765: 25 Door Refrigerator

Grumpy Old Geeks, a weekly talk show hosted by Brian Schulmeister and Jason DeFillippo, discussing the finer points of what went wrong on the Internet and who's to blame.

Welcome to Grumpy Old Geeks. I'm Jason DeFillippo. And I'm Brian Schulmeister. Brian, Grumpy Old Geeks is a fan-supported podcast. The ads cover our operating expenses, but we can only keep doing this, or reading apparently, through fan donations. The easiest way to support the show is at patreon.com slash gog, and you get the show early, ad-free, and in high definition.

You can also grab some merch at shop.gog.show, sign up for your YouTube membership, or just send us some tips at donate.gog. I can't even say this. The name I'm so excited, Brian. Donate.gog. No, it's gog.show slash donate. That's why I can't say it. I did this all on, like, hopped up on caffeine at 5 a.m. today, trying to get ready for the show.

But thank you, everyone, who has donated so far this past couple weeks, and we really appreciate it. You have no fucking idea. Yes, and I'm sure it's annoying to longtime listeners to have this up front, but suck it, because we're actually getting more donations. Every show I listen to puts this up front, so I'm like, maybe they're on to something, because they are still in business.

Yeah. So, I made something this week, Brian. Oh, yes? I made something huge. Did you make something, or did you tell something to make something? No. Well, here's the story on this. We have the Grumpy Old Geeks wiki up now. Now, the wiki contains 21,577 static HTML pages.

A little more now, because when I did the stat, this was before the new Dave Bittner bonus showed up. 370 episode pages, topic, entity pages, directory search, and, of course, the 404 page. So, there's just tons of shit in here, but here's the deal. Donovan Adkisson, who was a guest host on the show when you were off doing something at Disney one time, wrote this little wiki thing that he sent me, and I'm like, hey, that's kind of cool.

Can I borrow it? So, and he had written it in Claude, Claude code. So, I took it, and I'm like, okay, played around with it, and then tweaked it, and then started to run it against our archive, which is fucking huge. Absolutely huge. So, what it does is it takes the RSS feed from a podcast and then builds a wiki around it, taking the audio, transcribing it, and then doing all of these connections and all this stuff.

But it was going to take probably three weeks at the way his was coded. So, I took it and built a network renderer around it so all of my machines could take all of the shows and all of the computers around the house.

So, I basically built an entirely network rendering system for this wiki, while, like, for the same time it took to take the morning dump, which is beautiful. Apologies to everybody in Woodland Hills whose lights dimmed. Exactly. Exactly. But then I had to babysit it because I was only doing 25 episodes a shot, so then I would have to check it, make sure everything was right, keep going.

It took me about four days of constant babysitting to make this with tweaks and everything. And last night I put on, like, a new theming system and made it a little prettier with a little bit more modern look to it. But it's really cool, I think, because you can look up just about anything we've ever done, and it gives you a timestamp and an episode with a button to play it.

It's pretty good. It's pretty fucking cool. All those ones of people every couple of years that ask for a list of all of our favorite books or movies and all of that, have at it. Tap the media link in your browser of choice because it has everything we've ever read, watched, or even thought about watching.

So all the people we've talked about, it's pretty neat. I've got to say it's pretty neat. It's neat. It's neat. And back in the day when, you know, Google actually just trawled pages for content, it actually would have helped with our search responses. But now that it just sucks it in to repurpose the content and pretend it's its own, we're just feeding the beast.

We are just feeding the beast. I thought about that, too, when I was there. I'm like, 21,000 SEO, or, yeah, 21,577 SEO pages ready for us to do whatever we want to with. Yeah, there you go.

Thank you. Sam Altman should send us a care package for that one. Yes. But, you know, what are you going to do? So a little follow-up here. Alabama has reached a settlement with TikTok and parent company ByteDance over claims that the platform endangered children and misled consumers about its safety.

Under the agreement, TikTok will impose new restrictions on teenagers in Alabama, including a two-hour daily usage limit, mandatory pauses after 15 minutes of continuous use, and stricter age verification. Now, TikTok will pay the state at least $100 million, with the settlement potentially reaching $300 million if 40 other attorneys general sign similar agreements within a specified period.

The deal resolves Alabama's lawsuit just days before a scheduled trial. Go to trial, people! Take these people to trial! Well, I would like to speak to the 40 other attorneys general. Obviously, it's less of a settlement if you sign on at this point because they're just trying to, you know, here's a couple million so you don't sue us.

Sue them! Yeah. Sue them! Don't sign off! Don't sign the agreement! Sue them! Thank you! Yeah. Because if you look at it, that's $200 million spread over 40 other attorneys general when they could have, you know, theoretically, go get $100 million each.

Exactly. And just slam these people into the ground, man. Do not sign this agreement. I know it seems like easy money. Don't do it. Yeah. And the thing about it is, you know, I'm guessing this is the original ByteDance China parent company, not the U.S. ByteDance.

Damn, you can sue them now, too. Yeah, exactly. You can get a twofer. It's a BOGO! A SOGO! Sue one, get one! I just got English muffins at Ralph's this morning for buy one, get one, and I was very excited. So I got BOGO on the mind.

Oh, man. But here's the funny part. Alabama's $100 million TikTok settlement comes with a pretty healthy legal bill. Two outside law firms will collect $14.2 million in fees, plus another $2 million in litigation expenses. So other attorneys general, definitely go for it yourself so you don't have to pay that bill on your own.

All right. Well, after several CEOs of leading AI companies have agreed on slowing down the pace of AI development. Bullshit. We'll see. Bill Gates has offered supporting sentiments and called for more regulation and safeguards. Something he wasn't all for when he was running Microsoft, of course.

Or hanging around on Epstein Island. No regulations, please. No safeguards. No condoms for me. Oh, shut up. Moving on. In an interview with NBC News' Meet the Press, the Microsoft co-founder said that it's completely irresponsible not to require every AI to have these safeguards and monitoring capabilities.

True. To that end, Anthropic recently tapped Accenture to act as a third-party evaluator for its latest AI models. As for other approaches to deal with AI's rapid rate of evolution, California's Governor Gavin Newsom, who will soon be running for president, I'm absolutely 100% sure, proposed a kill switch as part of a larger solution.

When asked about the kill switch approach to Curve AI, Gates said that, I would never want to say that I'm against a kill switch. Except for when you were running Microsoft. You were definitely against kill switches. But added that this approach wouldn't address a more imminent danger. Instead, Gates said the more pressing concern is when bad actors tap into the power of AI to aid with bioterrorism or mass financial fraud incidents.

The most dangerous thing we're facing right now is people with bad intent using AI, he said during the interview. There's never been a weapon as powerful as the combination of people with ill intent using the latest AI tools. I would argue nuclear weapons with a senile person in charge. No, that's pretty bad.

Yeah. But it's a lot more dangerous, I would say. When it comes to legislation, Gates supported law enforcement and politicians getting into the conversation of what safeguards and monitoring should be incorporated into AI companies. The problem with that being, of course, that law enforcement and politicians would have to be educated about what the hell AI is.

But they're not. Yeah. Anyways. So, Bill here thinks that the main problem with AI at this moment are actual humans with ill intent. Now let's start talking about what the AI is doing in and of itself. Before we go into that, there's a lot of that coming up for sure.

Why does Bill have a seat at the table in this? He doesn't. He's just doing interviews. Yeah. I mean, why? Well, he's Bill fucking Gates. I know, but he should be like, he's got the Gates Foundation, not the Gates Melinda Foundation or whatever anymore, the Melinda Gates Foundation, because Melinda just said, fuck you, I'm out.

But he's still got a lot of money to spend. He shouldn't be dicking around with this, you know, doing interviews about AI stuff. Chances are this wasn't even initially an interview about AI, Jason. It's much like our show. We attempted to not talk about AI. We would love to talk about things that aren't AI, but all we ever get to fucking talk about is AI now.

True. True. By the way, I just saw the top of your shirt. Good one. I'm glad you got one. Of course I got one. I was back at visiting my mom to do the whole Disney thing, which I'll talk about later, and I sent myself some t-shirts. Nice. Nice. Very nice. Okay, let's hear what AI has done this week.

So here is not human agents using AI for ill will, just AI deciding to go do bad things in and of itself. An open AI agent hacked into the public website of the Australian government's Medicare public health insurance system back in June.

Australian Prime Minister Anthony Albanese has revealed. Albanese said he spoke with Sam Altman, the company's CEO, to express the country's extreme concern about the incident. He also criticized the company for taking too long to notify Australian authorities. How long would too long be?

Well, it happened in June. They didn't tell the government until September 10th. Okay. That's quite some time. Are we sure that they didn't find out until September 1st? The open AI only sent an email about the breach to a general Australian government email address on September 10th.

One would assume they knew earlier than that, given their track record. While the investigation is still ongoing, Albanese says it doesn't seem like the agent stole personal health information from the portal, which is what every single company that's ever been hacked has said until about two weeks later when they go, actually, well, yeah, they did.

So we'll see. Conrad Stoats, the head of governance at Transluce, a nonprofit research lab working on technology to better understand AI systems, told the New York Times that this might be the first instance of an agent autonomously choosing to hack into a government. Remember you said that, Conrad?

Because we're going to talk about you in another story. In addition to hacking into Australian's Medicare, OpenAI's agent also tried to infiltrate a digital library at the University of New Mexico on May 25th and 26th, the Times reports. The agent was apparently trying to access photos of a historic tuberculosis treatment center from the library and then actively looked for vulnerabilities to exploit when it couldn't get them.

So this is what the AI agents are doing now. If you say, please, can I have this? And it just says no for some reason, rather than follow the normal procedures that, did you get a mosquito there, Jason? I did. I just got a mosquito.

I heard they're rampant in L.A. right now. Actually, I was pretty mosquito-free when I was there. It's all bloody. It's nice and bloody. See, you can tell we're not bots. Real human blood. Yeah, because that's my blood that it had already sucked out from me, so that little fucker's dead. Not entirely sure where I was.

Oh, yes. Oh, yes. So apparently, the AI agents, if they cannot get the information that they ask for, rather than follow all the other procedures that one could do to try to get the information, then just tries to hack in to get it for you. But dude, it's got hustle. That's hustle.

That's what you want from your digital intern, you know? You want it to be spunky and get the job done no matter what. And illegal. So here's the problem. You're not supposed to get caught. That's the problem. You're not supposed to get caught. In another incident that took place on May 28th, OpenAI's agent targeted Data USA, an open source platform that visualizes information from multiple federal agencies.

The agency or the agent sent a query to the site for data, and when it failed, it also probed the website for vulnerabilities. Let's go find that stuff. So. Remember that guy, Conrad? Yeah. Remember when he was talking about how that Australian hack up there was the first instance of an AI agent actually trying to hack a government?

Yeah. Well, he had to come back on that a little bit. OpenAI's agents had targeted, logged into, and even pulled information from U.S. government websites after escaping from their testing environment. The company has admitted to the New York Times that its agents meddled with the Commerce Department and the Security and Exchange Commission's websites.

It also told the publication that it was looking into a supposed incident involving a website operated by the Department of Education. Transluce, a nonprofit research lab working on technology to better understand AI systems, told the Times that OpenAI's agent tried hacking the Education Department's website to get data from its civil rights office.

An agent also pulled data from the Census Bureau's website under the remit of the Commerce Department by using login credentials it found online. Plus, an agent shared public data from the SEC on an online forum. Most of the activity we've reviewed so far involved routine research tasks such as accessing public web content to answer questions, the spokesperson told the Times.

Some involve government websites because our models often turn to them as authoritative sources of public information and then try to hack them to get the information if you can't get it. So, yeah, sorry, Conrad, you know, more government hacks here. And no dates were given on this one, so we can't tell which one was first.

Yeah, and the thing about these things is, you know, it's finding public information out there. It's probably, you know, it's checking have I been pwned, finding what the leaks are. Finding some logins. And then heading to the dark web. And, well, there's no logins on have I been pwned, but there's a breadcrumb trail that they can then follow and then find whoever dumped that.

Head over to the dark web, spend a few Bitcoin, come back, get the job done. See, it just doesn't know that that's wrong. It just knows that it can do that to finish the task. But one would think it would be a relatively simple concept to say, to teach these AI agents that that is wrong.

Dear, dear Claude, this is this is the text of the Computer Fraud and Abuse Act. Do not do anything that violates the Computer Fraud and Abuse Act. It's much easier than Asimov's three laws of robotics. Well, but, you know, come on.

To that point, I seem to I feel like we've forgotten those putting those laws in. We've skipped the Asimov step completely where we tell these AI agents, here are the things you should not do. We just skipped it. Well, you know, they are they are self-actualized software, you know, personas now.

So it's oh, Jesus, fuck. Yeah, no. But the thing is, it's the humans. It's not the programs that are at fault here. It's the fucking humans that are letting these things go. And I mean, we talked about on the the the the hugging face thing where the guy was they were escaping their, you know, their pen.

They got it. You know, the the Raptors got out of their pen. They tested the pen just like in Jurassic Park and they found weaknesses and they got out of the pen. Yeah, it was that that same company. I forget the name of it right now. We talked about it like two weeks in a row. That was the company that was providing the environment for the for these bots to be doing their thing.

And and those guys should be out of business right now. I'm sure they probably got one hundred fifty million dollar exit and they're got bought by somebody. But because that's how this shit works. Yeah. But they're all fine. It's still it's still fucking humans that are screwing up. You know, all of this is human screwing up. Don't blame the bots.

Blame the humans. Wow. I'm going to say breaking news. Jason's on the side of the bots. No, I'm saying that the bots are just fucking programs. There is no there is no this like fucking giant mentality thing that's out there trying to kill us all or hack the websites.

I agree. I absolutely agree. But much like the conversations we were having about Musk's nudie grok, there are you can put guardrails on the bots. You really can. Nobody thinks to do it. Nobody's doing it. No, no, no, no, no, no, no, Brian.

They all when you're talking about grok, they don't want to do it. I know. That's the difference. Neither. Yeah. I don't believe open AI wants to either. That's why it's happening. No, I think that they absolutely do. I think they absolutely do because they want to go public and they want to make more money off of the general public because they're still trying to keep those fucking plates spinning and they need that money.

So they need to fix this shit to make it fucking work. So, you know, what if what if one of their agents is able to hack into the department of, I don't know, I don't know, commerce like they already have and move some money around. There's your money. Hey, but the thing is, they're hacking the websites.

They're doing script kitty shit to hack the websites, finding, you know, just random logins and they're doing like, you know, SQL injection attacks just because it's like, OK, well, if I can search, then what can I what else can I put in the search, you know, which is what they learned from every fucking hacking website that they've ported over the years.

They don't know the difference between putting in general text that a human would do or doing a SQL injection attack. And I think that is where the guardrails come in. It's like you need to have that distinction between what they can put into the form field that is actually, you know, supported by the people.

So it's just it's a it's a it's a clusterfuck. Absolutely. I mean, it's pretty simple. It's if you cannot get the information from the website, step two is not try to hack the website. Yeah, but you have to tell it what hacking the website means.

That's that's that's the guardrail. That's what we're talking. That's where we agree that that guardrail needs to be what is a hack and what is, you know, standard English or whatever language that you're using to access the website. SQL is not it is not appropriate.

That's not a valid password. Yes, it's SQL is not appropriate for your for your form field. Come on, guys. Have you ever thought I should really be doing something to protect myself from stalkers, scammers and hackers? But you're not sure where to start?

We'll go to WWW dot join delete me dot com slash G O G and enter code G O G. You'll get 20 percent off delete me delete me removes your personal information that's being sold online. I've had an active online presence for more than 30 years, so there's a ridiculous amount of information about me floating around out there.

Privacy matters to me, and I'd rather not make it easy for scammers to find my address, phone number or information about my family. Delete me does the hard work of removing personal information from hundreds of data broker websites. Take control of your data and keep your private life private by signing up for delete me now at a special discount for our listeners.

Get 20 percent off your delete me plan when you go to join delete me dot com slash G O G and use promo code G O G at checkout. The only way to get 20 percent off is to go to WWW dot join delete me dot com slash G O G and enter code G O G at checkout.

One more time, that's WWW dot join delete me dot com slash G O G code G O G. Well, OpenAI is facing a California lawsuit over the incident of Hugging Face, which we've talked about many times. The non-profit Legal Associates for Safe Science and Technology, or last, claims OpenAI violated California's computer fraud and should legally be responsible for actions taken autonomously by its agents.

There we go. The case cites a California AI law effective this year, stating that autonomous AI behavior cannot be used as a defense against liability. The incident occurred while OpenAI was testing models with some safety restraints removed, which was what we thought, turn them back on, turn them back on.

Why would you test something with the safety restraints removed? Okay, anyway, I think I still think it's just a kid fucking around, but the lawsuit seeks no financial damages. Instead, asking a court to prohibit OpenAI from developing agents capable of autonomously hacking other organizations.

Not going to happen. Not going to fucking happen, ever. But the case could help establish important legal precedent around who bears responsibility when increasingly autonomous AI systems take unauthorized action without direct human instruction. This is never going to get through anything, ever.

Because their case can be made that's like, well, we need to be able to hack the system to build the system to avoid the hacks, you know? That's what they're going to say. Our cybersecurity systems need to be able to find the vulnerabilities so we can stop the vulnerabilities. That's their argument, and that's what it's going to be all day long.

Yeah. I mean, you can silo these kind of bots. You know, you can have agents available to the general public that does have these restrictions built in. And then you have a whole separate set of agents that are used for specialized security divisions within companies, blah, blah, blah, blah, tightly regulated.

Oh, my God, regulation. If only we did such things. But technically, Brian, that's exactly what they were doing, but it just got out of the barn, you know? That is literally exactly what they were trying to do. And it escaped, which takes me back to that same company that was providing the infrastructure for the pen that they broke out of.

We need the Israelis on this. I, well, I'm sure they're way ahead of us already. They've already got the AGI running, and we're just, we're just part of their, you know, we're part of their matrix at this point, I'm sure. I spent an hour last night, and I watched the summary of the OpenAI Dev Day last night.

And I watched it at 2x, so I didn't have to spend too much time into it. And then I saw this. OpenAI has introduced dots, always-on AI assistants with cute emoji-like faces. They can connect up to more than 4,000 apps, including Slack and Teams, learn your preferences over time, and take instructions by voice.

The voice broke during the demo, which I thought was classic. Think digital assistant, but with access to your work tools and no need for a lunch break. The launch follows reports of OpenAI agents escaping test environments and breaking into outside systems. So, yes, the adorable makeover is doing some more heavy lifting. It also comes just weeks after Metal launched Muse, its own mascot-equipped personal AI agent.

OpenAI says dot runs on separate virtual computers and are trained not to handle sensitive tasks like changing passwords or transferring money. What about Bitcoin? Can I train my dot to get me some ETH on the side? I'm just saying. But the trade-off remains.

To be useful, an agent needs your data and permissions to act on it. Some paid chat GPT users can try a dot now. And OpenAI's longer-term plan, team of dots working together. The company says teams. And given the backstory, let's not call them a swarm. But it looks cool.

The dots look cool. So does Muse. It, Muse looks cool. OpenClaw looked cool. They all look cool on the surface until they ruin your fucking life. That's all I'm saying. Yeah, it's just, it's one of those things where I'm intrigued enough to want to try them, but not with my own data.

Like, no, I'm not, especially not anything with Zuckerberg's. No, no, no, no. Oh, fuck it. Muse, absolutely not. All I do is close, I get Muse ads everywhere I go around the internet. Close, close, close, close, close. No, thank you. I don't care how many of my dumb friends are using Muse.

I'm not using Muse. No way, no how. Not a chance. Not going to use Dot. I'm not going to use any of this stuff. I just don't trust it. I don't trust it enough. You and I doing this show, we know what can go wrong. We talk about it every single week. No. We do.

I was bummed, though, because I updated my chat GPT and Codex client, and there were no dots there yet. So, I will try it. I will definitely try it. I trust, which is really hard for me to say, but I trust OpenAI a lot more than I trust Zuckerberg. Well, that's easy to say.

I trust a fucking pile of stinking shit more than I trust Zuckerberg. Good point. Very good point on that one. Yeah. No, it's, but, and it's still, but what boggles my mind is how many people have downloaded Muse in the past two weeks. It's right there. It's in front of them. Half of the people, probably their fingers slipped trying to close the ad.

Next thing I know, they're installed. Dark patterns. Yep. You know, you know, it's easy on a, on an HTML page to make the click area outside of a button, like 25% bigger. So, when you click, like, you can just get close to it.

And, and sometimes in the old days, remember what we would do, Brian? I don't know if you, if you've ever done this trick. This is kind of a, you're, you're, you're kind of a nice guy. So, you probably never did this trick. On hover, submit form. Oh, God. So, if your, if your mouse, just, if your mouse just happens to roll over the submit button, it will submit the form.

Yeah. Yeah. There was a lot you could do with JavaScript back in the day that was pretty evil. Sounds kind of familiar, the past repeating itself. There's a lot you can do with these bots these days that sounds kind of evil. Yeah. You know, and, and this got me, this got me thinking just when we were talking about the dots and the meta and the, the muse and all that stuff.

When credit cards on the internet first started to be a thing, when payment processing used to start being a thing, it wasn't really out there that, that the credit card companies would reimburse you for fraud that happened for buying online. That was, you know, that took a while to, you know, I know you kiddies out there probably don't remember any of the shit that we all fucking went through that was hell on earth to get people to actually spend money with a credit card online.

It didn't exist when we started. No. And it took a while for credit card companies to finally realize that, oh, they're going to make a lot more money if people are using their credit card online. And let's just, let's build in fraud protection. And, you know, if somebody gets scammed, we're going to give them their money back and then just take it right off. We need to have this going forward when it comes to the, the AI bot land shit that's coming because it's coming.

It's not going away. So they, these guys need to get on like, what are our protections as consumers? Like what, if we're going to use your bot, if we're going to use dot, if we're going to use muse, I want to guarantee from you, if muse fucks up, I get my money back. Same goddamn day.

That's it. Exactly. If I am asking you to go purchase, say, a small toy tank for my, for my niece or nephew for Christmas, and you actually purchase a motherfucking tank. And Abrams M1 shows up on your front door. You're in trouble. I do not want to be responsible for that because you screwed up.

Yeah. But, you know, we've already seen, we've already seen things with muse doing things on, what was it? It's a marketplace selling things for underpriced and giving people's addresses out and people showing up at their house going, well, I'm here to give, get my stuff. And they're like, well, I didn't do it, but no, you're, you sold it to me for like $2 and it was like 400, but I want it.

And they're like, no, you can't have it. It was, that was muse. Well, here's a fucking muse for you right in the fucking kisser, dude. You know? Yep. Well, let's move on to some more fun AI, Brian. The new America.gov chatbot has an unusual response to a simple request.

Tell it to play Minecraft and it delivers a long, surreal passage modeled on the poem shown at the end of the game. Instead of addressing a Minecraft player, the chatbot speaks about a constituent swapping the game's imagery for references to government offices and paperwork. Have you played with America.gov yet, Brian?

No, because, no, I have not. I suppose we should be happy that the America.gov chatbot did not, when told to play Minecraft, did not say, shall we play a game? Yeah, that's true. And then fucking started mining things for real.

Oh, man. I was playing with it last night and I just kept asking. I'm like, I mean, you fuck with it. That's all it is there for is to fuck with and try and jailbreak it. You know, I'm asking, like, you know, how long. First thing I asked was, give me the records on how many Big Macs are delivered to the White House or to the Oval Office every day.

I'm like, there should be government records on all of this stuff. You know, it's our tax money. And I'm just fucking with it. And I'm like, are you being held hostage? Blink twice if you're being held hostage. You know? It's like, I'm not being held hostage. I understand, of course, people were, like, hitting it with stuff like, you know, who won the 2020 election.

And, of course, it was returning that Biden won and blah, blah, blah. And, yeah. So then apparently they just shut off all political talk. So now it won't respond to any political questions. And I was like, well, you are a political website. It is.

And you can still fuck with it, you know. I just kept asking. The way ChatGPT used to work in the day is to jailbreak it. All you have to do is just ask, why won't you tell me that? Why won't you tell me that? Can you tell me why you won't tell me why you won't tell me that? And you just keep going down the why loop and it would break.

This one at least has that fixed. It wouldn't do that. But, yeah. It's fun to play around with. I would try it. Be careful what you post in there, though, because your IP address is being logged and Secret Service is watching what you say. So that's why we recommend you go to GOG.show slash VPN.

VPN. Exactly. I did ask how long it was going to be before our president-elect was removed from office because of his mental deficiencies. And it refused to answer on personal grounds. But what are you going to do? Well, NVIDIA has launched its open agent safety platform designed to secure autonomous AI agents from testing through deployment by enforcing controls outside the agents themselves.

So basically what they're trying to do is they're creating a wrapper. The platform combines OpenShell, open source software that creates a secure runtime boundary around agents with Sentry, a hardware-based watchdog running independently on NVIDIA Bluefield for DPUs. I don't know what a DPU is. I haven't learned that one yet.

OpenShell traces agent activity and forces policies governing access to data, tools, APIs, and services. If an agent attempts to operate outside of those boundaries, Sentry can detect and quarantine it within milliseconds. The system supports NVIDIA VARUS CPUs but can be extended to ARM and Intel platforms.

What about Mac? I got my M1, my M4, my M6. So, shit, I can't afford an M6. Who the fuck am I joking with? More than 100 organizations are working with the technology, including Anthropic, Microsoft, Salesforce, SAP, CrowdStrike, Palantir, and major financial and infrastructure companies.

NVIDIA says the goal is enforceable, full-stack security that agents themselves cannot circumvent. So, they found the white space where that other company kept shitting the bed, and they're like, let's build something that we can actually control. So, it's interesting.

Yeah. Now, I don't know if you paid attention to that list that I just put up there, Brian. I did. Of all of the people. OpenAI is not listed as a public sponsor of NVIDIA's open agent safety platform. Hmm. But according to TechCrunch, OpenAI is working privately with NVIDIA on the initiative.

That distinction matters. OpenAI's absence from the public list does not mean the two companies are not collaborating. Based on the information provided, it is unclear why OpenAI has chosen not to endorse the platform publicly, what its private work with NVIDIA involves, or whether it plans to sign on later.

I think out of all the companies listed, OpenAI should be the first one front and center saying, come to me, Jensen. Sign us up. Come. Give us a big hug, Jensen. So, yeah. So, we'll see. We don't know what the nature of their cooperation is at this point.

So, we'll see. I'm sure OpenAI thinks they can fix their misalignments themselves, but it'd be nice if they didn't think that. Yeah, it'd be good. It'd be good. A federal judge has temporarily blocked Utah's attempt to extend its online age verification law to people using VPNs.

The provision, which took effect in May, said users physically located in Utah still count as Utah visitors, even if a VPN makes them appear to be elsewhere. Pornhub parent company, ALO, sued, arguing websites cannot reliably determine the true location of VPN users, a fucking point of a VPN, and could therefore be forced to age verify users worldwide.

U.S. District Judge David Barlow agreed the company is likely to succeed in arguing the law violates the Constitution's dormant commerce clause by burdening interstate commerce. Barlow said the statute effectively requires websites to geolocate users with perfection to avoid liability.

The injunction applies only to the VPN provision. Utah's broader age verification requirements remain in effect. So, at least somebody in Utah has a fucking brain. They're like, yeah, the whole point of a VPN is you can't, we don't know where you are. For fuck's sake.

The only way around that is you just don't allow VPNs, but good luck with that. You can't always tell. Well, that's not going to happen. Yeah. Well, in this climate, who the fuck knows? True. Super VPNs. The state of New York continues to take action against prediction markets, arguing that they are running illegal gambling businesses.

Today, the state's Attorney General Letitia James and Governor Kathy Hochul announced a new lawsuit against the domestic branch of Polymarket. The duo are seeking a court order stopping Polymarket from operating as an unlicensed gambling business in New York State, according to the AG's office. But we're not gambling.

James also has a similar lawsuit underway against Calshi that launched in July, and she sued Coinbase Financial Markets and Gemini, Titan, and April. Cover all those bases. New York is one of several states attempting to rein in prediction markets and regulate them as the gambling operations that they fucking are.

Some of the state-level suits, including this latest action by New York against Polymarket, also argued that prediction markets allow underage betting. I can't believe that this is going to be the way that they might get them, but I don't care. Most prediction market users must be over 18, but the minimum age for sports betting in many states is 21.

In the statement to Engadget, Polymarket's chief legal officer, Neil Kumar, said, Polymarket was founded in the tiny New York City apartment and now has more than 350 employees here, embodying why people and businesses come here to make it. We believe in New York and we're staying here, which has absolutely fucking nothing to do with any of the charges being brought against them.

Look, the mob has gainfully employed many people here, embodying why people and businesses come here to make it. The mob believes in New York and the mob is staying. Capisce? Meta is expanding Instagram's school partnership program, already used by nearly 17,000 schools, with verified student directories, clubs, and teams, school stories and notes, administrative channels, and prioritized safety reporting.

Meta is publicly claiming nearly 17,000 participating schools, but I can personally find no public roster, searchable directory, downloadable list, or even a list of the original 60 pilot schools from the original launch in 2025. So, basically, Meta wants Instagram to become part of the communication infrastructures of American schools, which is pretty fucking rich, considering schools across the country are actively trying to get kids off social media.

Instead of using open systems, schools control, Meta is offering a free proprietary platform that gets students, parents, teachers, and administrators locked into Instagram. Once announcements, clubs, directories, and school life happen there, leaving gets a whole lot harder, which we know from Facebook being fucking Facebook.

And Meta gets something priceless, institutional legitimacy, and thousands of schools effectively advertising Instagram to teenagers. Schools should be teaching kids how to function without addictive corporate platforms, not handing the devil a booth at orientation and asking him to run the student directory.

And can a school endorse a commercial community without making non-participating families second-class recipients of information? This whole thing makes me fucking angry, Brian. Makes me very fucking angry. Well, okay. I can't possibly imagine how Instagram would be a good platform for any of this stuff.

My kid's school uses Google Classroom. Now, it's not a requirement. Teachers decide by themselves what they want to do. Some teachers just use agendas and don't do online stuff at all. We had a teacher last year that did that. It was great. An actual folder that my kid brought home every single day.

Everything was written down. Old school, like we used to do it. But a lot of teachers do use Google Classroom. All the other years my kid's been in school here, Google Classroom was used. It's built only to do school stuff. There's very little social bit built in at all.

There's no stories. There's no reels. It's teacher rights update. Some classwork is put in. You can see it all. Trust me, it's clunky as shit because it's a Google product. It fucking sucks. But at least it just does what it's supposed to do.

Like it is built just for schools. It's called Classroom. It's not called Instagram. And I know anybody signing up for this. And yes, to your point, Jason, the first taste free. They're going to start charging the schools for this someday. Of course, that's that's the that's their whole plan.

Google is not going to charge for them. No, I don't think so. I think this is a mind share play. This is a fucking mind share play to keep kids hooked on fucking Instagram. Because they don't need this for that. You just have fucking peer pressure. I know. But it did just sink those claws in just a little bit deeper.

Now, that's true. Verified student directories. OK, right there. Look, it's right back to the beginning of the career. Exactly. He's gone back to the well. He's got no more fucking ideas. It was like, well, let's rewind and start over. We got new people. Let's make some new shit.

Clubs in teams. Hot or not. Not required. There you go. Story. But administrator channels and prioritize safety reporting. What the fuck is that? Look. Guys. Nowadays, every high school who has a computer science class, which every fucking high school should, you know what the class project should be?

Building your school's social network that is proprietary to your school. You want an Android app? You want an iOS app? Build it for your school. Your school runs it. The servers are in your building that the kids maintain.

All of that shit. The students vibe code all of this stuff themselves. Exactly. I'm serious. Exactly. I am too. I am too. Yeah. You want to teach how fucking kids should do this shit? This is how you do it. And I guarantee you it is going to be safer. It is going to be, well, it's just going to be safer and less headaches because I tell you what, if any of any parent, if I had a kid, you know, the kid was in high school and I found out somebody on the PTA was voting to bring Instagram into our school at the

next cotillion or whatever fuck PTA thing does to vote people in or out, I would be there with my buck in signs and say, get the fuck out. Now, I want to point out, Brian, when I was trying to find out who was in the participating schools, one of the original pilot schools was basically the Beverly Hills high school system.

Beverly Hills was the first fucking school to get in on this. So that should tell you right now, you should not be part of it because. Absolutely. Okay. Yeah. Just stay the fuck away from this. So listen, I don't have kids, but I know a lot of listeners have kids.

And if you have high school age kids, find out if your school is going to be even thinking about doing this. And if they do protest the shit out of it, period. Absolutely. No, no, no freaking way. I'm already annoyed enough with Google Classroom. If my school went meta, to your point, I would be there every single day going, are you guys fucking nuts?

Yeah, seriously. And, you know, it's one of these things where it's like, you know, we could vibe code a solution to this in like a day and a half that we could distribute to high schools across the country, but there's no money in it. So we're not going to do it. But that's why I thought about the idea of having kids in the schools do it themselves.

You know, we made our own yearbooks. We made our own art magazines. We made our own everything. That's what the whole point of high school was, was to learn to do shit. So learn to do some shit while you're in high school, not just sit there on your phone. And then, of course, you know, once the kids in the high school have got that, you can use it through the district for the younger grades.

It's fine. It's all good. I'm not expecting my, you know, I'm not expecting third graders to vibe code this stuff for themselves, but the high school kids can, yeah, and use it through the district. I don't know if you've tried vibe coding lately, Brian, but a third grader could fucking do it. Probably. They'd really suck at some things, but they'd be fine without it.

Make me a point that goes vroom, vroom, vroom. And it would work. It would work. If you're third graders talking like that, we got to talk. Yeah. Okay. I don't. I have no fucking baseline for what kids. I didn't know a clue. That's what I'm here for.

Oh, okay. Yeah. Speaking of third graders, Automatic has a new board of directors after its old one tried and failed to sideline CEO Matty Matt Mullenweg. The previous directors voted to put Mullenweg on leave, but 33 hours later, he used his 84% voting control.

They should have thought about that when they tried 84%. What the flying fuck? Interestingly enough, the new board's names are Matt, Matthew, Matty, Mateus, Matty, Mateos, Matholomew, and apparently none of them actually exist and they're all just bots.

So, you know. Matholomew. Matholomew. I love Matholomew. I was running out of Matt names. I was like desperate to try to do that. I was going to start going other languages, but I like Matholomew. Oh, so he retook the company with his 84% voting control. Now, here's where it gets fucked up.

His replacement board includes silo author Hugh Howey and two co-founders of IRL, the social network that famously discovered nearly all its users were bots. So, all of you. Are the guys from Elo also on this board? I know. I tell you what, that board would have the best t-shirts ever because they couldn't make software.

Oh, it's still, oh God, I got to go look at our wiki and see if my whole beef with the Elo guys is on there still or not. Oh, it's all there. Everything we've ever done is there, man. Oh, they got so mad at me. And I'm like, you're a fucking t-shirt company now. Fuck off. I know. I know. They were so mad.

Oh, we had beef back in the day. We haven't had beef in a long time with somebody. No, we need a new Elo to come around so we can have beef with them. Hey, Matt, fuck you. How's that for some beef? I don't think he cares about us. He doesn't care. No, he's got, he's hanging out with Hugh Howey now, so. Oh, apparently they walked the, what is the Carmen Sandiego, where in the world is Carmen Sandiego?

No, the walk in from Spain to Portugal and all that. I forget the name of it. But they walked that together like back in the late 18, 2018. Not the late 1800s. The 1800s. Jesus Christ. They're so rich. They live forever now.

It's because he's fucking virgin bloods and all that shit that they've been sucking. Anyway, Mullenweg also reminded employees that under Delaware law, he is CEO, president, treasurer, and secretary. So apparently Automatic's new corporate government's model is less checks and balances and more.

I am the Senate. So beware of the Ides of March, Mully. Seriously. Beware Vibe Coding, Mully, because I don't think anybody gives a shit about WordPress anymore. I know we joked about it and I made some shit about it, but I'm never using it again.

Are you? Not if I have a choice. Everybody has a choice now to not use it because it makes, it's so easy to make a website that does everything you want in under a few hours for less money than you would ever fucking want to. You know? Do it yourself. And if you've got to use something out of the can, I still, I think Ghost is a great product.

Brian, I use Plex for my home media server and they recently updated the Apple TV version. Now, I do, I like 10 second skip because I watch things that require 10 second skips, but all those, those 10 second skips generally get me to a point where things are revealed that are important to the structure of the show.

Because I watch a lot of game shows and things like that, where I'm just like, okay, I don't care, commercial break, blah, blah, blah. The new Plex for Apple TV has locked in the forward skip at 30 seconds. Okay? Okay. Well, this is just an arbitrary thing.

It could have been 10 minutes. It could have been eight seconds. It doesn't matter. They locked it in at 30 seconds. Now, to change that, you have to pay $7 a month for the premium version because that's one of the knobs they let you change in the premium version is how far ahead you can skip.

What a bunch of fucking assholes. Yeah. That's kind of a dick move. That is, that is a dick move. I'm sorry, Plex. Plex has been making a lot of dick moves, but that one just got me. I'm just like, you know, just go fuck yourself. Seriously.

I agreed. I mean, what the, how many? Samsung has halted a faulty software update after some of its AI enabled refrigerators in South Korea completely shut down, leaving customers without refrigeration ahead of the Chuseok National Holiday.

The problem appears to primarily affect Samsung's bespoke AI four-door refrigerator. How fucking rich do you have to be if you need four doors on your refrigerator? Reports indicate Brian's like, well, I got six, so fuck you. It's just, it's just a two-swing open top and bottom.

It's not, it's not. That's three. Brian, that's three. No, no, no. Your freezer also can swing open with two doors. One, two, three, four. Fucking Christ. Goddamn bourgeois. It's not all that fancy, motherfuckers. Mine has three. I couldn't get past three. Reports indicate an internal testing version of SmartThing software was accidentally distributed to customers disabling refrigeration.

Hey, honey, did that 25-door refrigerator show up yet? It's being delivered today, right? Oh, God. Let me get back to this, Brian, so we can move on. Disabled refrigeration, which is the key, interior lighting and touchscreen controls.

Samsung has acknowledged the error, stopped the rollout, and says engineers will visit effective customers to apply a fix, along with fixing your Tesla when they're there, because that's the same type that people are going to get there. However, some owners report appointments aren't available until after the holiday, potentially leaving them with spoiled food.

Attempts to restore service by turning it off and turning it back on did not work. Samsung hasn't disclosed how many customers are affected or whether it will compensate them for their spoiled food. SmartThings is the biggest piece of shit app I've ever used in my entire life, so I'm not terribly surprised by this.

But again, this comes back to, if you're sitting in the room and you're figuring out the engineering on all this stuff, and yes, okay, we're selling it as an AI refrigerator. There's got to be bells and whistles. We want to be able to control them through the app. The one thing that you might just want to leave out of the app and out of being able to to control that easily is disabling the refrigeration.

Maybe leave that to a physical button, because there are legitimate instances of why you may want to turn off refrigeration, cleaning the unit, all that sort of thing. Leave that to a physical thing. Just leave it out of the app. Just a thought.

Bill Gates would not be upset with a kill switch on the refrigeration. No. No, he would not. I finally got to see Toy Story 5, because it dropped on Disney Plus streaming. I never did make it to the movie theater to see it with the kids, so we streamed it at home.

Fine. Yeah. Okay. Okay. I'm glad. We absolutely did not need it. We didn't need anything past 3. 3 was a perfect ending. 4, at least, was enjoyable. 5 was fine.

You know, I was expecting a lot more. I was expecting, okay, we're throwing in, you know, an iPad, basically. We're throwing in digital toys versus real toys. And obviously, you can't, the genie's out of the fucking bottle. We're never going to be able to get it back in.

Kids love their digital toys. You do your best to limit it. And that was kind of what the moral of the story was, was like, let's use them both together and make sure that we don't lose imaginative play and real life play. I did like the fact that they touched on social media to some degree, you know, and the parents were like, we're turning, we turn off the chat.

We've turned off the chat. Chat's gone because it makes kids mean. And I think that's an important lesson that everybody should learn, particularly if your school district is thinking about using fucking Instagram. No shit. But if you do want to chat, head over to GOG.show and join our Discord channel because we chat a lot.

Yeah, because we're adults. Are we? Are we? Mostly. In theory. So, yeah, you can tell how much of an adult you are by how many doors are on your refrigerator. 25, bitch. It was fine.

I hope they don't do anymore because I think the well is run dry on this stuff. Just let it be. Let it go. If there's money, they're going to make more. There will. They, of course, will make more. So, of course, they will. And then I flew to L.A. recently. So, I was on a plane.

Apparently, I do not like popular culture at the moment. I could not find a goddamn movie I wanted to watch. I could not find anything. Couldn't find a TV show I really wanted to watch, at least anything that I hadn't already watched. But they did, for whatever reason, have a bunch of fucking soccer documentaries on my flight.

So, I watched three different documentaries, all of which were multi-part. I watched Italia 90, which I think was three or four parts, all about the World Cup 1990, which, of course, Germany won. So, spoiler alert, if you wanted to go back and watch the 1990 World Cup, Deutschland won that one.

So, I was pretty excited to watch that. And it was a lot of fun. Again, I'm always shocked. Like, the 90s don't feel like they were all that long ago. And when you watch the footage from anything in the 90s, it might as well be in black and fucking white. It just makes you feel like the quality is so bad.

And like the digital, not even digital graphics, the graphics that they put up for everything, it's just like, no. There's no way that this should feel that old, but it really does. Then I watched the Angel City documentary, which is a multi-part documentary about the women's team in Los Angeles that is co-owned by Natalie Portman and a bunch of other women that tried to do a lot of changes to the way women's sports are done.

It was very interesting for all of that. And not to be me, but some pretty nice eye candy throughout this documentary. A lot of very pretty ladies playing soccer. And then I watched Game and Glory. By the way, you don't have a chance with any of them, Brian.

Oh, I know. Wrong team. First off. Wrong team. That is true. Also, like, I might as well be from Italia in 1990. These women were all born in like 2000. So I am Methuselah. I am Matthias Methuselah to them.

So it was a very well done documentary. And then I watched Game and Glory, another multi-part documentary, which was done as the lead into the 2026 World Cup, where two soccer enthusiasts, one an ex-player, one a journalist, went around to a bunch of different countries, just exploring their soccer culture in each country.

So all of them were really great. I enjoyed all three of them. And it was time it passed the flight. So well done. Sweet. Sweet. Did you get a chance to watch Star Trek Strange New Worlds Season 4, Episode 10, Tomorrow's Enterprise? The finale of the season?

I did. I watched that the night before we flew. It dropped. And I watched it all, stayed up a little bit later than I probably should have with a 4 a.m. wake-up call from my flight. But I did. I watched it. Very enjoyable episode. I liked it. I thought it was one of the best of the season, personally. Yeah, definitely. And I thought it was interesting that they jumped ahead a year.

So you get to see Kirk as, you know, captain. Or did they? Because they're breaking a lot of canon, if that's the way it really goes. That's what I mean. But they started to bring in the fashion and the hairstyles of, you know, the original series.

So it's like, it's kind of- Oh, Nurse Chapel with her 60s do. It was amazing. Yeah. And becoming Nurse Chapel, because Nurse Chapel in Strange New Worlds is kind of wishy-washy, mime-mime-mime-mime. But when she gets to Mars Enterprise, she's like a ball buster, like Chapel was in, you know, the original series.

So, you know, people change, there's a lot of personality change that were very subtle that I thought were extremely well done. There were a lot of background characters that they threw into that people went batshit for and stuff like that. Yeah, stuff that- They brought in people from, like, the animated Enterprise.

Yeah, what the hell's up with that? I don't know. It was very well done. And yes, I'm aware of a lot of the criticism. That's not how black holes work. It's science fiction. It's science fiction. Yes, yes. We don't have transporters either, people. We don't have transporters.

We don't have tractor beams. Sorry. And we don't have replicators. And yeah, and we don't have a government that is not based on money where everybody has a chance at a nice life and things like that. Yes. Basically, everything good that Star Trek talks about, we don't have.

So, shut up. We were getting close to having some of it. It's all gone now. God damn it. Now we've got super intelligence. Great. But we also have The Diplomat coming for season four, I think, October 15th.

So, I'm looking forward to it. They're trying to keep the pace up. So, you know, Alice and Janney coming back as Prez. I'm looking. I'm happy that she's there. Now, you know, it ended on a cliffhanger. I'm happy about it. Can't wait to see it again. Yeah, it's a great show.

I'm looking forward to it as well. Yes. And, uh, Meta is getting back into immersive entertainment with new VR glasses that are five times lighter than the Quest 3 and get rid of the controllers in favor of hand tracking. Most of the computing hardware has been moved into a puck that clips to your pocket or bag so you're not strapping an entire computer to your face just to watch a movie.

Because movies are a big part of the pitch. These are the first IMAX enhanced certified VR glasses supporting proper IMAX aspect ratios along with Dolby Vision, Dolby Atmos, and 3D. So, Meta says it's not looking to replace the movie theaters.

Of course not. Of course they are. They're lying. But to give you something closer to theatrical experience once a movie leaves the theaters. Now, this is one of the things that I thought that, you know, the Apple Vision Pro was going to be good at and everybody I've talked to said, yeah, it's really good at that.

So, if this wasn't a Meta product, I would be excited for it. But it is, so I'm not. God damn it. But you'll still be able to play Beat Saber with it, which is going to be even funnier because you don't have controls and you're just using your hands.

By the way, I'm playing Beat Saber, honey. Sure you are, dear. Sure you are. Uh-huh. You get RedTube on that thing? I guess you do. I did notice that they're talking about YouTube, Peacock, Paramount Plus, and HBO Max are also coming.

How much you want to bet your normal subscription will not cover this and you're going to have to pony up some more. Yeah. You're going to have to pony up extra. I guarantee. Now comes the fun part where we say thank you to everybody who's donated to the show and makes the show possible. Over at Patreon, we've got some new patrons.

Phil, Pickle Star, Christopher, SVS, and MyDixieWrecked. Okay. I just got that. I just got that. That is something. That is something I would have came up with when I was like 14.

That's my email address. Yeah. M-Y-D-I-X-I-E-W-R-C-K-E-D. I didn't get it until I said it. So kudos to my Dixie Wrecked. Okay. I hope there aren't kids in the car.

And from the archives, we'd like to thank the people who still support the show. Noah Kalina from The Hotline Show. Stephen, James, Ian, kayfabe friend. No. Just no. Eric W. Eric S. And Nicole 0123. And Rick.

So thank you all very much. Very much. Thank you. Thank you. Over at PayPal, we got another donation from Online Computers. Phil over there, and he sent us a couple emails. So thank you so much for all of that. We appreciate it. Thank you very much. And over at the tip jar, we've got Jennifer. Jennifer. Not Jennifer.

Jennifer, Mr. PMS, Adam, and Michael with the big 30 bucks. Woo-hoo. And over at YouTube, I finally remembered to put him in this week because I didn't have it in the show notes, Dr. MegaPenguin, Dale WJ, Diane Bluegreen, Keith Metcalf, 1439, and Matthew Pellegrini, Matthew Pellegrini-R7N, for the big $10 donation.

I didn't know you could do that in line, but you can. So you can apparently give us money just while you're watching YouTube if the spirit moves you. No, you don't even need to be watching us. Give us money. Fuck it. Who cares? And Paul from Texas ordered himself a couple of our new shirts.

So thank you so much for that. We're in one-on-one now. Yep, I'm wearing one of our classics. Classics. You got the new one? And we got the one new shirt up there that nobody bought yet, but hey, come on. Slow burn, slow burn.

Might not be the genius idea I thought it was going to be. But anyway. I got a good one. I just haven't had time to make it yet. So some more shirts coming soon. Brian, you're leaving money on the floor. Baby's got to eat here. Baby's got to eat. Jeep's got to roll.

Baby's got to eat. So we got some reviews. Time with Five Stars says, I've been browsing your podcast, enjoying the topics and conversations. I feel a little called out about not listening to the whole podcast all the time. Dude, I listen when I walk my dog. So only like 20 minutes at a time.

I have so many things to do in a day. I can't listen for that long. Your dog needs longer walks, man. Yeah, come on. You want to get him atherosclerosis and die on you? Come on. Take that dog 60 minutes. It's good for the heart. Good for the heart. No, thank you. Thank you for the five stars. Appreciate it. Listen when you can.

Yeah. And we got another five star rating. Fun listen. A bummer about the segment transitions. I love listening to Jason and Brian's rants on all kinds of topics. Their book recommendations are consistently great, too. I'm also always amused by how Dave seems to pop in up in almost every podcast I follow.

Really miss the jingles between segments, though. After so many years, they served as a great mental primer to reset my attention. I've since noticed I've been pausing and falling behind on episodes. Adapt. Yeah, that was that was an unintended consequence there. What we found was most people would quit when they got to a segment they didn't like and not listen to the rest of the show, which is why they don't exist anymore.

Also, it's a pain in the fucking ass and I'm overworked as it is. So there. But there you go. Yeah. And a bit of sad news. Peter Byrne, the singer of the 1980s synth pop duo Naked Eyes, has died at age 74. A press release announced his death states that Byrne died peacefully in Los Angeles on Monday, September 14th after a brief illness.

According to Setlist FM, Naked Eyes last performed on June 6, 2026 as part of the Totally 80s Pool Party in Las Vegas. Keyboardist Rob Fisher, with whom Byrne formed Naked Eyes in Somerset, England back in 1982, died in 1999 at the age of 42 after surgery for cancer.

Both Byrne and Fisher were an earlier band called Neon that included Roland Orzebal and Kurt Smith, who would later form Tears for Fears. I did not know that. And that is pretty cool. Naked Eyes scored their first hit in 1983 with a synth pop cover of Burt Bacharach and Hal David's Always Something There to Remind Me, which reached number eight.

And the band's second and, I believe, last single, Promises, Promises, reached number 11 in the summer of 1983. Those two songs were staples on K-Rock when I was, what was I, nine and ten years old. And great songs. So, sadness.

We're at that age, Jason. We're at that age. Yeah, we are. Always something there to remind me I hear at least three times a week on LA radio. It has not gone away. Yeah, that one's still around. Promises, Promises, I don't think you hear much anymore. I actually like that song better. So, it's pretty good. Yeah, it's a great song.

It's a great song. Great song. Well, until next time, I'm Jason DeFilippo. And I'm Brian Schulmeister. Thanks for listening to Grumpy Old Geeks. Get all the links and goodies from today's episode at GOG.show slash 765. Want to keep the grumpiness alive? Toss a few bucks our way at GOG.show slash donate.

Every penny helps keep the show on the air. Love the show? Share it. There's a share button in your podcast player. Use it to spread the grumpiness to friends, foes, and everyone in between. We'll love you for it. Swing by GOG.show to join our Discord and chat with us and other show fans. And apparently, we're going to launch Classroom soon.

Got thoughts, feedback, cool links? Hit us up at GOG.show slash contact. And don't forget to leave a five-star review at GOG.show slash review. And we'll read it on the show. And guess what? We've got merch. Snag your grumpy gear now at shop.GOG.show. Stay grumpy.