time going into this store, and all of a sudden, I'm more than likely to be a criminal. That's a scary proposition. As four white men sitting here, in all honesty- Yeah ...

I'd love to hear what a Black person or another minority who have been historically targeted- Right ... think about that. This is not theory either.

" Mm-hmm. Give them a bunch of AI tools- Yes, you should ... the wrong incentives, and all kinds of bad things can happen.

Yeah. I think about law enforcement and I think about they traditionally look at historical analysis, human intuition when they deploy their patrols. But I think they also bake into it some bias, like a lot of bias, and I can't say that because I'm white, but I think this has been proven to be the case, and if it's in the wrong hands and it's deployed in the wrong way, you're basically opening the floodgates for irresponsible or rogue cops to do these real virtual simulations of environments and going after people.

When I thought of this, I thought of, and it's obviously "Minority Report," but I also think about all the like "Training Day" and all these different types of cautionary tales about police forces. And I'm not anti-police, but I have my misgivings about this technology ending up in the hands of somebody, say, and I used to live in Georgia, like some of those communities. I'll leave it at that.

I'll tell you what I am hopeful for. Theoretically at least, we should have a better conversation about where to apply our limited resources, and I'll give you an example. The safest place in New York City is outside of Yankee Stadium on a game day.

It's basically there are a small army of cops in half a mile in every direction. Try going there around midnight when the Yankees are out of town. It's a completely different experience.

And so, we need to be smarter about, well, where are we applying what resources, when and where to protect what community, and also maybe come to a conclusion that maybe we're just underinvested in the people that we need to actually achieve that level of security and safety. Because most cops they start out with a good idea and they're well-intentioned, but you're asking people who are making, I don't know, 75 grand on average to kind of defend us all from potential- Yeah. I'm going to date myself.

I think about one of my favorite movies when I was growing up was "Serpico," and I think about that. That's like an extreme example, and I know most, this is such a broad generalization, but I see it in my own community. This is a pretty safe area where I live, and I see the cops have a major presence here, and then there are other pockets of towns nearby where there is no presence.

And, anyways. No, but let me be clear here. There's a big difference between crooked cops, cops who are getting paid to look the other way or even to commit crimes on behalf of someone, versus inherent bias prejudices in police officers, right?

Mm-hmm. And I'll go back to my own past, right? I grew up Long Island, and I had several friends on both Nassau County, Suffolk County, as well as New York City police force, police departments.

I'm not aware of any of them taking money to be bribed, a la Serpico kind of conspiracy, killers with badges kind of thing out of Scarface. But I do know for a fact that the prejudices they grew up with carried over into their police work. And when something happened, a crime was committed or something, the fact of what color your skin was or something like that had a lot to do with how quickly they would pull you over or question you or arrest you or assume you're guilty.

And I'd like to think that that's changed since I was growing up, but I don't know all through this country if it, in fact, has. " It's always the variable or the factor that we bring up, I think rightly, is that it depends on the humans who use it, who have access to it, and how they use it. So I think this could be a good thing, but I think we also need to be extremely vigilant about this because I can definitely see instances of this going awry, especially depending on who's in charge, like the administration, the culture, the sentiment towards law enforcement, like the law and order type of attitude versus protecting our citizens equally, which there is a gulf there.

" This report comes from a group of people at the Vellore Institute of Technology in Chennai. I should point out that Chennai is one of the safest cities in India, statistically, and in fact has decreasing crime rates, as do most Western cities. And if you want to do something about global crime, what you have to address are transnational drug dealers and criminal networks.

You have to address the fact that 230 million people in third-world countries live in cities with very little to no police presence. I was just in a city of three million people that has fewer than 1,000 police officers on the road. And let me tell you, crime is not all that widespread there, but that's not because of anything the police are doing.

That's just because the people have just not started committing crimes, I guess. Point is, this is another instance of this sort of scare monitoring marketing behind AI, where they say, "There's a problem. We've got a solution.

" Is AI going to help? No. You know what's going to help?

Kind of the things that were mentioned earlier, paying police officers, getting police officers on the streets, having civil society established in these rapidly growing third-world cities that are actually experiencing crime, addressing transnational drug and criminal gangs. And frankly, I'm not sure that AI is the solution to that, but if you're a computer scientist, maybe you think it is. You know what?

But this isn't really an AI thing, what you're talking about, Steven. I live down here in Florida. In the part of Florida I live in, probably 60% or more of the people came from New York, New Jersey, Boston.

Mm-hmm. A few Detroiters maybe, Philly. If you canvassed those people, I would tell you that 75% to 85% of them will tell you that crime is rampant and growing in the Northeast, especially since that new mayor got elected.

Even though it has dropped by 40% Yeah. last five years. It's historically low crime.

The facts are that the rates are dropping like rocks. But you've got a commander-in-chief in Washington who spouts the nonsense, and it gets picked up down here by the folks in the villages. And so this is the public myth.

It has nothing to do with AI. That's the public myth here. You nailed it.

That's what concerns me is like we have another tool in our toolkit to fight these violent outbreaks. Right, and now this is just going to be used to reinforce it, and that's what I guess. I'm going to go the other way on it.

I don't disagree that this business of, "I'm going to be tough on crime to get elected," has been going on forever. The issue is the politicians never check the numbers after the fact because they would go against the marketing program. But I'm thinking maybe with some help from AI, these numbers become more transparent to everybody and we can have a more rational conversation about this.

Yeah. He who controls the AI controls the numbers, or at least what it reports, which kind of leads me to our wrap-up today, right? Cutting across all- I guess we need open source AI after all.

I don't know. Well, not Chinese, but it leads me to the current across this, right? Control of technology, control of AI.

Whether it's Microsoft, the US government, our local police, he who controls the technology really has their foot on the throat of our society. And I think as AI continues to embed itself in here, that becomes even more and more potentially dangerous and important for us to deal with. We'll be discussing it some more, I guess, though, right, guys?

Hmm? All right. Always bring your skepticism to these discussions.

Never leave it at the door. Steven, John, Mike, thank you for joining. Thank you all for watching.

Of course, Techstrong gang comes at you Monday to Friday every week at noon Eastern, which reminds me, we will not be doing a show this Friday, July 3rd, in observance of the July 4th holiday this Saturday, which also happens to be this country's 250th birthday of independence. And I'll be writing a thing or two about that. But we won't be doing Techstrong gang this Friday.

tv, Techstrong's OTT app on just about any screen you like to watch this stuff on, or our YouTube channel, Techstrong TV YouTube. But tune into the July 2nd episode where we'll be talking about July 4th technology-related issues. There you go.

We might even be lighting a slow fuse. All right. Guys, thank you very- All right, we're done.

Mike, John, Steven, thank you. Thank you for watching. We'll see you tomorrow.