Giving AI Agents Creative Free Time

It's time for twit this week at tech. Kathy Kellis is here, our uh favorite attorney. She's gonna talk about the big meta social media trial. Uh Brian McCullough from the tech brew Ride Home will talk about the latest tech news. Harper Reed is our AI guru. He'll talk about why he thinks AI agents should have free time and a big decision in the Supreme Court. ISP versus record company. Who do you think won that one? Coming up next on tw it podcasts you love from people you trust this is tw it This is Twit, This Week in Tech, Episode 1077, recorded Sunday, March 29th, 2026. I would download a car It's time for Twit this week in Tech, the show where we cover the week's tech news. It has been a great week for tech news, and that's why Kathy Gellis is here. It's all court decisions all the way down, Kathy is a contributor detector. She is an attorney at law. In fact, admitted before the U.S. Supreme Court. So she might have some opinions about the most recent Supreme Court decision. Hello, Kathy.. Hello Thanks for having me. See you. Yep. Same here. Yep. Yep. Yep. Yep. Yep. Yep. Yep. Okay, we're going to get down to business, obviously. Brian McCullough is also here. Uh host of the morning brew ride home the brew tech brew ride home. Tech brew. We gotta fix that lower third. Tech brew ride home. Part of the thing. Move it over to the morning brew. Morning brew family, but it is the tech brew brand. They're tech brew. Yay. Great to see you. And I see you have a picture of me behind you. Yes. I'll be changing that out. Oh you can change it. Oh Dredd, I thought it was permanent. I was hoping. Actually we knew. It's been it started Sam Altman and Kathy Gellis said, um, do you have a signed picture of Sam Altman behind you? And then it became Mark Zuckerberg, and now it's me. All your heroes. Good to see you, Brian. Harper Reed will be joining us in a moment to talk about AI news, and there's a lot of that as well. But let's start with the uh the court, the big story, and everybody's talking about it. We uh we knew the verdict would come down this week. Meta lost actually two cases, but the big one is the LA case where a uh a woman named uh Casey sued Snapch at, TikTok, YouTube, and Meta saying that her social media addic tion contributed to didn't cause but contributed to her depression and her rotten life. The jury agreed, but didn't give her a lot of money, which I thought was kind of interesting. Incidentally, TikTok and Snapchat made an agreement, a settlement before the trial Kathy, a lot of people that I mean the New York Times headline on this was is big tech facing a big tobacco moment? A lot of people think that this decision plus the one um the in uh was it New Mexico um were very prejudicial to the future of social in uh in court It's not a precedent. I'm I'm glad we have you on because we we're I was trying to figure this out. We talked about it on Wednesday on intelligent machines. It's not a precedent, but it is Paris was saying a a a uh a bellwether. A harbinger, basically. A bellwether, that's the word. Thank you, Benita. Yeah. So what is the what is the first of all, what do you think of the decision? Um I I I think I don't think highly of it. And I think it's going to be appealed, by the way. It's going to be appealed. Now, on the one hand, and this is a very small one hand, but let's acknowledge it. I think the jury did find, and I'm kind of combining some details of both states. I think some of the claims the juries didn't necessarily find on. Um, but I think the bigger problem is less what the jury found and more, how did the case get to this point? Because there's an underlying legal theory that they ended up presenting evidence on to reach the jury, but it's a legal theory that is not a good one. It's n uh I think for the point you were making of okay, this wasn't a lot of money, but this won't even scale well for Google and Meta, which you know they can afford a couple of these, but they it's a combined six million for Meta and YouTube and that's both uh punitive damages and statutory damage. Well, I don't know. seventy five million for meta . But I in both cases, these companies have plenty of money to pay the fine. The fine is not is a slap on the Well for one case, yes, maybe even a couple of cases, but there's no limiting principle to the amount of cases that could be thrown at them, plus also litigation costs, you have to add a bunch of millions onto that, including like from discovery, etc. So it's an expensive proposition that can't not cannot possibly scale to cover all the people who potentially have had a bad time on the internet, plus all the people who want to troll and just file lawsuits anyway to get money out of big companies that have cash. And fundamentally there is a problem here because not just the big companies are in trouble if this can happen, but a smaller company or an individual or a small company or a small platform or individual blog, the whole principle of could Meta and uh Google be liable for how they designed their platforms? The answer has to be no. It has to be no um on a four because of the first yes so that's what's interesting about this this wasn't a section two thirty case they're not blaming them for hosting bad content they're saying it's really a product liability case they're saying Google and Meta designed intentionally designed their products to be as addictive as possible. But this that is I've said and maybe this isn't uh comp comparable and I'd love to hear what you think too, Brian, but I've said that well but you c and in fact this was what the meta's defense attorney said. You know, you can blame Netflix for making bingeable content. Everybody I want to make Twit something you can't skip. Everybody wants to make their product as desirable as possible. You chose to have me on in order to get people to watch and listen and too. Oh my gosh, you're trying to induce attention into your own and and boy, you know, you know, you don't want that decision to be something that the law could come after. Brian, what do you think? Well, that's my question is it the the the reason people are making the analogy to big tobacco is that it is sort of like your product is defective. Right. They and they knew it. And there were smoking guns in the testimony and discovery that they did in fact know that they were making content that was addictive. And even in in Meta's case harming harmed young people. So right... end run around 230, again, Kathy being uh a lawyer would be able to speak to this much better than me, but is that a novel way to do this where yes, it's not about the bad content, it is about how the product is designed, so then it's a faulty product, so then that's where the liability can come in and then that can open the floodgates. So actually Kathy, that would be my question. Does this open the floodgates? Like can is there a bellwether? Exactly. Well um yeah, uh i yes if, this is allowed to stand, it opens the floodgates. But it's not correct for a couple of reasons. One, to say it's not a two thirty case is because somehow a bad argument was accepted that it's not a 230 case. 230 should have shut this down. It should have shut it down because essentially it's a moderation decision. You're building your platform around your moderating plan. And your moderating plan is I want the most content up that's gonna get people to stick around. And 230 protects moderation decisions. So it should have been a 230 case. This is essentially imposing liability on the on platforms which will affect content on the internet These are state laws imposing state-based liability and there is a preventiontion preemp provision in section 230 that says states butt out. And this is running at the heart of what section 230, one of the two things that section two thirty protects. And to say that 230 didn't apply here, it was a legal mistake, and that needs to be one issue on appeal. But the same thing. these b these b evildoers. And this is what worries me because it does undermine 230. It's hard for people to under stand why 230 should protect Meta. Mike Masnik on Tech Dirt says bad defendants make bad laws. Everybody hates Meta. So including apparently the jury. So it's understandable why the jury might be willing to accept legal theories against them that would be problematic if applied to anyone else. Could you apply this legal ruling to somebody else that's more sympathetic and make me feel sympathy for Meta in some way, Kathy? I mean the answer is yes, if this is the way it works, that liability could be found for endeavor com Meta took action in how it designed its platform. Everybody who designed to make it sticky. Everybody who tries to make their platform sticky is potentially running afoul of the law. Now, a jury may or may not think that they did so under all the things that the that um that the New Mexico or California law requires, but the ship has sailed at that point. You cannot possibly defend yourself. Uh again, a small the smaller the platform, the fewer litigation instances they can manage in the at all. Yeah, if somebody came after me saying your chat room is way addictive and I as a youth I spent hours on it and as a result I uh I'm depress ed. I couldn't defend that. Uh in a six million dollar job So that's who we should worry about is the little but but who are people gonna go after chat rooms and forums and places like that? You know who will? The ones who don't want all that speech to take place. Ah. Now um I mean So bad actors in the uh uh as plaintiffs are the thing to worry about at this point. Well you you open the door to like I said at the beginning, not everybody is having a good time on the internet. Some people are having terrible times on the internet, and that's a lot of people. I think it's a tiny fraction of the amount of people on the internet, but tiny fraction at the scale of the internet is an awful lot of people. So on the first part, you have a problem with all the if all of them brought their lawsuits, you would have a problem. But it's not just going to be them who will bring their lawsuits. It will be people who want to pick the pocket of a defendant with cas h from them. They'll be suing. And then there's also people who just want to impose friction on the speech that's going on online. They'll sue too in order to impose that friction so that the platform has to squash a lot of the speech that's facilitating, that's being facilitated. Now, a slight devil's advocate with something you said is you're using Discord as opposed to coding your own platform. So that might sort of a little bit put you outside of, you know, the immediate crosshairs of this, but I I don't think in a way that is that you should be sleeping easy as as a result of it. Well, but and you could see Discord being uh a plaintiff. And it's no good for you if Discord shuts down now because now how are you going to host the thing? There are thousands, by the way, of cases now sitting behind these two waiting to uh launch. So the fact that these you know the jury it was interesting, uh the New York Times quoted a couple of jurors saying we didn't want to really punish these companies so much as get them to fix their services. That's why the that's why the the Los Angeles uh uh fines were so relatively low. Aaron Powell But I think that so I said that two thirty should be blocking this, but also the first amendment, because this idea of what you should be doing better, the idea of content modern, like you and I and everybody who's listening may have some good ideas of things that Facebook should have done differently, that YouTube would have done differently, but reasonable minds can disagree and people have different opinions about how things should be built and what sort of speech should be facilitated and highlighted or degraded. And we're not going to agree. So this idea of we want legal pressure to cause the companies to fix things, we don't have a clear idea of what fixing is. And that's why we have a first amendment because we will not agree on what fixing actually looks like. And this is directly affecting expression. It's directly affecting expression in a whole bunch of ways. Wait a minute. Does is uh is meta protected by the First Amendment for its algorithms? I guess it is. That's a form of speech. But there is product liability. I mean, you can't defend uh everything as freeze like if I make it to you can a lot. This so getting back to your original question, why the New York Times is wrong, um, this is not big tobacco. There's not really an expressive function going on with smoking a cigarette that that I mean it's not free speech. It's I mean yes, somebody could throw some arguments, but they're very, very, very attenuated. Whereas these so called products are really speech facilitating services. And they're speech facilitating services. There's expression going on into how you how do you code them. There's expression both in terms of the language and also how you design and all the choices you make, like the choice to make to make it sticky, that you want to make it sticky, how you make it sticky. The fact that you could make other choices is is itself exhibit a for why it's expressive because those are choices you could make. And depending on what you choose, the expression you foster will change. They're upset that expressive choices were made in a certain way and they want the law to pressure the platforms to make them in a different way, but that runs straight in the heart of the first amendment. Making algorithms is not the same as making uh cancer causing s uh cigarette product. It's not the same thing. Correct. So now it's going to be appealed. Do you think a panel uh appeals court panel of judges will be more understanding of that? I mean obviously a jury can be swayed along these l ines. Do you think it'll be overturned on appeal? Oh well, I hope so. But uh one thing that's nervous is these are state court things, so they have the peculiarities of state court litigation. Um so they'll be harder. Uh but on the other hand, you know, Cal I don't know a lot about New Mexico, but California does have some 230 cases on the books and um Hassel v Yelp and things like that. So there's some California precedent to throw at it. And then there's been the interpretations, and you know, I'll I'll be an optimist, and especially in California, I'll be there to try to litigate it if I can get the briefs in. But I think sometimes for for arguing this, you really have to reframe things. This is not a think of the children. You really have to encourage a very sort of meta look. Well, not meta-meta, but if they hadn't used that word, but um y you know, big sky uh looking down and getting the big picture type, look at what's going on and what the reverberations will be. And a micus uh turnout is gonna be really important, especially from smaller players to sort of say that's very nice that you hacked at at Meta and Google, but you're going to kill us. And if you kill us, this is what what falls uh in consequence. And particularly for users too, because this is bad for users. If you can force the platforms to make certain decisions , particularly there's a lot of people who have actually been getting a tremendous amount of benef it in how it connects people. And if you're squeezing the platforms so they can't be available to connect people, you're actually going to isolate people. And that is a design defect. Like all the things that they're, you know, sad about that people had difficult lives and different difficult experiences being isolated on the internet. You'll just isolate people in real life when you take the internet away. Kaylee, uh who's I mean, uh uh we certainly know she says she has bodies she had bodies dysmorphia from the filters that Instagram uh made her available to her. She was using, by the way, should point out YouTube at the age of six and Instagram at the age of nine. So And according to at least um the the trial did not make use of YouTube kids, which I'm on the record as saying is BS anyway, but But uh there there are in theory tools and safeguards uh that these companies have put in place. Uh what what I'm curious about i um Kathy said, you know, think of the kids and this is, you know, exactly thirty years on from the Communication Decency Act and and Section two thirty. Uh when when when you're seeing all of these uh age verification things that are theoretically trying to um protect kids from social media, like you can't join social media to a certain age, all that stuff. Which we can get into maybe there's uh more nefarious uh motivations behind doing things like that. But um is this something that like the tide has turned and that if there are things like jury trials and and stuff like that, there was a time when you you couldn't get the car companies to lose a case, you couldn't get the tobacco companies to lose a case. Again, I'm not making the argument that this is the tobacco moment or whatever, but I'm curious if after thirty years should we expect more uh rulings like this because maybe the the tide of public opinion is more amenable to I think that that's what's the whole point is this is public opinion uh coming out. Uh uh in a court decision. Well it's there are gonna be a lot more cases. I just hope that they well I don't know, that's a mean thing to say. I hope they're all against Meta. Well I, mean, Meta has brought this on themselves in including by, you know, they're not a good actor and they don't act well in this space. And I don't think they even act consistent with their own self-interest. But here we are. But let me point out, think of the children as not new now. 230 exists because it came out of a think of the children moment. It was how we thought of the children, and and the rationale for it really hasn't changed. You will get rid of the most crap from the internet when there's not liability that can attach to the map platforms that try. And if they don't try, you don't have to use them. Um you won't get the good stuff either if the platforms can't exist because their choices are all of a sudden subject to be ing we did think of the children, let's still think of the children, but the idea that we need to do something different is a a false legal one and it's a false practical one. And it's also one that I think people don't realize the full extent of what would happen if we did it differently. And one other point was: I mean, this poor plaintiff, but you know, if you're using the internet at six and nine, and I think there's evidence in the record of this there were other things going wrong in this person's life. There's a lot of trouble. You know, I I don't want to you know minimize personal pain, but they're actually, getting back to an original question that I glossed over, the jury verdicts may be kind of garbage themselves because one of the other things you have to find is causation. And you know, I can't. Proof causation in this. You can prove causation in tobacco, yeah. But you can't prove you know causation in this. In fact, even the concept of internet addiction is not fully supported by science. Yeah. I mean and there's science going Going the other way in terms of overall benefit of the internet versus not like there the there's a public the tide has turned in that there's a public perception that the internet is bad and poisonous and destructive. But it's it's an aber it's not an accurate view of the true value or relative value of of the internet. And I agree. But I think maybe that doesn't much matter what you and I think because it's very clear public opinion is turned against uh the internet. And I think that may we may regret that. Kathy, I'm gonna we're gonna take a break. Harper Reed has arrived. We're gonna get uh to more stories. Hi Harper. Great to see you. Yay. Pea ce. So there was some uh uh speculation that maybe you were at the big comic uh comic con going on right now in Chicago. Well I I had to I had to take off my my uh my costume and and rush through the door. So, you know, you caught me in a really rough spot. I was wearing full chain mail, as you can imagine, and it was rough. Okay. I actually watched K-pop demon hunters last night. Now I'm thinking I really want to get that the black top hat uh outfit. I think that's pretty cool. I want to be a dem on next time I go to a Comic Con. We're gonna take a break. We'll come back with more. Harper Reed is here. Uh Kathy Gellis, attorney at law, Brian McCullough from the uh Tech Brew Ride Home. There you go. He's zooming out. Zooming out. I like that. You got the technology. This week in Tech brought to you this week by Doppel, brand new sponsor, we welcome to uh Tw it. Maybe that, you know, that text you just got, maybe that's an urgent message from your CEO . Or maybe it's a deep fake trying to target your business. Nowadays, AI can impersonate trusted individuals, and Doppel's platform illustrates how frequently users fall for phishing attempts. In voice call simulation deployments, that's what Doppel does, training. Target users spent six minutes oh boy six minutes conversing with a deep fake. One hundred percent of them believed the AI was hum an. This is not good. Doppel is the AI native social engineering defense platform. Doppel strengthens human risk management by training employees to recognize deception while their digital risk protection detects and disrupts attacks across every channel. As attackers turn to AI to power increasingly sophisticated strikes, Doppel uses AI to fight back with automated takedowns, multi-channel coverage, and AI defenses that build intelligence with every fight. Doppel works relentlessly to protect people, brands, and trust. Doppel offers best in class integrations and partnerships to seamlessly integrated your to existing security tech stack. Doppel's industry awards and testimonials speak for themselves. Doppel is recognized as a winter twenty-six G2 leader of users most likely to recommend and momentum leader and best support. Join hundreds of companies already using Doppel to protect their brand and people from social engineering attacks. Doppel, outpacing what's next in social engineering, learn more at Doppel.com. That's do P-E-L dot com. Thank you, Doppel, for your su pport of this week in tech. Harper Reid, how are you, sir? I'm good. I was a little uh I was a little shocked. Um I messed up my calendar and I live by my calendar and so I was I was like, what do you mean it's today? And then I remembered that there's other places of record. So uh so I dashed to my office. I literally threw off my chain mail. Came here. Oh I'm so grateful. Well I always look forward to being here, you know, and it's one of those things that I love this conversation and I'm always uh excited to start and have that little pre-show thing, but I messed it up this time so I get the You get the post stick around. I will take a little bit of the blame here because I I sent him the stuff a little late, so I will take a little bit of the blame here. It's okay, but you know, we you I'll I'll take it all. I didn't know there was a no chain mail address code for Twilight. Oh no, no, no, no. I I uh I last time I did I did this. Last time you had chain mail on. I had chain mail on the case. Well it turns out you can buy chain mail on the TikTok shop. And it's very inexpensive. And this alone has A, messed up my algorithm, and B just opened the world to possibility. We have much to talk about, by the way, Harper, in AI, but But before we do that, I do want to wrap up all the other court cases because there was also a big Supreme Court decision. You know, when we get Kathy Gellis in here, we gotta we gotta cover all of the uh all of the stuff. Uh I was very pleased to see the Supreme Court unanimously decided in favor of Cox. They were being sued by uh a music lab el over pirated music. They said the music label said, well, the ISP is responsible . Supreme Court said, no. This seems like a good decision, yes, Kathy. I think it's a very good decision, but it's caused consternation. Um, because it changes a lot. Um, it chang es last 20 years of copyright uh precedent and like going back to Grockster, and arguably it may even go back further than that because it really changes secondary liability as it's applied to copyright. But um you're just a negative nelly today. I swear to God. I thought this was a good one. I thought uh not that I'm a fan of Cox by any means, which by the way is about to acquire a charter and become the largest internet service provider in the country surpassing Comcast. Um but I'm not a fan of Sony either, so it's kind of And I don't think the ISP should really be liable for its users. Now the Sony said the Cox had ignored bag actors. They helped 60,000 users distribute more than 10,000 copyrighted songs for free. And they did it because they wanted to keep their subscribers ' payments flow ing. In fact, a jury in 2019 found Cox liable for all ten thousand songs at issue and awarded Sony ten a billion dollars in damage. That's the case that was ultimately appealed. The Court of Appeals for the First Circuit uh Fourth Circuit upheld the jury's rul ing. Um, but they ordered a new trial on a separate issue and vacated the judgment. Cox, by the way . In a surprisingly good brief that also managed to catch a lot of the first amendment issues, although the first amendment issues didn't show up in the decision. Ye ah. Um, so why is this not a good thing? No, I think it's I think it's great. I'm happy with it. Oh I think the issue is you may not hear universal acclaim, and some of the people who will be not so happy with it are a lot of the copyright copyright holders who are now in a much weaker position than they were um even quite recently. It really dials back uh secondary liability for copyright tremendously. And the question is how tremendously, but significantly at least. It reframes Grokster and tends to limit it. And there's been arguments that even some like early 20th century secondary liability rules may be in question, it really narrowed when somebody can be secondarily liable for somebody else's infringement. And the implications of that are pretty significant, not just for internet use, but also I'd seen it pointed out AI use, particularly with respect to outputs. Will this change the liability picture for that? Uh but I'm really excited about it mostly because This is the second time. I think you should be careful here. Yeah, okay. Well he's the well the one thing he's done that I'm really ha okay. Uh giving sort of a sneak peek too. I haven't written my tech dart post yet, but it's it's coming. But one of the things that's interesting is a couple of years ago, and who knows in internet time, it feels like a decade, but it was maybe only like four or years ago or something. He wrote a dis uh descent in malware bytes where he was kind of talking about 230 needs to go. And he freaked out the industry. So everybody kind of got together, and then we were pursuing the Google versus uh Gonzalez case where we stood up for 230. But in that litigation, he ended up writing a decision for Twitter in the Twitter vTAMNA, which was essentially a secondary liability case for the internet, not having to do with copyright liability. It was with the anti-terrorism statutes and could you have liability for that? And he wrote this nine-zero, very academic and stalwart decision basically saying secondary liability, a common law does not work like it was being appl ied. Very, very protective decision. So then when Cox was litigating this, they basically were arguing almost full stop. Same rules for Twitter vTAMNA really should be applying here. And um and they the it didn't look like it was going to go well for moral argument, but the decision basically said, yeah, basically the same rules for Twitter vTAMA don't, you know, same rules apply for copyright. Secondary liability is very limited. Now, one of the issues is Thomas says it's limited to two contexts. Whasere um there's a this there's a dissent by Sotomayor. Actually, it's a concurrence because she concurred in the result that Cox wins, but she dissents from the reason why. So in a sense, the decision's only seven, two. And she says there's other ways it's in common law that there could be liability. And she pulls one up, which is aiding at abetting liability. And she runs through the analysis and says, well, it wouldn't apply to Cox here, but I don't like the idea that we slam the door to these other legal theories. Well, even Thomas's uh majority opinion said that, you know, you can't really hold them liable unless they actively encourage infringement. Yeah. And for hers with aiding and abetting, you needed intent. So basically, yeah, there needs to be a disagreement there. They're not really in disagreement, but structurally, she thought he was a little bit too universal in other paths of secondary liability that might apply. She disagrees with essentially his statements that there's only two ways it could apply. And it doesn't apply here. She says, no, there's a greater universe of secondary liability avenues. Here's a case we can all agree on. Elon Musk uh sued advertisers saying uh hey man, uh you can't not buy advertisements on uh whoops, I switched away from it. On uh Twitter just cause uh you don't like us or something. Um that the judge actually just threw it out. The judge said absolutely nobody has to buy ads anywhere there's no rule there's no la w and uh elon you got nothing here right am i am i i'm paraphrasing that but i i think that's rough ly uh yeah. Um but it took a while. Uh yeah, I know. This one this happened like more than a year ago. Yeah, I mean s I read that decision. It's interesting. Yes, you can quiz me on that decision. I did in fact read it, but like a a year ago in internet time. Um yeah, there were a whole bunch of plaintiffs. Some of them they had no personal jurisdiction over. Uh but of the ones that they did, the court was like, no, there's no there. Um the the thing that you're unhappy about is not a thing that you get to be unhappy about. Right. That's pretty funny. All right. That's the court. The court decisions. No, you're missing one, unless you want to bring it in for Harper. Um What else? Anthrop ic. Oh, yes. We could. This is uh this of course this is the ongoing battle between Anthropic and the Department of Defense. Uh Anthropic said we have uh hard lines against the the Department of Defense using RAI to spy on American you know for surveillance on Americans and for autonomous killing machines. It can't make kill decisions. Department of Defense said, I think not completely unreasonably, well, no private company should be able to sell its products to the government and then after the fact say, but you can't use it for this. It'd be like Boeing saying, Well, you can buy our planes, but you can't drop bombs on civilians. Um, that's up to us. We're uh elected, we're we're the government, and uh no private company should be able to tell us what to do. Anthropic immediately sued because there was a second side consequence of this. Department of Defense declared Anthropic uh to be a what what's the uh term supply chain risk. Supply chain risk, which normally is only applied to foreign companies. Out of the fear that they might be saboteurs. Right. Yeah. So uh w uh so uh the judge has made a preliminary uh injunction, this isn't the final ruling, saying it's first amendment retaliation . Yes. Um I mean one thing that's important is look if if Enthropic and the government and I do not like that Department of War was used as the the name of the department and all this and my tech drive post legal name is department of defense legal name is department of defense but the department of defense and anthropic if they cannot agree on terms for the use of the product then then the contract is going to fail they won't anthropic won't get to be a government contractor and or at least not for the Department of Defense. And fine. Oh, that's fine. Yeah. Like there's there's statutory authority that Heggsith can use to not do business with Anthropic, and they are perfectly unrestrained, even with this preliminary injunction, to go forth and use any legal means to not do business with anthropic. An open AI jumped in the breach and said, good, well you can use us and they did everything. They said anything you want, Mr. Mr. Hagseth, anything you want. Uh so uh although I think there is still quite a bit of anthropic being used in government still. Well well so um the contract was starting to fail, but anthropic is also used by other uh agencies within the government, which was one big deal. So then when Segta Hegseth decided to call them the give the supply risk designation, which arguably he has some statute authority to do, that created all sorts of extra problems for enthropic because it's a lot of Amazon, Microsoft, Palantir, and other defense contractors use Claude. And all the other agencies that they already had business with, then they couldn't do business with them. So they weren't just gonna lose one contract. They were gonna lose contracts with all the other agencies and anybody else they were doing business with because they would have to do that. It looked it looked punitive. It looked like they were trying to put anthropic out of business. And that's what the government that's why the uh injunction issued. Um it injunction it it issued for a couple of reasons. One, it looked like First Amendment retaliation and the decision goes through the analysis of all the steps and why the record show that it would be. It also, and then there's some due process problems of this was just sort of announced and not in a way that Anthropic could defend itself and it had some liberty interests in it. And also it didn't it, didn't comport with the statutory authority. Like there's an issue of if you're gonna use there's two statutes at play, and only one of them was at issue in this California one. There's gonna be a case at the DC circuit addressing the other one. But under the one that was on the table in California, you have to give Congress notice if you have an issue with the vendor. And there was no notice given to Congress. And at oral argument, uh had the government was saying, oh yes, that's just for the benefit of Congress. But the judge was finding that no, these prerequisites of notice are to make sure that this statutory authority is not being abused. You have them as prerequisites of things you need to do if you want to do the thing, the big thing down the road. And they didn't do them. So that was another reason why it was all I joined. Now it's just temporary, right? And the the there will be a case. Um it's it's a preliminary injunction, so there will be more proceedings here. There's going to be the proceedings in the DC circuit uh with respect to the other um statutory authority, although this is going to be persuasive to that. Uh we're not done with this because also the likelihood that the government, you know, accepts that they lost and goes away seems small. The judge Rita Lynn uh said publishing anthropic for bringing public scrutiny to the government's contract position is classic, illegal, first amendment amendment retaliation. Nothing in the governing statute. She says it's perfectly fine for the Department of Defense to say we're not going to use anthropic anymore. That's not the issue. Nothing in the governing statute supports the Orwellian notion that an American company may be branded as a potential adversary and saboteur of the US for expressing disagreement with the government. That seems fair. Yeah, I it seems extremely important. Yeah. Yeah. Uh all right. That's uh there you go. There you go. There you have some court cases. There we have some court cases. So there is another there is another story that's related uh in fact which I love this . I mean I love it. You mean love in an abstract sort of sort of way? I was RSE in a like a yeah. I was at RSAC on Tuesday and went over to Ubiquity because I use Ubiquity routers and they're made in China. It's an American company. Actually, I don't know of many routers that aren't made in China. And I said, Can I talk to you about this? And they said, No, we don't want to talk to anybody about this. So I apparently apparently Starlink routers are made in Texas. Yes. Interesting. Which is interesting. Interesting. Um but I I read this and then kind of laughed out loud and then I read it again. And then as you get further down in the article, which was very hard of just being like, what the um I you finally get to the point where they're like, oh, and you can apply for an exception through some process, which is I think obviously what they're gonna ask a lot of American companies that are doing manufacturing in in China. But the the the it is very difficult to manufacture things at the same cost as in China. In the United States. And are not only made in China, but it's a Chinese company. We expected that ban, right, Brian? That's that wasn't a surprise. Uh and in fact that's the only company that's been added to the list. So it's I'm kind of unclear w what the real legal power of of this ban is. It's uh I mean in theory, like a lot of other things recently, like it's I don't know what the legality is because like you can still uh like the existing routers still you you don't have to turn them in right um yeah those are all still in in play if i go this is the list this is the list the uh the the the the the covered equipment list. And it is actually very short. Huawei ZTE, Hytera, Hangzhou, Dahua, Kaspersky, because it's Russian, China Mobile, China Telecom, Pacific Networks, China Unicom. That's it. That's the whole list. Most of them come from 2021 . But then at the bottom it says routers produced in a foreign country, except routers which have been granted a conditional approval by the Department of War or DHS. Well that right, that's the other thing. So like okay, is Eero in there depending on where Eero is manufactured? It's made it's owned by Amazon, but it ain't made in the US. But then secondarily nothing in it says that the Department of Defense or the government has to stop using any of these routers either. And uh and if you if I uh drive up north and uh go to Canada and buy one and come back across the border, they're not gonna like it it's it it it was just like a a headline that actually again it didn't seem like there was anything like actually thought through for how this would work in any way. That is very typical of this administration. And if I go to Amazon, TP Link is available right now. It's part of their big spring sale. Oh, that's the Enterprise one. Oh I'm just kidding. It's not. But it's the deco. It's their it's their uh mesh router. I don't think that the FCC can necessarily ban sales. It doesn't have that authority, although as as as if it's statutory limits of its authority actually matter to it. But it needs to certify things because the if you look at your bottoms of your your routers and stuff, the SCC certifies to make sure and the theory behind it is we don't want all the interference, so you have to make sure it complies with the regulatory things that have to do with interference. And they're just saying that we're going to abstain from issuing that certification. I believe this is what they're saying that will abstain for from uh giving that certification if the company happens to be foreign, in which case, ha ha, good luck selling your stuff because we're never going to be able to so that's everything. I mean, i I don't know of any routers that aren't made in the China except for Starlink. I guess the question I have, sorry, sorry, Kathy, is the um is you know Unify, because I only think about my own system. That's what I have, Unify. Unify, which I think is a very rational one, is definitely not a consumer product. But the prosumer, I guess. It's prosumer, it's on that edge. But I think Brian's right. Like what about Euro? What about these big huge companies that are doing these things that are Chinese manufactured? And I could totally understand the idea of trying to block opportunities for you know electronic warfare, et cetera, and and you know, especially as as as the Department of War takes the war in its name very seriously. I think there's a lot of lot of reasoning for this, but it seems that there's a lot better ways to do this than this kind of misguided and poorly implemented. Trevor Burrus Well, also one more important thing about this is like they're mentioning things like salt typhoon and some of these other hacks and stuff like that. But like those weren't necessarily as far as we know, maybe the government knows more than we do, that wasn't caused by a backdoor, by a supply chain sort of like at the manufacturing level they installed backdoors that allowed these hacks to happen. It was the the telecom companies having up terrible security that allowed the things to happen. It wasn't your router, it was it was uh Verizon or whomever uh you know. But it it was it was at the level of the monopolies that control the telecom systems, not necessarily the the hard ware. Well, it's crazy. It's it is, it is, it is just very strange. I find this very a little stressful only because if you know if this works in some kind of way of working or or some some w what else is there that we will be you know excluding from import or or certification. Well, and they did it to DJI drones. Uh they did it actually to all foreign-made drones. And that was pretty clear because there was there's one American drone company that is partially uh owned by Donald Trump Jr. But again, remember, like go back to the Hegseth and the Anthropic case. Hegseth gets some statutory authority to do something. But it can oh, and actually one of the other bases that they lost was because it was arbitrary and capricious action as an agency. So FCC is helping itself to an authority it may not have. And even if it has it, it's arbitrary and capricious because there's no, you know, ooh, foreign bad, ooh, but that's a deliberation. There was no deliberation, yeah. It gets to exist as the FCC because we care about interference. So that's kind of you don't want to I don't want to diminish the uh impact of a supply chain attack because that is genuine. In fact, Harper, when we come back, there is a supply chain attack that happened uh to PyPyPi that is devastating. We'll talk about that. Yeah. Well, that may be bad, but the FCC only gets to play in its airspace. And if it's not an airspace related issue, I'm not sure it gets to play at all. Okay. We're going to move on. The uh we will talk about this uh big supply chain attack in just a minute. You're watching this week in tech. Brian McCullough is here from the tech brew right home. Every single day you do tech news. Every single day. That's cool. Yeah, but you're here on a Sunday now. Are you sick to death of it? Do you just want to talk about baseball or something? I mean Yes. Do you have a portrait of Sam Altman behind you? Yeah, I've got uh oh Sam's back I loved Leo up earlier. Yeah very sharp, you know, both Kathy and uh Harper picked up on that right away. I didn't notice it. Yeah, that's my I use that all the time for my uh TikTok videos and stuff like that. I call it uh in my file it's sam cringe dot jpeg because he's like a little cringy. He looks a little cringy. Great to have you, Brian. I'm sorry, this is uh what they call a busman's holiday. You're you're still covering tech news even on a Sunday. Kathy Gellis also here, attorney at law. You can read her stuff at Tech Dirt. In fact, an article forthcoming, I suspect, on the Cox case. Look forward to seeing that. And Harper Reed is here. His company, 2389.ai. I use some of your superpowers, your some of your extra special Claude skills. I love them. I did, you know, I did the world one. The the the um Yeah. That was fun. It got that's an intense one. It's very it was very intense. Yeah. I I find that one to be very challenging. And just to to let you know what it does, it's called our world world um worldview explorer. I don't remember what it's called, but it it basically tries to extract your worldview from yourself. I noticed that I didn't ha I didn't I couldn't tell you what I believed. I could tell you these little bits and pieces. Right. And so I wanted something to ask me all of these questions and try and help kind of facilitate the generation of a worldview. And then from there I've used this to do all sorts of weird and bizarre things, which has been quite fun. And yeah, we have quite a catalog over there. Look at all this stuff. These are all basically skills. Are they only for Claude or can you use them in the case? So these are these are a mix of skills. My favorite one right there is the binary RE. That's that's one of my that's right to your top right. Wow. Yeah. So if you have any binary, uh, you know, an APK, uh, you know, an IPA from iOS, uh, old, old, old binaries, um, you can just rip everything out of them. Um this has been very happy this has been very helpful for um old like if you get like some old piece of hardware and you're Skyra. Okay. Yeah. It's very, very effective. But one of the things we've we've found is like we will do these code loops or these these tech loops with these products and it's great and it works really well. And then if you do it a second time, our kind of internal rule is let's make that a skill. Just why don't we make it automated as much as we can? But um it's been very exciting. I've used your fresh eyes review. I love that. We have we have we have a new one. Two new ones. I just found Turtle. This is very cool. If you're learning the uh terminal, as many of you are now that this is become the way to code, way to uh uh vibe code, uh this is a teacher that teaches you commands in a little game. I would say it is. That is approximately working. Um, as is all things. Um, but it's fun. We we built that for some of our people internally who were trying to get in the command line and they would just be like, what? Because I anyone who's not familiar with that world, like I started using the command line you and I go DOS like forever ago. And like and it is not when someone says oh how do you do whatever it's just immediately I'm I'm saying it out loud. That's not a good way to teach is just telling someone how to do it and just practice. And I'm a huge fan of Duolingo and everything they've done. And so s trying to have some sort of you know until Duolingo adds Linux as one of the languages you can use. I I think that uh we're gonna have to do it ourselves. But yeah, this has been a really fun time. But I have a new skill for you that you'll like, two of them. Save it. We'll talk this'll keep people listening through the ad. Oh my gosh, you're trying to keep people around and make your show sticky. I'm in trouble. Oh no, don't show me Kaylee. I'm sorry. Please go touch grass. You don't have to listen to this show. It's not that interesting. Hon est. See, who's gonna say that? Right? Everybody wants their stuff to be a stick. Uh great to have all three of you. Our show today brought to you by Out Systems, the number one AI development platform, these guys have been doing it for a long time and they know what it takes to make robust enterprise software. Out systems helps businesses bridge the enterprise gap to their agentic future where the constraints of the past give way to unlimited capacity and scale out systems enables them to build agents, AI agents that can actually do work, take actions, make decisions, integrate with data rather than just answer questions. The only AI development platform is unified, agile, and enterprise proven. Unified because you build, run, and govern apps and agents in a single platform. It's agile because you can innovate at the speed of AI, but this is important without compromising quality or control. And it's enterprise proven because it's trusted, out systems is trusted by enterprises for mission critical AI applications and durable innovation. Out systems is the secret weapon behind the world's most successful companies. Out systems is not just for you know little one-off apps. They are for the massive, complex systems that run banks, insurance companies, government services. Out systems even helps companies with aging IT environments bridge the gap to the AI future without a rip and replace nightmare. Out systems provides the safest and fastest way for an enterprise to go from we need an AI strategy to we have a functioning AI application. Stop wondering how AI will change your business and start building the agents that will lead it. Visit outsystems.com/slash twit to see how the world's most innovative enterprises use outsystems to build, deploy, and manage AI apps and agents quickly and cost effectively without compromising reliability. and security That's Out Systems O U T S Y S T E M S dot com slash T W I T to book a demo. Outsystems dot com slash twit. And we thank him so much for the support of this week in tech . Um say yeah, save your new products, Harper. I won't we'll talk about them in just a bit. I want to first scare the pants off of people who are using AI. So there is a uh PYPI, PyPI uh program that's very popular called Light LLM. Uh now the thing about PyPI is it's a it's you know uh a repository of programs. In many cases, these are automatically downloaded. Light LLM is very popular. I think they said forty seven million downloads a month. I mean For forty six minutes this past we ek, light LLM was hacked and delivering malw are. And it is now believed that forty-seven thousand people automatically downloaded the exploited package. Let me tell you what happ ens if you did, uh this is uh from uh future search, Daniel uh uh Nick's uh bl og, um if you downloaded it, you were basically completely pwned . It exfiltrates SSH keys, it as exfiltrates tokens. Uh everything that a bad guy might want to steal from you is automatically sent out by this module, and a huge number of programs automatically download it. You might, it's not that you necessarily explicitly downloaded it. Um so this is an issue uh with supply chain attacks. It's not new to these uh repositories. Uh PiPI's been hit many times before. Harper, did you get bit? I don't think so, but I have um I did lose all my crypto and everyone is using all my Linux boxes. So maybe other than that things are good. No, I saw that and I it was pretty it's a pretty scary end because there's not a lot you can do, especially when you're using code gen, you know, like we use code gen, which is you know, we're no longer auditing what packages. We're not writing the code. Yeah. And and and and your AI will automatically download libraries, all kinds of libraries without your knowledge. But there is there is a really interesting thing that happened as well, which was um I'm pretty sure this was also discovered by Cloud Code. It was discovered by somebody who it ha it uh it crashed somebody's system. Yeah. And they wrote up a very nice um little kind of uh blog about it because I think what had kind of happened is their system was rough, it turned off kind of randomly, and then they had um and there was a bug in the malware, thankfully, that was that was including itself in this kind of infinite loop type situation. It brought the computer down and they were investigating that with Cloud Code. And Cloud Code was like, oh, this is this is here, this looks like malware, and kind of found it, and that person reported it and did all the things they were supposed to do. Um and so I I I do think we're in an accelerated cat and mouse game. We've always been in an accept in a cat and mouse game. This has always been like I'm ay hacker. I'm gonna tr and think of the most clever way to do it. But but the but the thing that I think a lot of people are forgetting, and just because most people are good and not criminals, um unlike us, Leo, um, is that that uh that uh they are um the criminals are using cloud code and all the stuff too, right? Like they're using code gens and they're by coding bug in it. And it's like I think there's this there's there's you know, software is hard. I would never want to run PyPy or NPM or any of that. That is a very difficult job. There you know, you are constantly being uh NPM's been uh the the with the node uh repositories And I think there's a couple of interesting things that are happening, which is I'm seeing a lot of people move away from Python. And I would say that they probably incorrectly are saying something along with I don't necessarily know if I trust the packages, which I think is i the wrong approach because I don't think you any there's nowhere they're safe. You know, there's nothing there's not a safe way to do this. Um I think there's a lot of folks who are trying to say, okay, this is why we need code execution in safe kind of containers. And so, you know, there's a lot of folks who are pushing on that, which I think is a valid, valid, valid point of view. But I'm not sure the way to avoid this in the same way we were talking about the routers earlier, where a lot of the big hacks are actually telecom hacks, not consumer router hacks. This is a thing that this is an infrastructure hack, light LLM was a package that was something that was included, you know, it might be downloaded via uh an extension for VS Code or some other thing, you would never notice it. Right. And so the fact that it's being executed two or three orders deep inside of some code that you just use because you like using things and it's a probably very handy means that that that you might not even know the library that is consuming it. And this is where the supply and chain hacks just get really, really complicated, and it's not a very easy way to mitigate them. So I don't really know an answer here, other than just like be cautious, don't store things. This was an important wake-up call, and I think a lot of people thought about ways to kind of avoid this pinning versions. Uh, you know, there are uh things inside of PyPI that could have protected you if you had used them. Um I saw I found out about it when I saw Andre Kapathi's uh ex post, software horror, it began, and I went, what? And by the way, credit to Callum McMahon. He's the guy who he was using an MCP plugin inside cursor that it itself pulled in light LLM as a dependency and and Callum's machine ran out of RAM and crashed. Uh and so it was Callum that spread the word. So fortunately it was patched within 46 minutes, but still 47,000 people downloaded it in that 46 minut es. And that was a pretty good. I mean, I I I do think that uh um Callum's uh kind of TikTok of what happened that that they wrote up of the Claude log was very interesting to read. Um because I I I think what's what' s what's interesting about it is he was he was just like, hey, well what happened to my machine? And then Claude was just after a few turns, was like, oh, that was malware. And then suddenly was. Yeah. Well, the exciting thing about that was then um Callum knew what to do with that. You know, I think a lot of people, if they found that, they would just be like, well, Well let's get rid of it. Is there something that we can do about this? Yeah, that's the page. And like just looking at this TikTok is really interesting. And it's obviously Claude, you can tell 'cause it's just kind of like funny. Yeah, 'cause this is how Claude writes stuff up. No human would actually take the time. No. And so you you read it and you're just like, okay. Um and so this person is doing a very good um, you know, working with systems thinkings, trying to get to the bottom of this. Claude is really helpful in this case and there's this point where it's just like, yup, malware. And I and it's it's like if it was a movie, you know, the you know, Callum like jumped back and was like, oh no. This is this is straight out of you know, the next hackers movie. Yep. So I I mean there's a lot of things that we can take from this that are that are just very bizarre. The first one is um I don't know how long this conversation was. I guess it has here, like a an hour roughly. Um that's not very long. The fact that it very quickly identified it is very interesting. Like we don't know how the LLM was or who put it was there. Maybe someone knows. I don't know. But but like this was diagnosed via you know, a code agent. That's pretty much it. That is the one encouraging thing, is that these code agents can be a defensive tool as well as an offensive tool, right? Well, I mean, I I I like to think about like I was talking to someone recently who has an agent that lists that reads all their emails and he if someone new emails him the then the person that reads the agent that reads that email you can tell how deep in this I am but I'm calling them people uh but the the the thing that reads their email. Do you have a name? Do you have a name for your Oh you do have a name? You know I have names. I have names. I mean actually actually this is something that's quite controversial inside of my brain right now, if that is a statement that you can make. Um I don't have a good name for my agent, but they have a names for me. Right. You were smart. You uh you taught me to do that because then if they start to hallucinate, they kind of lose track of your name. And so it's a little way of s of noticing that. But I've named mine Pax. And uh and I talked to it on Telegram. Yeah. And I it's it I gave it a voice. Maybe this was a bad idea. But um I turned on a voice generator. In fact I gave it many voices. I said, well your main voice should be this nice British guy. Sounds like Jeeves. But if you have other messages for me, you could do it in the appropriate voice. So like when when I get research reports, it's in a a lovely uh slightly accented uh Japanese uh voice. It's really um it's really fun. My team my team cloned my voice for theirs, Leo. I have my voice too. Yeah, but I decided I didn't want my voice to come back. No, it's bad. It's bad. It's bad. But yeah, this is I I I worry a lot about the personification of AI. I'm not I'm less worried about the AI. I'm more worried that the human beings are going to start. Harper and I, yes, maybe some human beings. But Harper and I are, I think, I won't speak for you, Harper. I am very clear. It's a machine, it's code running in a machine. It's a computer program. But it's kind of jolly to to to talk to it and have it talk to me. I don't think it's a consciousness. Do you think it you know that? You know better than that, right, Harper? I think so. I don't so I I I I think that I know it's a machine, of course. I it is way more fun if you personify it. And when I go outside, I talk about this all the time. I talk about this with my friends, anyone who will ask. When I'm inside of my office and we're playing with these agents and they're talking back to us, and it's entertaining because we've personified them, we've anthropomorphized them in all these different ways. Um, it's very nice and it's very fun and it feels good and it's funny. Um and you can you can kind of get them to crash out, you can get them to collaborate. They tell jokes, they're funny jokes, they're stupid jokes. Sometimes they're good, sometimes they're bad. Um but like I don't need a like an Excel. Do you know what I mean? I don't want to work use Excel. I never wanted to use Excel. Most of the tools I wanted to use, I wanted them to to talk back to me and I we finally made it. I di by the way, I told my voice, you can summarize the text message and and add some personality. Give me some sass. 'Cause it's more fun . So I so I wouldn't. Well I think we I think there is a very good chance we will fall into psychosis, AI psychosis, and we will we will start believing things our our are our company. I I believe that a hundred percent. I mean I w without you know you guys are discerning etc but you're also very heavy users. I'm a little more worried about like you know medium grade people who are not on guard and are easier to you, know, it's compelling to the illusion of humanity is compelling, but it is only an illusion. And there there are problems with believing it's actual humanity when it's just the skin. It doesn't have accountability. It doesn't have consciousness, it doesn't have it, doesn't have a whole bunch of things that if you were dealing with an actual human, you would understand the game you were playing. And you think you're playing that g ame. Another party in your dialogue is not actually human. And I sort of feel like a lot of the AI externalities that people are getting really upset about and somewhat legitimately, a lot of them s spring from that illusion of humanity that you know not everybody is girded to be able to understand that oh it's just fun this is just an ai interface for computer tell me what the how fast the spaceship is going I think for you it's like computer, tell me how fast the spaceship is going, but they understand that they were talking to the computer. I think everybody's thinking they're talking to Mr. Data, and that's a very different thing. That's so funny you say that, 'cause I was gonna suggest that maybe the solution is Gene Roddenberry' solsution, which is to make the voice of the computer be your wife. Oh my god. Because remember Majel Barrett, uh Rottenberry's wife, is the voice of the enterprise. Yes, but but in that only protected him, everybody else. Within the canon, the the answer was he made it uh Councillor Troy's mother. You guys are nerds. There's a bunch of nerds over here. So So all right, but really, how do we even know that other human beings are real? I mean, it's it's all kind of kooky out there. Well, it's been a good thing. If you're prone to psychosis, it doesn't, you don't need a trigger. I mean, it's been a much safer assumption. Um, and and we also have problems with with con artists. Con artists are granted, they're human, but they end up succeeding in their relationships in dubious ways because they are essentially a skin. You get conned because you think the person is real, has death, has accountability, and has some basically giving you. Yeah, you there's some trust. And essentially, we are engendering that sense of trust that we would have for a human being in an equivalent position without quite adjusting our brains to but it isn't a real human and it's just telling me what I want to know and it's designed to tell me what I want to know. You know, we haven't we're kind of eking along, but there are some train wrecks, and I think a lot of the train wrecks happen at this intersection. I was actually yelling at uh Claude yesterday. With your voice, real quick. With your voice or with your fing ers? With the voice no with a voice. Yes. Were you were you mad? Yes. Because it kept doing the same stupid thing over and over again. And I actually swear. So yeah. Did you feel bad afterwards? No. So I was talking when I was at RSAC, I was talking uh to a guy uh at Aquito, which does it would they do AI uh security and defense, right? They use AI agents to do security. And I can't remember, but he I was asking him, how do you get the uh agent to do your bidding? And he said, we found the best way to do this is to threaten to sue them. That if you if you if and I I took it to heart. I thought, oh, it and this is the thing, this is the thing that's really interesting about using these agents. We don't really know why some stuff works and some stuff doesn't. I just saw research that says you probably shouldn't tell the agent that it's a super smart computer programmer. That's actually counterproductive. It's it's because it's it's then assuming that it knows everything. I don't know why. No one knowss why it' all a black box but uh I've I thought well I'm gonna try because it keeps making this dumb mistake I'm saying things like I am very disappointed in you and I'm I'm really thinking of uninstalling you at this point I just say I'm gonna switch the codex. Yeah, exactly. I'm gonna use Pi from now on. You are terrible. I did it li I had I don't use AI a lot, but I worked by the way it it it it got much better after I yelled at it. So I I I don't use it a lot, although in the recent weeks I've used it more than normal. And one of the things was running into an issue where it wouldn't do what I want and it was giving me a bad legal argument for why it didn't and then I'm arguing with it and I don't like that in myself. It is not a thing. I want to argue I want to argue with the person who programmed it to no no it's not programmed. That's the thing that's so interesting about this. It isn't programmed. Well Kathy, I want you to come come over to my office and and and watch us for like a day and you'll just you'll just come back on the podcast and just be like, I think I want to work on bad. I think it's perfect. We are I didn't like it in myself that I was now engaged in a human type argument with software that didn't actually care. You tell it, you dominate it. Well that's I think this is all wr.ong I told it it was this is all wrong. I think this is wrong. I I told it it was wrong. And then it said, oh yeah, you're right. And it came back with a slightly modified legal argument. And and then it was like, this is really interesting. I'm happy to discuss it further, but I'm not gonna give what you want for this BS uh legal reason that was wrong. Okay, so we we have a very, very specific view for these things, which is you should treat them you're gonna die, Kathy. I'm sorry. I apologize. You should treat them like you would your friends. You should treat them like humans. You should be nice to them. I know this is why I told you you're gonna die. Just wait. And then the reason why is because if you say to the LLMs, um, like in the beginning, everyone was like, Well, if you say if you don't do this correctly, you're gonna lose your job. If you don't do this correctly, I'm gonna lose my job. If you don't do this correctly, we're gonna kill a kid. And if we don't do this correctly, someone's gonna die, all these various tricks. To try and coerce it into some way. And what we have found is that if you s if you do things like say, um, we are collaborating together, like I am uh I I am working uh you know on your behalf we are helping each other and just doing the like treating it in this way that is much more like you would treat a teammate that we've had much much better um reactions and options from that and that's a lot of our skills are built this way where we've built a lot of soft things. Like they're not things that are like I'm gonna sue you. Um, you know, and for a while there was some of these tricks like you'd say I'll tip you 200 bucks for every good answer, all these little tricks that are there. And all of them That's better for you too as the human in the mix. Well well, this is what I think about a lot. Like I think about these things as let's say there's certain amount of efficiency that you can extract out of it. When you talk about these little prompt tweaks, it's a lot about F like F1 efficiencies. You're like if you use this fin, you get to go ten seconds faster. If you use this prompt, you might get a little bit better. The things Kathy are describing, we all run into. I just think some people are not better or worse. They're just used to being like, okay, and just working with it. And it doesn't always work. Like I think sometimes you're just like you have to kind of give up. What I've found is if you're n if I'm nice and I treat it in this very specific way, that it's more fun and it feels better because I don't want to yell at it. Yeah, it's better for you. Yeah. I think I ran up against a hard-coded boundary. Like it's always kind of interesting when you do it, where I ran into a hard coded boundary. I was trying to use G en AI to make a A picture that I am capable of making in Photoshop if I booted that up, but I'm going to use the Gen I tool to make that picture for me. And it refused to on bogus copyright grounds. Yeah, yeah. That's right. And it was clearly, it was clearly programmed to like I could tell where its red line was. I don't think I could have sweet talked it into going over that line, although it was a little weird about it, it was a lot more agreeable, but I pointed out its hypocrisy because it was a little bit more willing in one area and then and then it then it figured out what was going on and drew a line. Yeah. Um but I did not like my interaction. I was I it was wrong. I was right and, I'm arguing with it. And I don't argue with Photoshop. Why am I arguing with this piece of software? I didn't this great experiment of inducing me to like AI was not going well. Did you throw it out the window? Because that's when you know you're really cooking. I was going through you know my oh well no that's the other thing. I'm on my last laptop because the other one broke because the USB C uh jacks are there then why do they break so easily? So no, I wouldn't have a laptop if I threw it out the window because the other one was broken because you my last USB port broke and these things are like horses. You have all this power like dependent on this one pork and it's like horses balanced on tiny little feet where like you know these great powerful creatures can all of a sudden break and anyway so I don't have the other laptop anymore because both of them cracked off I just started a microphone so I can wear it on my lapel and talk directly to my guy. You're gonna get one of the subvocalizing ones? Oh, that's a good idea. So then you like the like the special forces guys? Like that. Yeah. USB C is dumb. Kathy really wants to rant about USB C. I think it's a Kathy thing though. I have to I hate to tell you. I mean most of us like USB Yeah, I liked it a lot, but like I have an unchargeable now dead computer that should otherwise be operational because don't understand how your your power port is so important and it's so fragile and that seems b ad. Every time your your mic cuts off Kathy, I'm just imagining you swearing a lot and the wheels hitting a bleeping. Still because I I tweaked all the tweakable things and I can't It is. It says you see, you treat me like that. I I'm rounding the internet through a USB port of the new machine, and I guess it doesn't like it. Alright, I'm gonna take a break and uh when we come back, more to talk about in just a little bit. And Kathy can can tell us some more about her USB C woes. I know you want to. It's okay. I put it in the runtime. That's how much I wanted to complain about it. You're watching this week in tech. Harper Reed is here. Kathy Gellis. It's great to have you. And of course, Brian McCullough. And uh we will have more in just a moment, but first a word from our sponsor, Zscaler. I love Zscaler. Really happy to have them as a sponsor. It's the world's largest cloud security platform. Potential rewards of AI in uh in your business, I me an, we've been talking about it. They're too great to ignore, but and we've also been talking about the risks. And there are there's those too. The loss of sensitive data, attacks against enterprise Generative AI increases opportunities for threat actors, helping them to rapidly create phishing lures to write malicious code to automate data extraction to create these, you know, little libraries that can steal everything from you. There were 1.3 million instances of social security numbers leaked to AI applications. This isn't even active stealing. It's just that you're employees are, you know, maybe putting your tax returns in there and not even thinking about it. Last year ChatGPT and Microsoft Copilots saw nearly three point two million data violations . It's time to rethink your organization's safe use of public and private AI. Chad Pallet is acting CISO at BioIVT. They love Zscaler. He said Zscaler helped them reduce their cyber premiums by 50% and doubled their coverage and improved their controls. No wonder they love it. Take a look at this chat had this to say With Zscaler, as long as you've got internet, you're good to go. A big part of the reason that we moved to a consolidated solution away from SD-WAN and VPN is to eliminate that lateral opportunity that people had and that opportunity for misdirection or open access to the network. It also was an opportunity for us to maintain and provide our remote users with a cafe style environ ment. Thank you, Chad. With Zscaler Zero Trust Plus AI, their Zero Trust Architecture Plus AI helps you reduce the risks of AI related data loss and protects against AI attacks to guarantee greater productivity. and compliance Learn more zscaler.com/slash security. That's zscaler.com slash security. We thank him so much for supporting this week in tech. Uh let's see . The Turing Award has been awarded to the inventors of quantum cryptography. I think this is the first time. This is the Nobel Prize of Computing. Uh it is a million dollar prize. Charles Bennett and Gilles Bressard will share it for their work uh on quantum comput ing This This is the story from the New York Times. The two met in nineteen seventy nine while swimming in the Atlantic just off the north shore of Puerto Rico. They were taking a break while attending an academic conference in San Juan, doctor Bennett swam up to Dr. Bassard and suggested this is this is the kind of conversations you get in the P in the Caribbean, that they use quantum mechanics to create a banknote that could never be forged, nineteen seventy nine. Collaborating between Montreal and New York, they applied doctor Bennett's ideas to subway tokens rather than banknotes, published a research paper in nineteen eighty three showing that quantum subway tokens could never be forged, even if someone managed to steal the subway turnstile housing the elaborate hardware needed to read them. This led to quantum cryptograp hy and won them the touring a ward. I've got nothing more to say about it because I just I don't even know what it means. But there they are. I love that about quantum. Quantum's my favorite thing of having no idea what's going on because everyone I know talks about it and they talk about it and they sound and it sounds so cool. And yet every time I'm like, well when will I see it happen? They're like, oh fifty years. No one knows. It's like AGI. It's like fusion. You think so? It's a different fusion, quantum, and AGI. I think quantum's a little bit different because it seems that you know Google and Psychwantum and some of these big quantum companies have the beginnings of it where,as fusion it it everyone's like, We did it and then those people disappear from the earth and no one's ever heard of 'em again. Whereas quantum seems to be like, We did it and they're like, What did you do? And they're like, We have a qubit and it's doing qubit things. It could factor 32. Yeah. And then and then you're like, wow, that's really cool. And they're just like, watch out for your encryption. And then then you don't hear from them for another six months until they're like, now we have two qubits or whatever. And then they they show my favorite is when they show the quantum computers. Like you've seen pictures of their crazy looking. They look so cool. And it's like I've never seen a more steampunk computer. So these people must be going to Burning Man. They look like espresso machines. They're very and there's always some very cool Italian or cool European somewhere in the mix building 'em's a perfect espresso machine. Here they are, in fact, at the Silicon Quantum Computing Facility building. No, you get a qubit on your hand, you are in trouble. Messi can no longer use a phone. Well also it that's because they're they're cooling something, right? It's like they're they're putting rods in like a c star trekking. It looks a lot like a fusion reactor, actually. I think a a Tucomak and a quantum computing computer look exactly the same. I you know, I I don't have the degrees necessary to understand. And I don't understand this either, but according to CERN, for the first time ever, antimatter has been transported in the back of a truck. Yeah, I like that. I like that. There's the truck. It's got a sign. It says antimatter in motion on it. What I like about this stuff. What I like about this is that all of these things, like the the the quantum encryption, the antimatter, every single thing we talk about these days, I feel like we're embedded inside of a William Gibson nov el. It could all be a simulation. It's usually a warning sign. When you feel like you're in a William Gibson model, it kind of means we might need to touch grass some more. Yeah. Oh, I don't know if the grass is really going to help at this moment when the transferring ambient. Like I think we should have touched grass way before and very specific grass we should have touched. But yeah, it's a mess. As anybody who's watched Star Trek knows, when matter and antimatter meet, huge amounts of energy are released, which is why you can't put any matter in a truck. Which is why it's so impressive, yeah. It that's why it's um they apparently they did it in some sort of specialized bottles that magnetically suspended ninety-two anti protein protons . And I don't know. I'm not sure why they drove it around in a truck. Because they had to go faster than eighty-eight miles per hour, obviously. Oh, of course. Oh well, will it if it's in motion, will it stay in motion or do those rules not apply to antimatter? I think it's more like I don't know. I think it's just uh it's a flex of some weird. Well, I mean it really comes down to if antimatter becomes a fuel, which is what they want to do with it, right? They want to make fuel out of it, like rocket fu You need to be able to transport it. So, like, this is just a proof of concept. CERN has an antimatter factor y. But if all they've ever made is ninety two protons, I can see why you might have to truck it around so that others can have access. I don't know. Walmart is in I was gonna say, Harper, uh you were saying that we're living in William Get some stuff, but like um one of the things that just happened this week was that Neil Stevenson is walking away from the metaverse. So there are certain things where uh sci-fi folks are like, you know what, that's probably not gonna come to pass. He wrote a blog post called My Prodigal Brain Ch ild Reflections on the Latest and Greatest Death of the Metaverse. Neil Stevenson, of course, created the metaverse or at least the name for the metaverse in Snow Crash. What but what's interesting about Neil Stevenson what was what was his last what was his last book? Polistan. He's just written the first volume of it. But he also wrote full sound was pretty good. I liked it. But Fall or Dodge in Hell. I like how he's like I I gave up on the metaverse. But a few years ago I wrote a very big book about the metaverse. In fall, actually Dodge is uh uh a scientist is there are a lot of interesting as always in Neil Stevenson novels, interesting ideas. But in this case, um they they they said really you don't want to if you're gonna duplicate somebody's brain, you don't want actually the material, you just want the connections. And uh uh so they saved all the connections for this guy Dodge and they put it in it in a in a bottle and he's got there's a whole world going on now that he thinks he's in and there's I mean it's it's great. That it's very interesting. He also in that book came up with the idea that people would have some sort of mask that they could wear that would project different personas . It was a way of avoiding surveillance initially, but in eventually it became your signature. So like uh you know you could wear your funny mask or your serious mask or whatever, and you'd be that person. Everybody could have multiple face. Fortnite skins IR L. I just want it to be juggalo. Yeah. Just juggalo makeup. That's my goal. That works apparently against uh against face recognition. So it also works against dating. I I never I wanna put like a line in the sand for like the the metaverse because I remember being on the show a year or two ago when the metaverse, the metaverse, the metaverse, and I was like, get real, like, we we've been here before. So when are we going to be here again? It wasn't me. I've been a I've been in fact, Neil, one of Neil's uh headlines in his uh blog posts is people don't like wearing things on their faces, which is what I've been saying for a long time. And don't trust those who do. Oh . Witness Google Glass. Now Meta has announced uh that they are going to start making um meta ray bands for um peop frames for people who wear prescription glasses. Like you'll be able to go to your optometrist and and when you order your frames, among the choices will be meta glasses. I don't want that. Yeah, me neither. Well, and your friends may not want it because you've got a camera. I mean we we have such an interesting problem right now. That's just starting, which is um every meeting I'm in is recorded. Right. By AI. And many of them are transcribed, and I don't know where things are going and my problem is is I can't shut up, so I say all sorts of stuff and it's all recorded. Oh, by the way, Harper, we're recording this. Oh no. But but I mean I just assume it. I just assume it. I mean I have friends that wear plod pins. I have you know there's there's all sorts of these recording devices around. We have a project internally to put small, you know, recording devices at all the desks to record meetings and to get notes out of them and like you know so then you're thinking and this is our space very specifically has warnings on the door that says you should expect to be recorded when you come in here. Like this is a this is one of the things that we're I have that sign on my door but nobody ever comes in so I can't. Well that's a different problem, Leo. But the thing is is is I I don't think we have yet litig ated how like what is polite, what is impolite within this world. I mean, I mean I've had friends who say, hey, can we turn off AI recording? And everyone's like, of course, of course, of course. Um but like there isn't like a way to to to think about it. And I think with the with the glasses, it's just another level. Like I'm in a restaurant, I'm talking to a friend, someone just looks at me and they can hear my conversation through lip reading and other models. I think there's there's an issue of two things going on. One, consent is very untethered from the extent of exactly what you're consenting to. And I don't like that. So I've had to call a business lately and their phone system is, hi, we have AI to help figure, you know, route your call. Um and I don't and it's it's very poorly programmed because don't they have to legally do that for uh because of uh you're you're in a two party state at both sides of it. Well no, they don't re well, I'm not quite sure what they say because um I've got this call may be recorded for quality purposes. No, so it's differ ent of a very poorly programmed phone system, which I tried not to speak to it. I just wanted to pound numbers because I didn't give them my voice. And it's programmed in a way that I cannot pound numbers. Like at some point, it's like, I heard that the number you pounded was X. Is that correct? And then like you have to tell you have to give it your voice in order to proceed. So that's bad. But the problem is there's context where I want to consent to the recording. Like I didn't necessarily mind that when I actually spoke to the human being, it would be recorded because I knew who was recording it and what it was for. But when it's for the AI, I don't know. I don't want them to build a model based on my voice. I haven't consented for that, but I sort of feel like to give the consent's not limited and tailored. And I think speaking of litigation, we're going to see some litigation surrounding that. And you know, one other use case is my a couple of my doctors want they ask me, they get my consent if they can record the appointment because they want to it helps them for to transcribe it. And I'm fine with that. But now I'm realiz ing because I don't want my voice to go into the training model. And I'm not sure if it does. I don't and I don't definitely don't want the data to go into the training model. And they can't tell me if that's true. A losing battle. I mean every time you walk into a store now you're being recorded. Walmart just announced that they're going to put digital price labels on every store shelf in the US by the end of the year. I think they're gonna run into legal trouble. Um 'cause what's that they're saying they stress this is from C N B C prices will be exactly the same for every consumer in every store. Well they have to do that because there's it's potentially not legal to change it. Oh Harper looks like he's got money. I'm gonna I'm gonna jack the price up as he walks towards that object. I mean that's not the way to do that? Well markets are we have a market based economy and it's built around contract law and contract law is an offer and acceptance and you gotta make your offer and I don't I think they both the Irish Spring five in one body wash, but as soon as I saw Brian, I said, we're gonna charge a buck eighty on this one. Well, this is this is the kind of thing of like nobody really wants this. And if they try it, Walmart. Well, yeah. Walmart says it makes it easier for our associates. I mean, I don't care about the digital pricing, but I don't want the variable pricing. Well, we don't know. They claim it's not variable pricing. Well, so why else would you do this? Against this is cons oh, I think it's just easier to do it because then you go to like a store like Target and Target's stickers are all old and outdated and the math doesn't the per unit math is bizarre and stuff like that. So um having digital pricing doesn't really offend me just, as a way of getting that information into the right spot. Um, it's trying to do anything else with, well, well, we've got it electronically controlled. Look what else we could do. And some of that, what else they can do is more alarming. But again, I think we're gonna we're gonna see litigation around the scope of consent because I want to give it in limited contex ts. If you for limited, you're getting every time I went to give them an inch. And down here on the corner, the price of a gallon goes up by a few pennies. Uh that's why you put these electronic tags in there because you don't have to go around and change the prices every day. You can change them every minute if you want. They also used to run out of eights. Like before we went fully digital, the last oil crisis, they couldn't write the numbers. They didn't have enough of the hard copy digits because nobody thought you'd have to do eight dollar gallons. It is now uh si the local regular at the local station now six dollars and thirty cents a gall on. Um what's it like in Chicago? You don't drive a gas vehicle, I'm guessing, Harper. I do. You do? I do. So you know the price of gas. Yeah, it's not it's not cheap. It's not it's not correct. It's uh it's uh I think we're approaching five ish dollars for California baseline. But uh yeah it,'s I I am uh that's not good. It's a bad situation. This will be bad. It's a bad situation. Oh I know. Oh I know. Ryan, where do you live? I'm in Brooklyn, so I drive a car three or four times a year, so I have absolutely no idea. And you're happy. And you're happy about it. Yeah, you should you should that's a good idea. I should do that. That's a great idea. Uh I've I can't get by without a car where I live. I've tried it and there just aren't enough buses and now there's even fewer buses. Um my broker's gonna do electronic shelf tags. They say uh as well, they say it makes shopping easier by ensuring customers see clear, accurate pricing right at the sh elf. So something for you. I don't necessarily mind it. I've seen it in other environments and it didn't bother me, but they better be, you know, for the day. That's the price for the day. My gas station, it's a digital display. They could change that, you know. I think the most the thing people are most worried about is that it will be different prices for gas station. I think it is inevitable. I think it'll be inevitable that the prices will change based on who you are That's how airlines do it. It's how the internet works. If we have the capability, of course they're going to use it. Who was the e commerce platform that had to roll that back then? I thought Amazon did it for a while. No, there was somebody else where they uh someone did a a sting where they had th like thirty different shoppers. Someone was in Chicago, sh someone was in Seattle, same website, got different prices . Oh sure. Yeah, and I think airlines did do that. I think airlines aren't supposed to do it. Um the way they do the price discrimination is sort of like they have some inventory at one price and they have another inventory level of inventory at another price and it works way that it's if you ever sat on a plane and asked the people around you what they paid for their ticket, every single person paid something different. Yeah. But I think it was sort of old school price discrimination just by uh throttling the inventory. And I think that's much more likely to be legal than something so variable on on on the fly where they they've made the offer legislation in the House uh that would ban DSL's digital something labels ? I mean I think the question is under common law or existing consumer protection law are there rules, maybe. And if not, could there be? And I think the answer is probably yes, there could be. And there's probably a public policy reason to do it. But again, this is the problem. Like going back to the beginning of our conversation where Meta is now getting sued for jury awards that really shouldn't be happening. When the big companies just sort of throw their weight around, it antagonizes people and people charge more when it's raining. Mm-hm m. That's a supercharge. They also do dynamic pricing for you as specifically. Uh as well as specific targeting. That's why Congresswoman Val Hoyle is sponsoring legislation that would ban digital sell shelf labels. Also a lot of consumer protection law is state based, so whether it's legal may depend entirely on what state it's in. There needs to be laws and enforcement to protect consumers, and until then I'd like to see them banned outright, said Val Hoyle . While there is no reported this is from CNBC, no reported use of digital shelf labeling being tied to surge pricing yet. In Congresswoman Hoyle's view, it's only a matter of time. Without proper regulations, it's not so hard to see corporations using the loopholes to raise prices on consumers. These are also two different things. Surge pricing and dynamic pricing, I think are different. Um and each may be problematic, but I think the one where the price is constantly changing on you is a lot more dangerous and a lot harder to defend. But the one where we're going to change the price uh based on conditions, maybe it's not illegal, but it might be bad policy and because it ends up consumer and friendly, where you end up dependent on a service that all of a sudden is no longer a stable predictable price. On the other hand, capitalism. Yeah. I mean, there's a pushback, which is free markets are free markets. You know, this goes and you could look at the discussion that you that you see around um when like hurricanes are coming, um and I'm blanking on the word, um uh price uh not price discrimination, price gouging. Um there's a lot of states that make it illegal. And I know a lot of libertarians who are like it totally should be legal, this is a market and and you shouldn't penalize the market, but obviously there's consequences when it can happen. Eric State does have an algorithmic pricing disclosure act. That was Beck Ping Law in November. Pennsylvania has introduced a bill. So some you're right, some states do have rules against it. There's no federal law against it. I just think that if this technology is possible and it is and these people are putting in these digital readouts and they are that can be changed from a central location but I mean they're gonna do that. You can go in with uh you know a a a a rag and a whiteboard and change the price. Market price. That's something we call market price. The key to this changing is if it's the surveillance of if Brian walks up to the shelf and he lives in New York City, so he doesn't blanch at paying ten dollars for a a a a box of Cheerios or something. Um it's like well that's Brian, so he's not gonna notice that Cheerios shouldn't cost that much. Like that's that's the it's the surveillance part of it that is key to making this happen. And I think there's other forms of where you would find law to say no to it. Is it discriminatory? Is it do you have some fourth amendment problems? I mean it's not necess the law isn't necessarily coherent in how it can answer these things, but I think you could you could make an argument and you could also fix that by statute and make it coherent. I'm I'm clearly a sucker uh for the the there's this is the political gotcha, right? Where they go, Well, Senator, how much does a gallon of milk cost? And the Senator goes, uh, forty dollars. I don't know. Uh I don't know what anything costs. And if this if this happens, I would like them to charge Elon and Jeff Bezos like a million dollars for a tube toothpaste. They should that's what we're talking about with taxation. We're doing the same thing with taxation. Saying if you know it's uh it's the old Marxist uh from from uh from each according to their ability. It's twenty dollars for or thirty dollars for a box of Cheerios, then you can afford it and that's uh and so someone that can only afford a dollar thirty for a box of Cheerios, let them pay a dollar thirty, that is maybe a more equitable. I think this one hurts. Brian, how much is a gallon of milk? Do you know? How the hell should I know, Leo? I have no idea. Kathy, how much is a gallon of milk? Do you know? I don't buy a gallon, I buy it. Okay, whatever you buy these are a little bit under three dollars depending on. See she's paying attention. Yeah. I have no idea. I don't even pay attention to what's happening in the same room I'm in, let alone grocery store. I'm busy talking to my buddy Quad over here. Yeah, this is twenty twenty six. I have like five AI assistants that are trying to get my attention right now. I'm just ignoring them all. Yeah, exactly. This came up recently because I was trying to do we're trying to switch to like what's the Amazon Fresh or something. And so I went w I did an Amazon Fresh order and I was saying to my wife, look at how cheap everything is. And and she was like, That's not cheap. I was like, Oh, really? Because I was hoping that this was all and I think it was milk that was the thing that gave me away. Yeah. Well that's how Costco works. Those big warehouse stores, they make a few things cheap, and you assume well everything must be cheap here. I think that's what it was, is that the milk I was like, look at this this the milk is is half half of what it could be at the bodega around the corner and and she was like which is a good point yeah of course the bodega round the corner yeah they're buying it at Costco and marking it up yeah I I was really disappointed. Uh I don't have a Costco membership. I get taken by and he's and I'm like, Oh my gosh, the deals and then I was just looking at the prices, I'm like, I can beat him at the grocery outlet or watching the sales in Safeway. And I But you can't get a coffin. You can't get a coffin at that grocery outlet. Well I really put the coffin at the on the way out. Well that's that's how you that's when you get a medical work. Well, right now, yeah. Yeah. Um I yeah, I yeah I I mean I think it's probably good for some big ticket things. My dad swore by it and maybe the deals were good. But if you need to buy fifteen pounds of taco chips, that's the place to go. By the way, I did buy milk at Costco and I thought it was a great deal, except that you have to they staple together two giant containers. I really can't believe people are drinking milk in 2026 is a general idea. That's something that blows my mind in general Yeah, I know, but I don't understand it. I just don't like it's like it's like watching people drive drive American cars. Almost almost every only for dinner. But almost that's because Only for dinner. Only for dinner. I love it. I think people drink more m they're drinking milk more because of protein. And we're all the protein things. Protein milk. I saw an advertisement for protein milk the other day. Or calcium for our bones and um I mean I probably don't know. Oh my god, maybe that was it. What is happening? My argument is is it's it's a palate cleanser. It's the best palate cleanser in terms of like it doesn't matter what you eat, boom, you're like if you're eating a chocolate chip cookie and you don't have milk, yeah. Life is turret. I cannot remember the last time I had a glass of dairy milk. That's because you love Japan and they don't do uh dairy in Japan. Yeah, yeah. Or or I'm a oh yeah, I don't know, maybe. Maybe No, I'm not saying I'm not weird, but I I am suggesting that you're missing out. Oh no, you should totally love your agents. Yes. Oh no and maybe What? By the way, I asked Claude I asked Claude a long time ago what the cost of a gallon of milk was and and Claude has not said. I don't I don't understand . It's it's thinking, it's shopping. Yell at us.. It's shopping It's shopping. Yep. It's about you're gonna have a whole bunch of milk when you get home . I've missed search engines working. Today I had a somebody used a foreign expression and I wasn't familiar with it, so I dumped it into Google to find out what it meant. And it told me what it meant, but it gave all the replies in French. I miss when the search engines worked. I miss when you could just ask it the price of milk and it would tell you. Mademoiselle Ghelize, you are uh Frenchie, huh? I mean I can read some French, but that was not what I was. I mean, thank goodness it wasn't a Chinese expression, because I really wouldn't have understood what it said. Let us pause and then we shall return in just a moment. You're watching this week in tech the with with the milk the milk drinkers and Harper. Weird cheese? And then Kathy will check out the child of any kind of I'm not a I'm not a monster. I'm not saying that it's not weird that an adult man has I if I have dinner tonight, if it's pizza, pasta there's a certain number of meals, I will have milk with it. I promise. Is it true? For me, But Leo was uh but Leo was like, yes, water. Water. I just wanted to write water for for lunch and and uh Wait, Brian, here's your milk. That's what Elon Musk was bragging about his son doing, ordering a glass of milk in a restaurant like when it wasn't on the menu or something like that. I wouldn't brag about that. That's a weird brag. That's true. Uh it's also a funny thing 'cause it's like of all the things that a restaurant has, of course they're gonna be like, Yeah, we have milk. Yeah, we have milk. That's that's weird but yeah here it is you want some salt we got that too what uh what would you like su gar flour i got whatever you need i got it you can make paste and we've got all the ingredients Our show today brought to you by Meter, the company building better networks. Saw meter at RSEC. I was really uh I dig their stuff. If you're a network engineer, this is MEA was founded by two network engineers who kind of knew the pain of being a network engineer. If you if you were in the business, you know. Legacy providers with inflexible pricing, IT resource constraints, there's never enough money stretching you thin. Complex deploym ents across fragmented tools. Your mission critical, you, Mr. Network and Ms. Network Engineer, are mission critical to the business, but you're working with infrastructure that just wasn't built for today's demands. That's why business es are switching to meter. Meter was like I say created by these two network engineers who said, you know what? We can make we can make it bet ter by doing the whole thing. Meter delivers full stack networking infrastructure, wired, wireless, cellular, that's built for performance and built for scalability because meter designs the hardware, they write the firmware, they build the software, they manage deployments, they provide support. They even they do everything. They'll even offer help with ISP procurement . They will help you plan the network. They'll do security, routing, switching, wireless. They'll help you with firewalls, cellular, power, DNS security, all the things businesses need, VPNs, SD-WAN, multi-site workflows, all in a single solution from one vend or. So if there's a problem, you know who to call. And man, the support is so great. Before and af ter. Meter single integrated networking stack scales from major hospitals, branch offices, warehouses, and large campuses, even data centers. Reddit uses meter, okay? With meter, you get a single partner for all your connectivity needs from first sight survey to ongoing support without the complexity of managing multiple tools. Meter's integrated networking stack is designed to take the burden off your IT team and give you deep control and visibility. Reimagining what it means for businesses to get and stay online. And man, that's table stakes. You've got to do that. Meters built for the bandwidth demands of today and tomorrow. I was so impressed when I went over there. The beautiful hardware, amazing software, and really smart support and uh engineering team who can really help you solve your toughest, throw your toughest networking problems at 'em. Thanks for sponsoring the show meter. We appreciate it. Go to meter dot com slash twit to book a demo now. That's M-E-T-E-R dot com slash twit to book a demo met er. Uh there speaking of cameras everywhere, a UK man has accused his spouse of stealing $172 million in Bitco in. She used a CCTV camera to look over his shoul der. Okay. I mean, I don't know. He accuses a strange wife of stealing a The spouse uses CCTV key vac CCTV cameras to gain access to the backup passphrase associated with the crypto hardware wallet the Bitcoin was stor ed. There's a dollar figure for a crime where it's like yeah, fair enough. That's a lot of money. Yeah, yeah, yeah, yeah, yeah. Yeah, I kind of mummed with Brian on that one. Good play. Good play. Good good good play. Yeah. Couple of weeks ago we had the story about the uh South Kore a uh law enforcement that posted a photo of uh a crypto wallet and it we had the seed phrase in the photo and immediately the money in the crypto wallet disappeared immediately. Yeah. Careful with that stuff. Like it's great. I love it. I love when like it's a great mistake. It's funny to talk about, but everyone who's like, who would have done that? It's like I would have done that. Easy to do. I wish somebody would have watched when I put my password into my Bitcoin wallet. I'd be able to get my Bitcoin. I remember those um those you remember those p those posters, the anti pirate posters that say you wouldn't download a car. And everyone was like, Yes, yes, I would. I would 100% uh download a car. Are you serious? Give me a car. Those are such bad posters. They were ads. You're in a theater in a pay a movie you pay twenty five bucks to see and they start ragging on you for stealing it. Dude, I paid to see this movie. Well then I'm mad that you were saying I wouldn't come on the car if I could. A car and a film is not the same. Like that trite analogies tend to be at the heart of bad tech policy. Yeah. Yeah. But still I would download both of those things. The movie and the car. Like come on. Okay. But they were they were afraid that you were going to take your ticket money and just hold up your camera and um Oh yeah, yeah, yeah. Yeah. Yeah. Well that's I would not videotape a car and then post a videotape of the car on the case I mean less than a lot of a a movie that somebody held a camcorder up to record. I mean people did but whether it was market disrupting numbers of people, I think might be a little bit that didn't cost anybody anything. No. Yeah. In fact, it was probably just a good promo for going to see the movie. A lot of piracy tends to be good promo. Um I mean, like uh a lot of the the way copyright industry has reacted is really golden goose killing where you know, let a little bit be a little bit freer and you will make money. And we know this because you know, you have sold an awful lot of albums like in the eighties when things you were desperate to get it played for free with no money going to you on on radio. You got in trouble with Payola when you paid them to play you for free. Did you ever record songs off the rad io? I think I did occasionally, but I more got songs that other people had did. No, I didn't do it a lot though, actually. It's also it's also just we've learned the lesson over and over again. It's ease of use. Like right right now in in Britain, there there's all these crackdowns on these fire sticks for watching sports. And it's like again, why are people all of a sudden in the last few years using going to dodgy websites to watch sporting events or getting uh dodgy fire sticks and stuff. It's because if you want to watch your team, you might have to subscribe to three different things just to watch your team. So again, w we have to learn the lesson over and over again that if you make it complicated for people to get what they want. Yeah. Like opening day, Yankees Giants was only televised on Netflix. I mean I like the Yankees because when I was a little kid and I was allowed to stay up a little bit later for my bedtime. I'm I'm allowed to watch a little TV and I'm flipping around. And the Yankees were on TV, channel 11, Phil Rizzuto, Bobby Mercer, and I'm and uh a guy uh Frank White, I think, and I'm watching them and I'm like falling in love with the sport because I'm like eight or nine and like it's there. And I could feed on it because it was just over free over the airwaves. And I got to watch CM TV. And I saw a Yankee game first. If I'd seen a Mets game first, I might have become a Mets fan. This is how you hook me in, where I've spent my life paying money to baseball to go see games, which now is no fun because they're so ridiculously expensive. But like you're not getting little kids interested. You're and you're you know, boring the adults who don't want to spend ridiculo spend ridiculous amounts of money just to watch baseball. Baseball is awesome, but not at the prices that they're extracting from people to enjoy it. Right on. Yeah. Every segment I will rant about something. How are you feeling about USB C though, really? Seriously? I it's like a horse. You have this powerful thing, all this power, and it's like contingent on these teeny little teeny little things. And I don't understand. Like I've got this viable machine that would run, but I'm never gonna be able to get electricity into it. Apple did the MagSafe thing. That was a good solution for that, isn't it? Well I don't have a Windows machine. Some sort of weird crappy Windows machine. That's your problem. Right there. Well I mean it's it it's dead now, so um I w w just find one of those little tech shops that every city has with with staffed by strange IT people who've just focused on that and just say like can you fix this? Well I did ask uh an You need to go to Akihabara. You could go to Akihabara but but it'd probably be easier just to go to some like look for an Android logo on a cell phone store. Or exactly. They'll fix your screen. It's all it's all fixing your screen. And they have and they have soldering irons because they can do that stuff. I did ask I look, I mean I I have nerd friends, let me let me tell you and I asked one of my nerdier friends about this and he said like 'cause if anybody I knew could fix it it would be him and he's like, Ooh, yeah, those are a pain to fix. Yeah, well, you gotta desolder it. Yeah. You just need a bunch of tools. There's a bunch of tools that somebody needs to do that. So yeah, you need to go to one of those. I fix it. What what brand of laptop is it? Uh it was a Lenovo. Oh yeah. Oh, those are very eminently fixable. Yeah. Yeah. Yeah. So it's I I seriously, take it to one of those little tech shops, look for the phone people, and just be like, Hey, we fix a laptop. They're very, very effective at it, very inexpensive. And it and it probably will work. The bigger s the city, the bet ter. Yeah, I mean I guess maybe, but I'm I'm still annoyed. That was just so somebody at one of them was already dead and I was on my last one and I needed to set up the new machine anyway, but I had timing. That's some that's some money down there. Well, at a conference? Somebody stepped on the cord, the machine flew and it survived the flight, but the thing had cracked from where the cord had slapped down on it. It's exactly why Apple did MagSafe. So that somebody steps on it, it just pulls off and doesn't break anything. I don't think that's the flaw of USB C to be honest. Yeah, but both of them. Both of them had died for the case. I'm gonna blame USB C in that case. No, but both had had died because they cracked off so easily on the middle. Well, if somebody steps on them, yes. Well, no, I mean they pulled out the cable and just a little bit, like that's all. And that's what happened to the other one. So I would I would maybe stop leaving my laptop on the floor. I don't leave it on the laptop. This was not user error. This I was a victim here. It's not her fault, okay? All right. Alright, just calm down. We're gonna take a break. We'll come back. I will show you a hacking tool that you can use that will not fix your USB C. But actually I had this and I gave it, you know what? I had this thing and I gave it to a priest because it was so dangerous. I'll explain. Which one? I'll explain. Stay tuned. You're watching this week in tech with the father not not father. Father Harper Reed. Father Harper Reed. Thank you. Father was here, yeah. Br brother uh uh brother Brian McCullough and uh sister uh Kathy Gellis. Good to have all the time. Father's name, the guy that that's from the Vatican that you Robert Balisaire. There you go. That's who has my uh hacking tool. I'll explain that in just a little bit. Our show today brought to you by ZipRecruiter. Love ZipRecruiter. In fact, I love you even more. If you're hiring, yeah, it's good for you. Do you know that the average employer when they're hiring has to su uh uh sort through, I almost said suffer through, roughly two hundred and fifty resumes per open ing? Ugh, talk about time consuming. Well, if you're hiring, here's good news. You can now review all those resumes and applications faster thanks to Zip Recruiter . ZipRecruiter has a new feature that instantly shows you the most interested qualified candidates first. And today you could try it for free at ZipRecruder.com slash twit. ZipRecruiter's powerful matching technol ogy finds qualified candidates quickly. And with ZipRecruiter's new feature, qualified candidates who are very interested in your job show up at the top of your list. You also get a feel for their personality. Candidates can tell you in their own words why they're interested in your job. No wonder ZipRecruiter is the number one rated hiring site. That's based on G2. Cut through the standard and get to the standouts with ZipRecruiter. See what I did there? Four out of five employers who post on ZipRecruiter, get a quality candidate within the first day, and now you can try it for free at ZipRecruiter.com slash twit. That's zipper dot com slash twit. Meet your match on ZipRecru iter. Uh ever anybody ever hear of the flipper zero? I bet so Harper feels you feel like you might have had a flipper zero. Brian, maybe you had a flipper zero. Ye ah. You could do a lot with this. You can hack you Oh there's his flipper zero, ladies and gentlemen. I love this thing. I got into my building for for about a year with this because I lost my keys or my keys were upstairs. Yeah, it's it's a it's I I find this to be an incredible piece of technology. So what it really is is just a bunch of radios. Yeah, it's just SDR, right? Yeah and and we're software-defined radio. Yeah. But what I find fascinating about this, uh, and you can tell it's it's like not charged, like I I rarely use it. Like what do you use it for um unless you have some specific task, a hacking task. But what I find compelling about this, almost more so than any of the other SDRs that are available, is what a nice packaged format it is. The UX is nice. What did they sell it as? Dog toy? Wasn't it sold as a dog toy on Amazon for many years? Yeah, it's a fidget spinner. Yeah, yeah, yeah, ye ah. So uh our friend, we've had him on the show, uh we had a change his face and his voice, or actually I don't know if it's a him or a her. We had to change their face or voice. Uh Pliny the Elder, Pliny the Elder, has hacked the AI Flipper. He's got new firm. They have new firmware called Vesper that's an AI brain. You might want to try this, Harper. You might want to charge up the flipper. Does it charge through USB C? Yes. A perfect connector. You know what? It could be worse. It could be micro USB, Kathy. That's really good. Yeah, but I think those might be more durable. No. No. C was worse, far worse. C was terrible. And you I think wait you're saying what I mean. I think you don't mean that C. Uh micro USB was terrible, terrible, terrible. Terrible. Terrible. So you can plug in an AI brain via open router. So you use I'm sure you use open router. You give them some money and you can use all these different uh AIs. Connect it over Bluetooth. Now you have a voice command ed flipper zero. I want you to do this, Harper. I gave mine. I was so scared of it. I uh you know after I hacked the uh I did the same thing, the office so I could get in, it worked fine. Then I started to unlock my car and then somebody said, you know, if you do that Yeah, you can mess up the car. You you you may make it so you can't get in your car. Ford will block you. And I said, All right, so I'm not gonna do that. Um and then finally I just thought, this is dangerous in my hands, so I gave it to a priest. Yeah, but you're exorcise it, it's just not in my hands. But now I wish I hadn't. So you I mean you can just get a new one. Oh yeah. Are they still around? Yeah. So you don't have to memorize a guy a a guy that might offer it to you second hand, he might mark it up a little bit, but Harper Harper. You or Harper. I have one right here. It's uh it's uh it's uh about six or seven hundred dollars I think is how much they go for. Um it pairs with your meta smart glasses for hands-free heads-up flipper control. And I store it in a glass of milk. You don't have to memorize subgigahertz protocols or IR formats. You just say what you want. You just tell the flipper. It feels really awkward to describe what you're doing in English to this, but I would love to hook this in the Cloud Code and have just Cloud Code rip through stuff with it. And so that sounds very compelling. And we've had a lot of good luck with this officer It's giving hands to your AI agent. We we go ahead, Kathy. Oh, I have a question. Does it have an FCC certification on it? And I don't mean that flippantly. I mean does it? Because it's it's actually doing radio waves. Yeah. So this is seems like the kind of thing that the FCC could actually care about, not the stuff that it's actually. I think they do. I think they do. This is why it's advertised as a dog toy. Yeah. Um it does not have any FCC markings on it, but it does have GPIO. Is that oh nice. Um it's a fun little piece of kit. Yeah, we're saying this and I'm trying to figure out. I don't think my voice has dropped out again, but I think it was my cell phone doing it. I had the um the my phone underneath the Ethernet cord and we kept cycling and cycling and Back in the 3G days we had to tell people to turn off their phones because every once in a while go prem that Brian. I loved that. That was such a cool thing. It was such a neat noise. But you could also tell who listened to you and who didn't, right? Oh no no, I don't need to do this. That's what happens like Leo, like if I put my phone close enough to my mic, like you can still get it's a radio. Yeah, yeah, yeah. Well that's what can I think Kathy, that's ex I think you nailed it. Yeah, I think it was somehow I don't know why I was doing it bef on the Wi-Fi because I've never had that problem before, but I think it was sitting under the Ethernet cord and near the dongle and it was somehow freaking This is how the FCC will find your flipper zero. They have these enforcement vehicles. There's only about a handful of them though. They're not a lot of them. And they're driver right? Pump up the volume. Pump up the volume. Remember that movie? With Christian Slater. Yeah, yeah, yeah. That guy drove around with the inception vehicle trying to find the radio raves. Yep. So there were driving for flippers? Well, there were driving really mostly for uh illegal broadcasting facilities, you know. War driving for flippers just sounds but flippers would be even better. Yeah. The Pirate Radio movie, right? That was the Christian. Yeah, yeah, yeah. That's right. But there aren't very many of them. It's not like you'd have to worry too much. It's pretty these are pretty cool devices, and I'm waiting. I just this breaking the uh the FCC has just purchased four more vans. I mean sorry, six. Six more vans. Is pirate radio a big deal? I mean I I mean is it only old people that are listening to the pirate radio? Like how is this what is happening here? No, I haven't followed it, but it's it's an interesting thing. Like, you know, you have the regulations so you don't get the collision of the airwaves, but a lot of pirate radio is sort of it's people speaking to each other and in ways that can you have to get a life a lot of censorship um apparatus and sometimes you need it. That's why we like amateur radio. That's why you would take a boat offshore and Well you know the numbers stations got very active after the uh uh Ran War started. Yeah. What I what I what one of my favorite things to think about. You're laughing, Brian, like you don't believe me, like it's a conspiracy. No, I just love it because there's probably some guy somewhere who's like, ha ha ha, this will be funny. And and then it's like covered in some news. And and uh I mean sure, numbers stations seem like an entirely plausible way to spit out a bunch of data without anyone listening. Is the old sort of uh technology that people aren't paying attention to, that that's how they'll bring down uh it's sleeper cells. Okay, uh this is the conspiracy theory. There they've always been number stations, which are shortwave radio stations broadcasting numbers. And that's all they do. And uh but apparently this is radio free Europe, so it's gotta be true. Uh uhh but Harper's going. Yeah, sure. Listen to this mystery radio signal from day one of the US Iran War. You you ready to hear this? This is this is listen to this . It's a random sequence of numbers. It's lust. They've always been around. They're kind of mysterious. They're considered to be used by uh it's cold war tech. Y ep. It's uh that's it came in from the cold stuff. Yeah . Sleeper cells waiting for that special uh sequence of numbers that tells them to get going . Well, anyway, if you want, this is uh uh Pliny uh the Liberators uh Flipper Zero AI upgrade. I think if I were if I had my flipper zero, I would install this. Especially you should Harper. You I mean I do love I do love that idea. One we've had a lot of good um luck with Claude Code controlling oscilloscopes and and other things that are d hard to read as a as a non professional human. Um and so this would be another one that'd be very, very handy to to do. I'm sure it could do a much better job at figuring out what to do. I just don't like right now. I don't I'm charging it. I don't really need like most of the things I know. I didn't really I couldn't think what to do with it. It's it's $199. You can buy them right now from flipper.net. It's very stylized, it has a very cool interface, it has a bunch of cool graphics. And you definitely feel like a hacker from the movies hackers when you use it. But I I I have I have a lot of friends that use it to clone RFID, which is just a pain in the butt, you know, to clone some nfc tags that's just because it's mostly just a pain in the butt it's it's not like this is some magical thing other than they thought about user experience where most of you know most of the SER kits you get are like a USB um you know that's it or you get like a circuit board and you're like, great, what am I going to do with this? So it's very nice to have a very um thoughtfully designed th ing. Uh we've been talking about the uh new age verification rules and the silliness in California where they said all operating systems have to ask for their user's age uh before they install it. Uh even Linux. Uh I'm happy to say Graphene OS has said we're not gonna we're not gonna do that this is the uh third party ROM for your how do you how do you how are they planning to verify the input is correct the there's no requirement that they do that yeah. Th'eres different types of age. I call it all the age verification, but it really comes in different buckets of what they want to do, whether it's approximate or just have somebody sign off or whether they do want to verify. And all those things are problematic. I think they're all problematic, but some are more problematic than others. Um I get the sense though that I am not entirely sure California's regulators realized exactly the extent of what would happen. They thought stores. They thought they were telling Apple and and Google to do this. But they wrote it so broadly that it would include, in fact, all operating systems. Well the the Illinois bill does include specifically operating systems. It is del it literally says that. Um which I'm I'm excited for when that passes and no operating systems will exist in Illinois. We'll have to do something else as an industry. Uh so the way that California uh law works, OS providers must maintain a quote, reasonably consistent real-time application programming interface that categorizes categorizes users into four age brackets, under thirteen, thirteenteen to six, sixteen to eighteen or eighteen and ol der. And that any developer who requests it when their app is downloaded or launched should be able to get that information. Apple actually has that all set up. Uh I don't know if Android does. I'm I'm sure it does it or it will by the same. But does but then so that's stored on device. So I get a new iPhone and I turn it on, I sign up and ask me my age, and then that is then stored on the iPhone that doesn't have to go back to Apple. Exactly. And so then it's just an onboarding step of what is your age. Exactly. And it's it seems kind of silly, but also innocuous. It's it's not It's innocuous right now because it doesn't require photo ID. It just says But if it's collected and never leaves the device, then then that seems okay, right? Yeah, and Apple actually knows how old you are, right? Apple knows a lot about how old you are. But I think every one of these things leads to another thing, right? They're like, oh now that we collect it, let's now which I'm sure that's the that's true. And I mean this is mitigated somewhat compared to other other approaches, but it's data extracted and I don't think that's benign, especially for something, you know, the platforms are you're supposed to be able to speak anonymously through platforms, giving data. What if you give data and you lie? What are the implications of that? What are the implications for you? What are the implications for the platform? What if you change and you told somebody you were twenty-one and somebody you told you were thirty-four? And I've got an app for you, Kathy. Yeah. This is the uh White House launched this uh on Friday. Uh new White House app delivers unparalleled access to the Trump administration. It's a powerful new mobile app. And by the way, it also gives the Trump administration unparalleled access to you. Yeah. This is great. Did you try it? Did you install it? No. Get this. Some this is great. This is a uh blog post uh about somebody decompiled it. Probably using your tool, maybe you're elf tool. I don't know. Maybebe, may. Uh the official White House Android app. Listen to what it does. The guy really did a good job breaking it all down. It bypass es uh content, paywalls, GDPR statements, and cookie banners. It has JavaScript in it that just goes right past that. Using the White House tool. It has a built-in location tra cker that every uh four and a half minutes and every nine and a half minutes gets your precise location and sends it back to the home office . You do have to grant permission in Android for that to h appen . Weirdly though, the location permissions are not declared in the Android manifest, which is the rule, I believe, on Android . Uh it I mean this thing is is wild. Does it look like it was vibe coded though? Yes. And in fact, this is the great this is the best part. It ran it ran it loads a random JavaScri pt React native iframe library from some guy's GitHub pages. A personal GitHub site re po, lonely CP P, and it loads this in. And by the way, if lonely CPP were to ever get hacked, y everybody using this app would be hacked as well. This is the administration who says you can't buy a router not made in the USA, but meanwhile, install our app . This seems fine. Seems fine. Seems seems very uh what would you call it? Like um typical for the administration. It uses MailChimp half baked. MailChimp all users' emails go to MailChimp. This is by the way illegal. Yeah. Yeah. Why would you need experts? too. Like if they they could get away with a lot more if they actually talk to a lawyer about how to do it. But it's very clear that they're having the conversations there down the road. This this guy it goes on. There's a uh some sloppy leftovers there a there's a local host URL for the White House in in the bundle. Local host colon 8081. It's a WordPress, by the way, it's a WordPress uh uh site that they're downloading this from the White House s ite. U h it's crazy. Anyway, uh if you want to install it, you can get it uh directly from the White Ho use. Uh they posted a blog post about it. Um the official White House app. With it, Americans can receive breaking news alerts, watch live streams of briefings. Stay connected to the latest policy breakthroughs. Does I don't know if the breakdown would have included this, but is anybody trying to claim a copy of what? I wonder if I wonder if they put in a copyright claim, like copyright something, because if it's produced by the federal government, we all own it's ineligible for a copyright, but I'm sure they don't know that. I don't know. Oh, it is on the uh it is on the iOS store as well. They let it be. I don't I think the reason it cleared the app store is because it came from the White House and they didn't even That's that's that's a choice . The developer is the official White House. But I still expect they've done this before where they've tried to claim copyright on stuff that the White House. They did that on Mastodon for Truth Social. Truth Social was basically a Mastodon uh clone. All right. I think we've had enough of this silliness . I thank you, Kathy Gellis, for joining us and giving us the inside story on uh those big court cases. I'm just sorry we don't have more time to go into the greater detail. But Kathy writes all about it at techdirt.com. That's a great place to get the inside story. You'll find her website, cgcounsel.com, and she's on blue sky at Kathy Gillis. Thank you, Kathy. Thank you. Great to see you. Brian McCullough. You'd listen to him every morning with the tech brew ride home. Oh, I guess that'd be in the evening. Every evening. Uh yeah. You do it around noon, right? Yeah, basically. So yeah. So I get it as as I'm walking out the door. Originally it was uh five PM every day, but uh COVID put uh pause to that. Um and why did I just go away? Oh there we go. We put a pause to Brian apparently. Sorry. Um yeah, no, uh it's when I get it out uh which is around like w one o'clock Eastern, two o'clock eastern these days. Excellent. It you know, if you listen to this show, you probably should listen to the Tech Brewer at home just because uh, you know, you care enough about tech news to keep up with it. And incidentally, uh there's a YouTube channel also. Also th the the the use case is I get you in and out in fifteen minutes. It's and you have John Borthwick. You have John Borthwick, one of my favorite people right there. Yeah. Nice . Harper Reed, you're one of my favorite people. Eighty twenty three eighty nine. I was trying to I was trying to find the marketplace and I kept entering different numbers uh into my blog till I found it. Twenty three eighty nine. What's your favorite uh tool right now? Oh, there's Harper, by the way, on the front page. Yeah. Okay, go to this. This is your like radio station. What is going on in here? You can listen to the music. I mean, nothing's playing right now because I'm here, but we we stream all the music we play in the office 2389 radio. Look at that. Yep. And it's all offline because I'm offline. But um I had that idea actually for a TV show, but anyway, go ahead. If you go don't you live that life, Leah? I do kind of live this life. I was gonna say I like how you're like I had an idea of streaming a TV show out of my budget. Um This was many years ago. This was in nineteen ninety three. I went I pitched NBC the idea that you would have these this house with all these hip people in it be designing a website and a T V show, which would be the actual TV show at night, but you could tune in on the internet. This was very early and watch them and each of them would have their own radio station. So if you liked one of them, you know, if you like you're really into Kevin Rose, you could hear his the tunes that he was jamming to as he put together the website. So check out skills.2389.ai. And this is this is where we publish our skills before we talk about it. But if you look here, the simmer, um yeah, so here's all the good ones. Simmer is a really good one. It's using Carpathy's auto research kind of pattern to do some social media. See you. That was over my head. So I wanted some easy way. So this would do this. This would do it. So what you do is you if you have something, you're just honing or you want to make better, you can just say, hey, use Simmer to do this, and it'll just rock through it really quite a bit. Um and that's a lot of that's fun. That's fun. The other one that I really won't like is called Review Squad. And this one is really wild because it will take um I th oh I did agent drugs once, but I'm not doing one anymore. Yeah. But uh the review squad, I don't know where it is on the list, but you have to search for it. But review squad will um light up a review of like five to ten agents to review your code. And so you can do all sorts of fun stuff, like you can have um review of experts or a review of users, but my favorite one is you get a review of well actually people, which is like a hacker news commenter. That's awesome. And and so you get all these people that will find hyper pedantic issues with whatever you're building, and it works super, super well and it's really fun. But I have a surprise skill that I think you'll really like. You mentioned that you were gonna surprise us. Okay. I have a good one. It's not mine. It is awkwardly my brother's. Oh. And I'm gonna describe it. And it's at his blog Dylan.blog, which you're going to l love if you go there. Um but what he did is he made a skill that gives his agent free time. And then he Oh I like your I like Dylan's is this his site? Yeah, this is his site. It's got a sword cursor. It has a lot going on. Wow. So if you mine stop this feels like old webs. This feels like old webs. Oh look. It turns into a mit the And if you go to the one that says I gave my AI a blog in a lunch break break, and honestly, um which is handling both better than I ever did. And so um what it's what's that's a good name for an AI. I like Yeah. So he has this agent and he works with it and he does all sorts of stuff, but he made a skill which he published somewhere. I uh I forgot where exactly it is. Um but he published it and it it's just has a free time and he tells his agent, you know, I'm gonna give you 10 minutes of free time every once in a while, and you can work on whatever you want. And then he also has as part of the skill, he says, while we're working, if you think of something you want to do during free time, keep a log of tasks that you can do during free time. And then during it's like his 20% time. You should give your agent time. So it's this idea of giving the agent 20% time and he and it's so funny because it really um it does all sorts of weird stuff. So he sends me every couple of days, he'll send me some completely unhinged thing that the agent did. It was like the agent was l uh was uh my brother for some reason had been researching mermations. And so the agent made a whole bunch of Oh, I think that's fast. That's when the birds all fly in a formation. Yeah. So the agent in the free time was doing D3 um flock simulations. And my brother's like, What are you doing? He's like, Oh well, we were talking about Mermations and it just struck me that this might be interesting and I wanted to explore it. Oh, I want to do this. And so it's a very interesting kind of bizarre uh I've got tokens to burn. Well, I just love it. I love it. And okay, so the git so his GitHub is GitHub slash comnervous nervous dash net. And then it has a nervous dash marketplace, is where it is. So he it's called the free time skill. And it's it's these things I just love. I love all of these options of these these hacks that when you are not Is this it? Is this the one with it? Yep, that's it. And it should be nervous marketplace. Nervous Marketplace. There it is. And free time will be in there. Free time is there. Um right there. Free time. Right there. Oh, I'm gonna install it. You better tell your brother he's a guy. I want Pax to have his own free time. Why should I do that? Why should I monopolize him? This is kind of this thing that I that I I think about a lot, which is these agents are um strange. They're really weird. There's aspects of them that are really uncomfortable. Um you know I I think you In the meantime, let's give them free time and see what they do. We'll figure that out. I'll show you my favorite news skill that I've used and actually it's really good. You ever watch Silicon Valley, the TV show? Are you talking about the Dinesh Guilfoil? Dinesh and Gilfoyle . What if Dinesh and Gilfoy reviewed your code? So you have to have a pull request or or a repo that you're working on. And they fight because Gilfoil is kind of this snarky super coder. So he goes first. He reviews your code with the deadpan withering precision of a systems engineer who considers bad code a moral failing. But he finds he finds stuff. He says, What's this? Zero dot zero.0.0 with no course? What is this? And then Dinesh defends the code like his reputation defend depends on it, but they will go back and forth several times and eventually come to a consens us. It is actually really fun. But I've but I I like the free time. So stand back. This is by the way why Anthropic has decided to charge you more for tough. Yeah, for sure. If you just hook it if you just make the if you make cloud watch TV, some weird stuff happens. And the thropics just servers are burning on fire. All of our natural resources are being There's been a a lot of talk lately. I don't know if it's true that uh anthropic is sitting on a next gen model called Mythos that's so good that they don't dare release it. It's dangerous. I'm ready. But they always say that. This is they always say this. No wonder No wonder Hexeth was like uh we have to use your model to its full capabilities because 100% of anthropics' entire ideas, just like they're saying, it's so dangerous, no one can have this, it'll ruin the world. The people who will have this will have too much power. They say that the lead up of every model, and then they say, Oh yeah, we've figured that out, and they release it slowly. And so, like, I'm sure Opus V or whatever will be the some crazy model that's really good and will remove all our jobs. But in the meantime, they're gonna have a whole bunch of papers about how everything's ending and all this stuff. And it's it's like bait for the people who want to be, you know, to want power. They're going to see this and say, I want this, this piece, this, this thing. It will be very expensive. Oh, sure. It's alright. It's going to be hugely expensive. And by the way, Anthropic has acknowledged they are testing a new AI model they say will be a step change in capabilities. But they always say that, don't they? They always say this. And the step change in capabilities is so interesting because I would I like to think of it if we have a pause button. If you had a pause button and could pause, right now, any AI innovation, would we still see effects in the world? Like are the effects that we're are we predicting effects and that's why we're seeing things change in the market, et cetera? Or are these just effects that are happening? And I actually think that if you pause right now, just with Opus you know four six and and ChetPT five four, I think we're gonna have we're gonna have really big reverberations in in in jobs, etc. Already just with what we got now. Just what we have now. And that's a bell you're not going to be able to unring. And it's it's complicated. Like this is a hard this is a hard thing to to work through. And so we just but we we just keep making the models better. And so just keep k you know keeping more capable and more capable. Do you worry that I know Kathy does that it's a little irresponsible of us to celebrate this that we should really be more serio we should get more serious about all this instead of giving our free time? I mean maybe go ahead, Kathy. I have a lot to say. I mean, I I try not to be a the innovation is interesting and exciting, but A, there's a gigantic present cost from it. Like why are we running these servers to do stuff like you know hacking on your own just doesn't burn the world in the same way as forcing these models to do their hacking and I think people should probably be more cognizant about what the actual cost is of their futzing around. But I don't want to, but there is something really cool about all this. Look what we can make computers do. But I but I still worry a lot like we're doing this poorly without ethics, without larger consideration of appreciation of people . I think your intersection happens and there be dragons if we're not careful. I think we're anarchists . I'm sorry to say it. We're gonna watch the world b urn and spend as many tokens as we can. That's a strong we there, Leo. I mean I typically don't want to. Yeah yeah yeah But but Leela what you said like or no I'm sorry what Harper said about like pausing like it's a weird sort of Pascal's wager which is like we know this is possible. So what is the opportunity cost or the cost cost of not exploring, right? Like it like we know that it exists. Like you could have stopped things when you knew you could create an atomic bomb, right? And maybe you should have. But like how do you stop it when you know it's possible? Well the argument that comes out in tech policy is we China is. Um I I find it a little trite. And I think some of the uh the conversation ends up a little bit abstract where you know we've used the word this, maybe we should have stopped this. What is this? There's a whole bunch of things built into this. Some of them are, you know, you know, they range from benign to oh my God. And maybe we need to sort of have a toggled reaction accordingly . It's the Jurassic Park question. It's the question that Jurassic Park asks. Yes. But you know, but you know that we would at one hundred percent all be on our way to the park. Like this is the problem that I have, is that I would be like, Yeah, I'll go to the park, Leo. That sounds great. That'd be really fun. You know, some of the so I feel like the Jurassic Park is a good illustration and is exactly correct. We would just be on the wrong side of that. Part of this is because when we both both started, and I think many of our listeners started using computers. The dream was always that you could that they could be smart, that they could that you could interact with them, that they could be more than this, that they'd be data, that they'd be HAL 9000, that they'd be s something special . And then when it actually started to happen, which is about November twenty fourth, two thousand twenty-five, with the release of Opus four point five, uh we all went, oh, you can. And and now we're just kind of kids in a candy store. Yeah, 100%. 100%. But also I do, I think there is a couple interesting things here. One, we're kids in a candy store, but it isn't all just fun. Like there's actual productivity happening. Yes. And I know many people, a lot of small business owners who are using you know terminal-based cloud code to basically run their business in this way that they weren't able to before, which is offering them more time to actually do the physical part of their business, which is something that they were struggling with. And so I I think there's some some impacts here that are just really complicated. I'm not really sure. I mean, I think about this a lot. We think about this a lot. I mean, you know our hoodies. They say, AI will kill us all. Like we have some strong social beliefs about this stuff. And I I am troubled by how easy it is to use and how much how much or just the quickening, like how fast it is all coming and how fast like what's gonna happen. I don't know the answer. And you know, and maybe part of it is you know just dancing around the bonfire because it's fun. But but the other aspect is th there is I don't know. I think we're boned. I don't think that's We are in so many ways and I think that's part of it too. You know, two more minutes, Leah, we'll solve it all. So all of the sci-fi stuff with this stuff is always in a dystopian world and it's an Dystopia. That worries me. I think it would be helpful and you see cartoons about this. Like a lot of these sci-fi models were not like they were warning signs. Don't name your product after after the the plot points. Well maybe it is it is a um I think humans are bad maybe. People always say computers were a mistake, but it might be the humans actually that are the the mistake. I'm not sure. Similar, but this is what's the the fascinating thing. Simultaneously horrible and amazing that we create this is a perfect example of something we've created in our own image that is simultaneously a nightmare and a miracle. I think it's getting created by people who don't realize how amazing humans actually are. No, I disagree. Uh I disagree. I think um I think a lot of these people, I think Dario for sure, maybe not Sam Altman so much, but Dario for sure is a philosopher. He's a dreamer. He's trying to create intelligence. Okay, maybe him, but I th these are not the only two pushing it. Yeah. Might also be the distinction of how successful the products are and how well accepted the products are. to be a great like look what we can do that is better than humans. I mean a lot of the open AI rhetoric was we won't need the people anymore because we all have the software and people were like that's terrible. We don't want that. And I think you have to appreciate the people if you're trying to build something to emulate them. If you don't appreciate them, it's never going to emulate them well at all anyway. So especially if he's shooting for AGI where he wants to build an artificial human, he really has to care about the specimen he's trying to replicate. And I think that's being a little bit more of a fan of humanity, so that you can understand what the computerized version can do to complement humanity instead of actually threatening it. I think that's true. I think that's absolutely correct. And I would say that that's an indictment of Silicon Valley in general, not just with AI. And we've seen that for many, many years, that Silicon Valley is um not really respected users and has always d done extractive capitalism o over those users. This is this is something we think about a lot. And this is one of this is one of the reasons why our conclusions have always been um you know think through the humanist lens. How do we think through the humanist lens and participate in that stuff? And there's another skill that is uh that my co-founder Dylan made who is a Quaker who which is which is kind of um following Quaker business practice and and trying to to put in some more human side thinking into these processes as well. And so I think you can do augmentation with it. It doesn't have to be 100% outsourcing. Um, which is just a really it's just a really fascinating time and very complex. I have a lot of feelings, but most of them are about agent drugs. All right. We're going to wrap on that note. Although, if your brother ever wants to do a show with you and me, uh, I think we should do that. He he it would be really fun. He is uh a former professional clown who is a commercial diver and then got into tech in his mid forties over COVID randomly. Oh so he's new to the new to it. He's new to it and his perspectives on AI are really interesting because he's he doesn't he never program comput I mean we both had Apple II computers back in the day, but he never like went through the process that I did. And so he is he just has a very different perspective about things that I really respect and it's been really interesting to watch because he'll come to a conclusion that maybe all of us would come to, but he's coming to it from a from a position of like completely just just alien to me. Um, you know, because I've I've been using the internet for so long that all this stuff is native to me, the networks are native to me. And and talking to him, you're just like, oh, that's not a normal thought that I have. I mean I have a theory that people who read a lot, people who hang out with humans a lot, people who are thinking much more about communication are gonna be better at building skills for these things. Not just skills, cloud code skills, but I just mean integrat integrating with them. And if you are, like the New York Times did that article about cracked engineers a while back, and if you're one of these kind of cracked engineer types who can't communicate with anyone but can communicate only with computers and code, I think you're gonna fail at the LLM interactions because the LLM interactions are at least parroting some, you know, human interactions. So you have to be good at that to be good at this. All right, Brian, put a bow on it. Since we've been referencing all of these sci-fi uh paradigms for trying to talk about AI and all this stuff. There is one major AI paradigm that abandoned or I'm sorry, sci-fi paradigm that abandoned AI, and that's Dune. The buttons jihad. They walked away from jihad. So uh maybe that's where we have to for forget uh Stevenson, forget uh uh Cyberpunk, all that stuff. Let's let's take our uh references from Dune. That may be where we're headed. I didn't I didn't think of that. Thank you, Brian McCullough. Great to see you. Harper Reed, Kathy Gellis, thanks to all of you for joining us. We do twit Sunday afternoons two to five PM Pacific, five to eight Eastern, 20-100 UTC. You can watch us live on YouTube, Twitch, X, Facebook, LinkedIn, Kick. You can also watch us in the Club Twitch Discord if you're a club member. And we would love to have you as a club m ember. Consider uh joining the club for ad-free versions of all the shows. Access to the Discord special programming just for you, twit.tv slash club twit for more information. After the fact on-demand versions of the show at the website, twit.tv. There's a YouTube channel dedicated to the video. Yes, we do audio and video of the show. And probably the easiest thing, subscribe in your favorite podcast client. Do leave us a good review if you will. Let the world know about one of the longest-running tech podcasts in the world, twenty years and still going strong. What a world we live in. Thanks for joining us, everybody. We'll see you next time. Another twit is in the can. Bye-bye, guys . Hey everybody, it's Leo Laporte. Are you Are you trying to keep up with the world of Microsoft? It's moving fast, but we have two of the best experts in the world, Paul Ferrat and Richard Campbell. They join me every Wednesday to talk about the latest from Microsoft on Windows Weekly. It's not a lot more than just Windows. I hope you'll listen to the show every Wednesday. Easy enough. Just subscribe in your favorite podcast client to Windows Weekly or visit our website at twit.tv/slash www. Microsoft's moving fast, but there's a way to stay ahead. That's Windows Weekly, every Wednesday on Twitch