[00:00:18] All right, happy to have you along, my fellow tech friends. I'm Sam Bushman. Jay Hill's with me. We keep an eye on tech so you don't have to. Brought to you by NetworkProvidersInc.com. You've got a partner when it comes to IT, strategic, help desk, security, the whole gamut. NetworkProvidersInc.com. I'm Sam. He's Jay, and we're talking tech. Jay, on the last episode, we talked about America.gov, this AI or SI superintelligent website that's supposed to take over for 29,000 websites. And, you know, that's great. I hope it really works.
[00:00:47] But the problem is I ran my scanner on it to check accessibility, and it came back and only could scan the first page. And it said there's all kinds of accessibility problems because there's some animations and different things like that on there that need to be worked on. And then it tried to scan the other pages and got a 403 error for all of them. Wait, what is a 403? We know what a 404 is. I can't remember. 404 means the page not found. 403 means, hey, this thing is, you know, protected. You don't have rights to access that page. Oh, okay.
[00:01:14] So you've got a website that's supposedly secure and all good and everything else. It's so secure, though, that I can't even access most of it, Jay. Dang. Can you click on the different pages and stuff? Well, I haven't browsed around much on, yeah, on America.gov. All right. Well, anyway, kind of interesting, though, that you get 403 errors and it's inaccessible. I mean, that's the ultimate inaccessible, right? Yeah, exactly. Anyway, I just find that very strange. All right. We talked about the Dutch police, though, and I hope they catch these people.
[00:01:44] Hacker group Shiny Hunters that hacked the FBI. I hope they get to the bottom of that because, man, we just can't have people hacking this kind of stuff. I guess it's repercussions for the FBI going after them for their criminal activity, right? I think that the FBI has a lot of egg on its face on this one, and I think they're going to throw a lot of resources at getting these guys. I think they're a little steamed about it. Yeah. There's no doubt an epic battle throwdown happening now. You'll make the FBI look bad and then get away with it usually. Yeah. There you go. All right.
[00:02:12] Anyway, there's this. I don't know who this guy is. Brian Chesky. He's what? Aaron's CEO or whatever. Airbnb. Airbnb's. Sorry. Airbnb's. He says that AI agents need their own operating system, Jay. Right. What do you think? I don't know. I mean, a lot of these Google or not Google. Chat GBT has just rolled this out with I think they're calling it pods.
[00:02:41] Claude has this. Other things have this where they're spinning up their own virtual PCs. Now, I guess they're running on Linux. I don't know. They're running on some operating system, but they can do these agentic things, be an agent for you, and they can run a browser with your logged in sessions, and they do all this stuff. So when he says AIs need their own operating system, I don't really know what he's meaning by that. They kind of have their own operating system that they're running internally, and if they're spinning up their own VPCs or whatever, they're using that.
[00:03:10] And if they're running on your computer, they're using Windows or whatever you're on, Mac or whatever. Yeah. And a lot of times, believe it or not, they spin up, even if you're using, say, Cloud Code on a Windows machine, oftentimes it'll spin up a little Linux. Linux kind of environment, believe it or not, that it runs a lot of things in, Jay. Yeah. He says his focus says how digital services should evolve as AI agents and take on more tasks for users. And I understand that, and I understand the need to keep things sandboxed and stuff, too.
[00:03:40] But creating a whole other operating system for it, though, I don't know. Why don't I just run it and keep them in check on the ones that we have already? Well, and then some would say, hey, you know, you can already use Docker, which is a Linux container kind of environment where you can literally lock out and create their own, as you mentioned, sandbox kind of protected environments and everything else. Super powerful. So I think we have a lot of the technology for it, and I think things will, over time, adjust. And when we say it needs a new operating system, to some degree, I agree, because most of it is headless.
[00:04:08] And headless means you don't have a monitor or a keyboard or, you know, it just kind of runs autonomously. It doesn't have a head or controller, and it can do things. And so my point is, you almost think of DOS, Jay. DOS was so simple, and everything was so small. That's what they should run it. You would almost fit an operating system for, well, it would be a cross between DOS and Linux in my mind. It would be simple like DOS in the sense, but it would have security like Linux. It might be some of these, you know, they take these portable versions of Linux or mobile Linux kind of things, or, you know, dumbed down.
[00:04:37] I shouldn't say dumbed down, but simplified distributions or hardened distributions and stuff. I think some of that would do really well for security. I think Linux is the operating system to do it. It's already the operating system of the net. And all you got to do is create a flavor designed specifically for AI. And I think you've got what he's talking about already, Jay. Using Docker and equivalent thereof and sandboxing things and then narrowing it down and getting rid of a lot of the bloat and things. You don't need a support for a gazillion printers. You don't need support for a ton of, you know, because it can use other tasks.
[00:05:07] So the question becomes, can it be an operating system that runs its own agents, but yet it taps into the operating system for other resources when needed? If you could do that, you could create a firewall between the two of security. There's a lot to what he's saying. That's exactly right. And I've played with some of the early on agents, ClaudeBot and some of these things, and they did run just from a command prompt or command line interface. And I know they're getting more advanced than that for sure now. Well, now once you set them up, they run and they're just kind of headless as they run. And they don't need a monitor. They don't need a keyboard.
[00:05:36] They don't need anything to do a lot of what they do. Yeah. And they just spawn up what they need. If they need a web browser, they can do it either on your machine or they just spawn it up in the cloud and like VPC kind of thing. It's very strange. And we're going to see more and more of this, folks. And I'm just, here's what I'm going to tell you right now, my prediction. Whatever you think is going to happen, it's going to happen faster than you think. Yeah, that's right. And more than you think. It used to be slower. It used to be when they said something, we used to go, ah, yeah, wait five years or whatever. Now it's not true. And in 18 months, you'll probably see a lot of that already happening because people need
[00:06:06] to get a handle on this and they need to get a handle on it fast. For sure. Because the tech is moving along at breakneck speed right now. One of the examples is Shopify, Jay. Yeah. Shopify is unveiling their canvas. They're going to build and customize online stores where people can just talk to it and they can just explain with their new site. It lets merchants create customized online stores through conversations with Sidekick, its AI agent. Merchants just describe the changes they want and they watch their storefront update in real
[00:06:35] time, making a conversation the direct way to shape their online shop. That's interesting as could be. You watch the site unfold as you type and give it direction and I want this and I want that. And I think a lot of things are going that direction, Jay. And I think that's more and more going to be the case. Remember though, you still need people to kind of make sure that access that your sites are accessible. You're going to need people for kind of custom coding. And it's better. One of the things that I've really learned about AI, Jay, is if you have a lot of knowledge,
[00:07:05] you can really give it the prompts and it can do a ton of great things. But if you don't have the knowledge to prompt it in the right direction, it can be a struggle. It can be not only a struggle, but it can be full of bugs and problems and holes like nobody's business. Yeah, because you don't even, not only do you not understand where those holes are, but you don't understand how to explain it to fix it properly. Well, and you don't even understand the hole exists. So let's just say I say add credit card processing to my website. It'll do that.
[00:07:33] But no one's said, hey, but let's make sure that it's secure. Let's put mechanisms in place to say stop international sales if I don't sell internationally. Right. Or block IPs if I need to. Or make sure that it checks the CCV plus the credit card number plus the expiration date plus the zip code. Okay. All I'm saying is that it might implement it, but it might not care about or take care of those things. And it's not that it means to be negligent. You just didn't ask it that. No.
[00:08:01] See, Sam's supposed to be smart enough to know all that stuff, even if you don't ask for it. I'm just telling you it's not there yet, though. That's what I mean is that, you know, if you have enough knowledge to get it done, you can. So I just give you my example of my website scanner. It didn't know what to do with 403 errors. Why? Because I hadn't accounted a 403 error and I didn't think about it. But now that I see it, I've got to update my thing. What does it do with 403 errors? Does it report them as broken links, Jay? I don't know.
[00:08:31] It's not a certificate thing, is it? Well, it's providing reports. So I wouldn't say they're broken links, though. Yeah. Just because you are denied access to a link, that doesn't mean it's broken. Time to dig into it. So anyway, I use that as a simple example because it's complicated. And I really appreciate that websites are easier to develop that way. But what you're going to find is when you develop a website that way right now, if you're not careful, you're going to have a technical person come and say,
[00:09:01] look, this has got too many problems that I've got to rebuild it for you. Right. So remember that. And I'm not downing these bots. I'm just telling you that they sound like they solve everything. But when you really dig into it, it's not as simple as it sounds. Chat GPT, Jay. You're into fashion, aren't you, Jay? No. You're a fashion guy. I'm not known for being a fashion guy now. The reason he's not on video, ladies and gentlemen, is because if he's on video and I'm on video,
[00:09:25] he'll look like the suave debonair kind of businessman, I'll just look like the redneck guy. And so he's just like, Sam, I'll let you be on TV. And that way, people won't really realize how in style I really am. But Chat GPT is changing this, Jay. If you're not stylish, you can fix that with this plan. Well, they say they're going to make a virtual dressing room with photo-based try-ons. That's you, by the way. Yeah. No, I've never done this. Because Amazon has had similar things to this before and stuff, and I've seen people do this,
[00:09:55] but I don't think it works as well as they do. I mean, I guess they just want to see what they look like in that shirt or whatever. But yeah, so they're going to make a new feature where they can- And while you're trying on all the clothes that plays that song, I'm too sexy for my shirt. For my shirt. I guess. Right? Is that what it does? No, I'm sure it doesn't. Oh, good. So you can upload a photo of yourself. You can tell it products or whatever, and it will create AI-generated previews. Do I need to give it things with my photo, like little tags like fat, short?
[00:10:26] I don't think so. No. Okay. It'll figure that out all on its own, Sam. It doesn't need a prompt for that. Okay. No. But yeah, you know, so you can make- I think it really could do this before. I just think they're making it more of like a separate, standalone, you know, AI dressing room thing. Again, and it's more- Every time they do another round of this, Jay, it becomes more AI competent. But even a year ago, you could have sent in a photo of yourself standing up and said,
[00:10:54] hey, and here's a picture of a shirt from a product website or whatever, and say, hey, make a picture of me wearing this shirt. And it would have probably done it. Yeah. But all I'm telling you is it's becoming easier, slicker, and better at it than ever before. Yeah. And this generation is going to be better, and then the next generation even later is going to be better. And it's something that we kind of laugh at at first because it's just so awkward and so goofy. But I'm telling you, over time, I think it will be a thing. And over time, it will become quite accurate, Jay. Yeah. I think it'll get better and better, as we always see with these things. All right.
[00:11:23] Open AI launches. What do they call this thing? Revolutionary AI, Jay? Well, they're calling it DOTS. It's the Revolutionary AI Assistant called DOTS? DOTS, D-O-T-S. DOTS. Wait a minute. DOT is that Amazon Alexa player thing, right? They did have a DOT. DOT. Okay. This is DOTS with an S, though. Or a spot or something. Yeah. But this is DOTS, plural. And this is coming out from ChatGPT. We talked about this a little bit in a previous episode.
[00:11:50] But they are basically making interactive digital assistants by performing tasks autonomously. They announced in their new developer conference that DOTS will compete with recent and similar introductions from Meta, Google, and AI startup instinct. This release comes as OpenAI addresses increased safety scrutiny. And unlike its competitors, OpenAI emphasizes DOTS will have enhanced capabilities for robust safety measures.
[00:12:21] Aligning with consumer expectations. The service is now available to select subscribers. And what this is is this is basically AI running an agent for you. So you could say, hey, check my email every hour or whatever. And if you see, for example, this is one of the examples I saw from them. If you have, you know, check your email, check your GitHub, check your whatever. If you see several complaints coming in from customers about a specific bug, you know, go ahead and look that up. Test it. Act on it.
[00:12:50] Write a patch for it and send it to me so I can approve it. And these things will run continuously. It's not something that you have to continue to prompt or ask for update. These will run in the background and look for things and do things on your behalf. Maybe they sort your emails. You know, I was talking to a guy that's been working with GrokBot. And it's like a $30 a month plan or $300 a year if you pay for it annually. And, you know, he said he had like 40,000 emails in his inbox. This is a personal inbox. He just gets behind.
[00:13:20] He doesn't really use it that much. And it was a Gmail account. And he had it just run through that thing. And overnight, this bot on its own went through and filed all this stuff. He put it into folders. It said, hey, all this stuff is safe to delete. This is some stuff you may want to archive. You know, all these files. And these, you know, whatever, a couple hundred or maybe not even that much, maybe 50 that were like, hey, these you kind of need to act on and they need replies. And I went ahead and drafted the replies for you. You just need to go through them and preview it and send them if you're ready.
[00:13:50] And he was blown away. I mean, he was just like, this is incredible. This thing did weeks' worth of work overnight in a single day. And he doesn't even know how long it took. Maybe it only took an hour or two. But it just did all this work. He woke up in the morning and all of his inbox was squared away. It was incredible. Incredible. And I think that that's a panacea, that it's going to be that simple and that easy and that holistic. I do think that this stuff is real. And I've been building bots and agents to do a lot of this stuff for some of the employees that I consult for now.
[00:14:18] And I've used, believe it or not, Fireflies for a lot of that. Fireflies.ai, which basically is a note taker in your meetings that joins your meeting. And Fireflies sits there and records your meeting. Then it can provide summaries and to-do lists and action items for people. And then it can literally send it to Slack and different places. And then I've got a bot that can say, hey, go read my email. Go see my meetings. Go see my Slack.
[00:14:42] Give me a summary of what's going on, what I need to do today, what meetings I have, what to-do's, who needs to be responded to, and what are the top priorities. And it comes back and gives you a summary of all that. It's like an e-butler just taking care of everything for you. Yeah, man. It's like the Jeeves of our time, right, Jay? Yeah. Remember Ask Jeeves, the old website? That's why I'm bringing it up. I actually went and saw Joseph of the Technicolor Dreamcoder. Great, great play.
[00:15:09] And it had a lot of great modern-day theater, theatric stuff that was really cool and everything else. But anyway, in there, one of the songs that says, he's the Jeeves of our time. And so that's why when you said that, I'm just like, okay. Kind of funny. We got a break and then we'll come back, right, Jay? That's right. All right, ladies and gentlemen, I'm Sam Bushman. He's Jay Harrison. And this is Tech Watch, where we keep an eye on tech so you don't have to. Brought to you by NetworkProvidersInc.com. If you go to NetworkProvidersInc.com, you can get a free consult.
[00:15:39] You can also sign up for our newsletters and get tech tips and more. And best of all, you can listen and watch the archives on that site as well. I'm Sam Bushman. He's Jay Harrison. This is NetworkProvidersInc.com's Tech Watch, the podcast for those who want to keep an eye on tech with us. Bypass the mainstream narrative with Liberty News Radio at LibertyNewsRadio.com. Engage with charismatic hosts live or on demand.
[00:16:08] We cover the crucial news focused on God, family, and country. News that other networks simply refuse to use. Think of LNR as hard-hitting news and podcasts at your fingertips anytime, anywhere. Join us at LibertyNewsRadio.com. Empower your day with the truth. Because the truth will set you free. LibertyNewsRadio.com. Cybercrime is exploding.
[00:16:32] Take Sarah from Sweet Delights, whose world crumbled after having to close with not being able to bounce back. Small businesses are prime targets, but the right strategies can keep yours safe. Jay Hill, CEO of NetworkProviders, has co-authored The Cyber Playbook Simplifies Cybersecurity for Business Owners with Strategies to Avoid Costly Breaches and Fines. Build a strong cyber attack response. Secure your business with key protections.
[00:17:02] Cyber threats aren't slowing down, but you can stay ahead. Protect your business. Ensure its security for tomorrow. Get the Cyber Playbook today at NetworkProvidersInc.com slash cyber dash playbook or call 385-446-5500 now.
[00:17:36] All right, ladies and gentlemen, Sam and Jay, this next tidbit is a little bit scary in my opinion. Headline says, NVIDIA. And NVIDIA used to be a no-name company, really. I mean, if you were a gamer or something, you knew about NVIDIA. They make incredible video cards and stuff like that. And unless you really needed a defensive video card, you didn't really know about or hear much about NVIDIA. But now that AI is here, everybody needs a great video card because it uses GPUs and, you know, the processing cards that NVIDIA has does that really well.
[00:18:04] So they've become a rock star of recent, and their company has just exploded. And now it's a mammoth, mammoth company. And they're virtually at the center of everything, AI and everything important nowadays. So anyway, NVIDIA is concerned about all this hacking that's going on where these agents go rogue and hack all these sites. And it's a big problem that everybody's talking about in the mainstream press right now. So NVIDIA wants to have an answer for that. And here's their answer. NVIDIA announces a, quote, security platform for AI agents.
[00:18:33] NVIDIA unveiled an AI platform that includes a chip level monitoring layer called Sentry. And they say the company says it can intervene when an AI agent acts outside its, quote, assigned task. That's a company description of how the new product works. Not an independent, established safety result at this point.
[00:19:01] But, Jay, my concern with this is who's to decide when my agent's off task? Exactly. And is this going to put NVIDIA really at the helm of all things and say NVIDIA decides everything? Yeah. I'm just telling you, I'm super uncomfortable with this whole discussion. Well, you know. I get the concern and I get the desire for a response. But this is not the response. It is a concern.
[00:19:27] But if they control the guardrails, they can say what you can not research. I mean, what if you're trying to research a new graphics card to compete with NVIDIA? Is it going to stop you from doing that? You know? Who sets the rules? Like, are you going to be able to edit them? It reminds me also, it is kind of a good idea. But, you know, in automotive manufacturing, you have these robots that are doing all this crazy stuff, moving in and out. They'll kill you if you get in the way.
[00:19:51] So you have, not only do you have the robots, but then you have robots that monitor those robots to make sure they're not getting out of line or, you know, doing something wrong. Because if something does go wrong. It's like a 3D print. You know, when your 3D print goes out of wire, the whole thing just goes haywire. It just messes all over the place and makes a big mess. And the robots will do the same. And so I think it's cool to have. Yeah, people do that sometimes too. Yeah, that's true. But it's cool to have these guidelines. It's just who's going to be in control of it. Is NVIDIA going to make the final say on it? Are you going to be able to read the rules?
[00:20:19] Are you going to be able to see what those guardrails are and or be able to adjust them if you want to? You know, those questions remain. But it is good that they're working on this. And, again, what I think we need to do is I think we need to have a private consortium, you know, that can put together guidelines that make sense, that don't lock people down, that don't deny people freedom, but yet do protect, you know, there's a balance there, right? With all these things, there's always a good balance. And some people want to take it one direction too far. The other people want to take it the other direction too far.
[00:20:47] The answer is something in the middle that makes common sense, that protects people, that yet doesn't deny people freedom and does create a process in which people can be represented. And when it gets it wrong, there needs to be a way to mitigate that, right? Yeah. And, you know, NVIDIA had this issue where after the Hugging Face breach, they wanted to implement safeguards. Not only are they using the Sentry thing, they're also using – Yeah, they need to rename Hugging Face to punch you in the face though because that's what it kind of did, right? That's kind of what it did.
[00:21:13] They've got this open shell software that's limiting what agents can access along with the Sentry system that's designed to detect and quarantine agents that are crossing boundaries. And so they say the new safeguards are going to help. They're working on both sides of it. So they're working on both inputs and outputs. Yeah, and where they say it's like an open shell or whatever else, so I assume that other people can interact with it and everything else. Look, if they do it right, it could be an incredible thing, folks. Really, that's the bottom line. Is there a need for this?
[00:21:43] Absolutely. And so the question is where that balance is and who's going to, you know, be involved in that balance and how does that work? And that's where we really want to know. But the U.S. military has to be very, very careful too these days, Jay. They do because of tracking links and tracking ad IDs on government devices. Apparently, the military has disabled advertising identifiers on government-issued devices after warnings that commercially available phone location data could expose U.S. personnel in the Middle East, like their movements and where they are and where they're being stationed.
[00:22:12] The Army says that the setting has been off by default on Apple and Android devices since last February of 2026 and that people can limit this form of tracking on their own phones. iPhone has a tracking setting. Android has the option to. The step doesn't eliminate all tracking, you know, because they can click on links and do other things. But it does prevent apps from collecting location data through, like, default permissions, which is a good thing. I think everybody should do that, really. Yeah. Anyway, I think everybody should do that.
[00:22:41] The problem is it's easier said than none for most people, right? Not really. It's very, very easy. Do people put that slight resistance of, I've got to learn about this. I've got to follow, you know, five steps in a tutorial to go into settings and go into. But and even that I've got to find the five-step tutorial and then follow it. And if anything's changed on my phone, it doesn't match. It takes about one gram of resistance to stop people. So it doesn't. Well, that's the point.
[00:23:08] And the problem is even if I do get the five-step list, if my phone's outdated or not the right phone or and then you say, OK, I got to give you 40 different instructions for the different phones. Now I got to try to find my phone and my phone's not on the list. And it's all that, right? Yeah. Well, and not only that, but a lot of times we see this on Windows all the time. This bugs me. But Windows, you know, you uninstall Solitary. Here, you uninstall OneNote. You uninstall whatever. The next update comes through and they reinstall it all. They put it back. They put OneDrive on your computer and they do this. Yeah, don't they understand? I don't want that. That's why I uninstalled it. Apple will do the same thing.
[00:23:34] It may not actively put it back on a phone that you disabled it, but sometimes you upgrade to a new phone. And if you don't go through the upgrade process exactly correctly, the new default may be back on for ad tracking. And so there's these things where people kind of get stymied, even if they've done it one time. They've taken the effort and the time to disable this thing. And then next thing you know, they turn around and it's back again because that's the default. I understand. Man, anyway, very interesting.
[00:24:02] And I agree that it's easy, but it's only so easy. Hey, so you got this Plod thing. We've discussed this several times over, I think, the last several years, Jay. It's PlodNote Pro. Yeah, very cool to like it. It synchronizes meeting transcriptions with AI precision, they say. And the CloudNote Pro is an innovative AI-powered device. It's tiny. Anyway, the reason I bring this up really quick is I saw one yesterday. It was the pen version. And this guy wanted to pick my brain as an IT consultant. And he's like, can I record what you do?
[00:24:32] And he actually had one of these Plods. And he showed it to me. And I checked it out. Pretty neat device. I said, yeah, send me the transcript. So I'm waiting on the transcript to see what it shows based on what we talked about. Very interesting stuff. I'll report on that soon. All right, thanks so much. Make it a great Tech Day, will you?