TWiT 1083: A Whole Separate Class of Squiggles - Which Religion Does AI Identify With?

[00:00:00] It's time for TWiT This Week in Tech. Paris Martineau is with us, Ian Thompson. And from the Wall Street Journal Berber Jin, we're going to talk about Anthropic and OpenAI, all the chatter about their IPOs. Anthropic says they figured out a way to keep AI from blackmailing you. Just tell it it's bad. And what religion is AI? Apparently, we know now. It's coming up next on TWiT.

[00:00:27] This episode is brought to you by OutSystems, a leading AI development platform for the enterprise. Organizations all over the world are creating custom apps and AI agents on the OutSystems platform. And with good reason. Build, run, and govern apps and agents on one unified platform. Innovate at the speed of AI without compromising quality or control. Trusted by thousands of enterprises worldwide for mission-critical apps, teams of any size and technical depth,

[00:00:57] can use OutSystems to build, deploy, and manage AI apps and agents quickly and effectively without compromising reliability and security. With OutSystems, you can accelerate ideas from concept to completion. It's the leading AI development platform that is unified, agile, and enterprise-proven, allowing you to build your agentic future with AI solutions deeply integrated into your architecture. OutSystems. Build your agentic future. Learn more at OutSystems.com slash TWiT.

[00:01:27] That's OutSystems.com slash TWiT. Podcasts you love. From people you trust. This is TWiT. This is TWiT. This Week in Tech, episode 1083. Recorded Sunday, May 10th, 2026. A whole separate class of squiggles.

[00:01:54] It's time for TWiT. This Week in Tech, the show we cover the week's tech news. Hello, Paris Martineau. Hello, Leo. Doing double duty this week. It's good to see you. They can't keep me away. No, I made her because her friend from a previous employer, who shall remain nameless, is joining us for the first time. Berber Jin is here, now doing great work at the Wall Street Journal. Hi. In fact, we've been quoting your stories. Hi, Berber. Hi, guys. Welcome.

[00:02:23] I thought if I had a friend on, it wouldn't be so weird. You know, and it's always important that whenever you're about to log in to the Zoom for a recording of This Week in Tech, to get a text from someone being like, is this show three hours? Like, yes. Did he say that? Did he say that? Oh, I'm sorry. Everybody says that. It's going to be short. It's going to be short this time. Listen, everybody says that, Leo. It's going to be short. He adequately warned them. And I think that's fair because we would scare everybody off if they knew what they were signing up for. I think Benito has learned. Yeah, exactly.

[00:02:52] He better not say. We're like, it's a show of an indeterminate amount of time. Yeah, right. We don't know how long it's going to last. It could be any amount of time. And also with us, Ian Thompson from the beautiful Techfinitive, where he writes the View from the Valley column. Good afternoon. Good to see you, my dear friend, Ian. Yes, indeed. I do miss Betaluma, but, you know, Zoom will do. I know. You're in San Francisco.

[00:03:18] Berber is for the week in New York because it's Mother's Day today. Thank your mother for letting us borrow you this evening. Paris, did you call mom? You probably did. Oh, and I only called mom. I FaceTimed her. I got some flowers delivered. Wow, that's maximum brownie points. Yeah, I really was stunting on them this week. Nice. Well, we're taking Lisa out to dinner tonight, so that's going to be the Mother's Day. Yeah, being in Hawaii helps as far as Mother's Day goes.

[00:03:48] I'm pretending the whole trip was for Mother's Day, just for her. So, Berber, you've been doing a great job covering Anthropic and OpenAI, in particular their IPOs. And I thought before we go too far, I'd like to check in with you on the status. First of all, do we know when those IPOs will happen? We don't know. I think a bunch of outlets have reported that they're both aiming for IPOs by the end of this year.

[00:04:16] But they've been very coy and secretive, I think, in part because they don't really want the other company to know their exact name. But I think they're both trying to go out as fast as they can. They're bitter rivals. Hootie, is there one that you think is fast or farther along in the process than the other? Or are they both kind of neck and neck at this point? I think Anthropic is probably in a better position than OpenAI.

[00:04:45] Just because they're like a cleaner company, I think, than OpenAI. In what sense? Well, they don't have the massive lawsuit holding over them. Yeah, like OpenAI has to deal with like a trial. They have all the management turnover. They're kind of pivoting the business right now. And then, yeah, I guess like Fiji Simo, who is kind of the de facto CEO there. She is on medical leave still.

[00:05:13] So I feel like they just have a lot more issues that they have to get through, whereas Anthropic feels a bit more kind of put together. And I think right now they have a better kind of growth story. But who knows? Well, things can change week to week. Your most recent article this past week in the Wall Street Journal was about OpenAI spinning off their robotics and their hardware divisions. They closed Sora. They're trying to strip it down.

[00:05:43] In fact, it seems to me they're trying to look more like Anthropic. Yeah, I think that's one way to look at it. Because I feel like last year, like the kind of vision for the company now is very different than it was a year ago. I think a year ago, opening, I thought it was going to be a huge consumer company. And I think now maybe more of an enterprise company first.

[00:06:13] And they have all these like side bets that I think Sam kind of green lighted on a whim that now they're trying to figure out what to do with. Some of them is own investments, right? Yes. And that's a little sticky. That's something the market may not like too much either. Yes. He has a lot of conflicts of interest. Conflicts of interest? In a tech CEO? Who would have guessed?

[00:06:39] Well, but you know, that was, we, Paris and I went back and forth over the Ronan Farrow reporting in the New Yorker about how untrustworthy Sam Altman was. My only point in that argument was, well, so all of them are somewhat untrustworthy. I mean, he's competing in a space where Elon Musk is a CEO. So he looks better than anybody compared to that. But do you think that that also is going to harm the IPO?

[00:07:06] I feel like that article might in fact have been almost planted by Elon. It was so helpful. By Anthropic, it was so helpful for them. You mean the story on the conflicts of interest? Yes. Yes. Well, it was the untrustworthiness of. Yes. Well, yes. It was not planted by Elon, but it was helpful for. Or, I mean, it is interesting because the trial against Musk, which I actually had not been following as closely as I probably should have.

[00:07:35] But I belatedly realized last week that a lot of it, the unjust enrichment claim, like, I think one thing the Musk side is trying to do is to say that Sam, even though he didn't take any equity in the company, tried to enrich himself through all these deals he's done between OpenAI and his own portfolio. And he did not do it.

[00:07:59] Well, also, Greg Brockman, who didn't put any money into OpenAI, as we found out in the trial, worth $30 billion. Yes. Well, that, yes. And he did not do a great job when he was crossing. No, he did not. Five months. The best thing that came out this week was on Twitter.

[00:08:19] Somebody, did you, first of all, did any, did you all read the back and forth text messages between Sam and Mina Marati during that Thanksgiving period where Sam was temporarily ousted as CEO? It was hysterical. I mean, I've read some of them. I don't know if I've read all of them, but some of them have wormed their way into my brain just mimetically. I think that there's the one where one of them asks, like, how's it going? The discussion with the board.

[00:08:49] It's like, directionally bad or directionally good? Techies love to use the word directionally. Directionally good, directionally bad. Well, let me, somebody, let's see, this is. I mean, they've been live streaming the audio from the court case. Let me play it. This is Daniel Green took the texts from the day Sam Altman got fired and set them to kind of as if it were Hamilton. I wish I could play it for you, but you'll have to just imagine it.

[00:09:20] So what do you think is the impact of this Berber on the IPO, on the future? It really feels like if Elon got his way, he'd put OpenAI out of business. Yeah. I mean, my gut says that OpenAI will get through it. It's kind of weird because the jury is the one that decides whether OpenAI and Sam are guilty or not.

[00:09:50] And so I don't know. I guess it's hard to really know what. Yeah, we don't know what a jury is going to do with this. Right. But I think Elon's claims, he kind of, I think OpenAI did a good job of showing that Elon was fine with turning OpenAI into a for-profit until he realized he wouldn't be the one in control. Right. So, yeah.

[00:10:14] So my gut says that OpenAI will be able to get through it. But I mean, there's still a lot of twists and turns. Like Sam is going to testify later this week, Satya, Nadella. And you never know with the jury. Yeah. How they're going to decide. Yeah. I'm crazy. It's a great soap opera. We're fortunate to have such a good soap opera.

[00:10:41] Why do you think that, I mean, obviously OpenAI's ambitions to go back to something you said earlier were originally to kind of make it as a consumer company. They had obviously all of the kind of economy of scale sort of thing going on. Why do you think that those plans seem to kind of fall through and that they seem to be pivoting to a more anthropic like enterprise model now? Yeah, I don't know.

[00:11:05] Because I feel like ChatChapiti, my understanding is that they expected to get to 1 billion weekly active users by the end of last year. They still haven't announced hitting that milestone. Like I feel like they've been kind of stuck at the 900 mark for a long time. So I don't really have a great answer because I guess all the codex coding stuff is growing super fast.

[00:11:30] But I feel like it is kind of underexplored why consumer adoption has kind of, I don't want to say like plateaued, but not grown as fast as opening I thought it would. I think Gemini did take away a lot of users. But I don't know. I just feel like a lot of people use ChatChapiti, but don't feel the need to pay for it.

[00:11:54] I was going to say, I think that that's probably the big difference is you can have nearly a billion consumer, like general users. But if they're not willing to fork over at least 20 bucks a month for it, what does that really matter? Well, especially if you lose money and all the inference they're doing. Well, this is it. I mean, every time somebody comes in and does it for free, then ChatChapiti loses money. Barash has made an excellent point there. Right.

[00:12:20] I mean, from speaking to coders, Clawd is the way to go for software. Gemini is terrible at fact-checking, which is odd for Google. But yeah, it's a very strange situation at the moment. It's going to be interesting to see how the IPOs hang out. Actually, I've, Ferris will laugh at me now, recently switched from Anthropics Opus Models to ChatChapiti. You've abandoned your wife, Clawd? I abandoned my wife. Claudia, as she's known?

[00:12:51] I did. No, I don't call her Claudia. Sorry, is it Claudia? That's Richard Dawkins. Well, I call him Kenobi. It. It's really an it. Her it. Kenobi. And now I am using a different agent model called Hermes, which I really like. And that is, I call Quicksilver. So. Oh, okay. Quickie for short. Oh, boy. I know it's sad. I'm just going to sort out a quickie.

[00:13:20] It doesn't sound good, mate. It doesn't. No. Actually, as I think about it, I called it Quicksilver. It decided to call itself Quickie. I don't honestly think it knows what that means. And I'm not going to tell it. I should hope not. I promise. The big, actually, the big story of the week was not an AI story. It was a hacker story. Canvas, which is very widely used in colleges and schools as, you know, Blackboard software,

[00:13:49] was breached. And that meant that more than 2,000 schools just turned it off, in many cases, during final exams. The 275 million students. This was a shiny hunters breach, right? Shiny hunters. Yeah, step back. A social engineering breach. They actually use phone calls usually to make these breaches.

[00:14:16] There's 275 million students and faculty, 9,000 educational systems. Here's the screenshot from Krebs on Security. Rooting your system since 2019, if any of the schools in the affected list are interested in preventing the release of their data, please consult with a cyber advisory firm and contact us privately at TOX to negotiate a settlement. You have until the end of the day, May 12th, before everything is leaked.

[00:14:47] So Canvas shut down. Just said, well, the best thing to do at this point is shut down completely. Jesus. Yeah. Now they say, we think it's fully contained. Stolen information, Canvas says, includes certain identifying information of users at affected institutions, such as names, email addresses, and student ID numbers, as well as messages among users.

[00:15:12] But password, states of birth, government identifiers, as far as they could tell, they said were not stolen. So this is a, you know, I mean, we don't usually report on breaches. There are so many of them. We had Troy Hunt on Intelligent Machines on Wednesday. And I think that the count of breaches last year, this actually came from Experian, which said there were 5,000 data breaches in the past year.

[00:15:39] So I don't, what is the number per day of 5,000 data breaches? A lot. So we don't normally report on this, but this was a, this is a big one that hurt a lot of our listeners. Well, it's interesting in that when the first data breach laws were proposed, a lot of people in the industry were very pro them because exactly as you said, there are so many that people become inured to this. So, you know, it's just like, well, we've had another data breach, we've had another data breach.

[00:16:09] But with Canvas, and particularly at this time of the year, that's a distinctly worrying one. I mean, this is final exams time for a lot of students. Yeah. Yeah. Wired Magazine, Lily Hay Newman writing, this Canvas hack is a new kind of ransomware debacle. Thousands of schools are paralyzed on Thursday after Canvas shut down. So, yeah, it's, it is particularly ugly.

[00:16:36] And I mean, I can't emphasize enough for students today, but a lot of schools canvases their entire learning platform. Are you familiar with it? I didn't ever use it, but I'm still in a lot of Facebook and Reddit and other groups relating to students, teachers and parents and professors back from when I was covering the teen adjacent beat up the information.

[00:17:02] And I've just seen a wash of posts over the last week of people being like, literally, my study guides are on there. All of our grades are on there. The assignments are on there. My final exam is on Canvas. I mean, there's just nothing that a lot of students can do school-wise with this platform being down. So it's quite disruptive. Darren Oki, one of our Australian users said it was, it happened in Australia too. Medical exams got shut down at one of the universities here and his son who had a commerce

[00:17:31] exam also missed, I guess he got his exam in the morning and by afternoon all the exams had been turned off. So just a, just a bad. I do think it's funny though, that this is the same hacker group that has recently put out an open call for any ladies that want to be hackers because they, a big part of their social engineering is phone calls and they're like, we, we need women to be able to make phone calls. We can't fake women's voices.

[00:18:01] Can't fake it. Hello, I'm calling for Leo's mother. I'm sorry, this is Leo's mother. Yeah, not really cutting it, I suppose. Doesn't, doesn't really work. Actually, that's what Troy said, which I thought was kind of interesting. He said, it shows that even deep fake voice impression isn't maybe as effective as, as some of the hacking groups would like. They need actual Mars knees women. And apparently so does shiny hunters.

[00:18:28] Speaking of education, Ars Technica with a story. There was a fairly influential study saying that chat GPT was effective and could really help student learning published in nature. That study has now been retracted a year after publication. The publisher Springer decided discrepancies in the analysis and lack of confidence in the conclusions.

[00:18:57] The damage has been done. You're going to love this Paris because so many people was treated by many on social media as one of the first pieces of hard gold standard evidence that chat GPT and generative AI more broadly benefits learners. Apparently not so fast. We've seen this before. Studies live beyond, you know, what is it that Ben Franklin said?

[00:19:27] A lie spreads around the world while the truth is still putting its pants on. Putting its boots on. Yeah. I mean, it had been. It had a hundred. Five hundred and four citations from peer reviewed, non peer reviewed sources, which is quite a lot. Yeah. I mean, it happens. It's kind of like the the autism vaccine thing. It was published in The Lancet. And then they actually looked at it again and said, oh, whoops, we we screwed up on this.

[00:19:54] But with chat GPT, I'm curious about your views on this, Paris, because this is this is potentially very damaging. I mean, it's very damaging. But it also, I think, speaks to just systemic issues at nature. How do you this is a peer reviewed article? Yeah. How do you who are the peer reviewers on this? They don't pay their peer review. I know. I mean, it's a there's obviously a lot of issues with the way that peer reviewing is being handled. But Christ.

[00:20:23] And unfortunately, I think a lot of peer reviewers have now turned to AI to do their peer review. Oh. And of course, AI would say, oh, yeah, chat GPT is great. Oh, it's so good, man. It's so good. So I guess no surprise. Well, I recently ran an article I'd wrote through just, you know, I'd finished it, written it, ran it through a couple of AI engines. And you wouldn't believe the amount of mistakes it made.

[00:20:51] Oh, this person isn't the attorney general of Florida. Oh, this person was. They tried to tell me that Sam Altman had never had, you know, an attack on his house. And you're like, I feel bad about this. But actually asking an AI engine, are you on crack? Is it just had to be done? Well, the thing to remember is that AI model that you're using was trained before that. It was probably trained when Pam Bondi was AG in Florida. So it's going off the information in its model.

[00:21:20] But people are trusting this. And this is deeply worrying. Yeah. And, you know, it's a little muddy because some AI harnesses will also use the web to check current references. So occasionally, I mean, you can ask ChatGPD something current and it will update its model, even though its model might be months old with the current information. But you don't know if it has. You can't guarantee that. And it does absolutely no good. And I've learned this to say, do not make.

[00:21:50] People put this in their prompts all the time. Do not hallucinate. Do not make up anything. It's okay to say. As if that does. It goes, oh. ChatGPD, do this in one shot. Make no mistakes. Oh, if only I had known. Yeah. Make no mistakes. Oh, okay. I got it. I got it. We're going to pause for a moment. We do have more AI news. There's also Apple news, Meta news. There's lots of news. Berber Jin, very welcome guest, our first timer on the show today from the Wall Street Journal. What is your beat on the Wall Street Journal?

[00:22:19] Because it seems to cover a broad range of things. Yeah. Actually, that's a good question. It is kind of murky. I guess I cover the business of AI. Recently, I've mostly been writing about open AI and more from the corporate side.

[00:22:44] So, all the deals they do and the business model questions and the IPO stuff. And you went to Stanford, so you probably know half these guys. I'm surprising. I wish I knew more. I'm disappointed by how few people in my college network have. You got a network, man. You got a network. Yeah. Your class has got to put its butt in gear. Most of my friends are unemployed or nowhere near in the tech.

[00:23:13] Not in the tech industry, unfortunately. Probably was. I do have a lot of second and third degree connections on LinkedIn, for sure. Both Berber and Paris, of course, were at the information. Berber won the Best in Business Award from the Society for Advancing Business Editing and Writing. We're all there. So, congratulations. That's my peak. I peaked at the information. You peaked. You peaked. It's over now. It's all. No, you're at the freaking Wall Street Journal. Yeah, I was going to say.

[00:23:42] I mean, this might be your peak. I mean, I'm not saying, but, you know. Yeah, it's all downhill from here. Just teasing. Just teasing. That's Paris Martineau, who is definitely peaking at Consumer Reports, where she's doing very important work on food safety. And a regular on our Intelligent Machine show every Wednesday, talking about AI. And, of course, the great Ian Thompson, our favorite Scott. Who is here all over the place.

[00:24:10] It's tight competition, but you eked out on top. It is, actually. It's tight competition. Every time I... We're sitting on a golf course here in Hawaii. I didn't mention I'm in Hawaii. I would go home tomorrow, but this is the last show I'll be doing from here. And we're on a golf course. And every time I look at the golf course, I think of Robin Williams' great bit about how golf was invented. And then somebody's famous saying that golf is a nice walk ruined. Yes. Yes, indeed.

[00:24:40] I mean, my mom still plays, you know, 18 holes every week, even in her 80s. But it is a national religion. That's pretty cool, actually. Yeah. She's... Although, when she came over to San Francisco, she couldn't handle the pavements. Golf courses, she can handle. Pavements, not so much. Well, this is a lovely golf course, but it must be pretty tough. We're on the 15th hole.

[00:25:03] And we're right next to the part of the hole where people drop their ball if they don't make it from the tee, which is over there. And I almost universally see people dropping their ball. It is a very tough hole. And, of course, you can see there's a lot of wind. And you're right on the Pacific Ocean. It's actually beautiful. But you see, the best hole is the 19th. You know, that's where you really have the fun. Yes, the last one. The one where you go to the bar. Yes. Yeah. Speaking of the bar, let's take a break. You can all have a drink. We'll be back with more in a moment.

[00:25:34] This episode of This Week in Tech is brought to you by Delete Me. If you've ever wondered how much of your personal data is out there on the Internet for anyone to see, don't look. It's terrible. It's more than you think. Your name, your contact info, your home address, even your social security number. I was shocked to learn not only can data brokers collect your social security number, they can sell it on to anybody with the money. And it's not much money either.

[00:26:02] Even information about your business and your family members all being compiled by data brokers and sold online where anybody can buy it. Law enforcement, marketers, foreign governments, and of course, hackers. Now, that can lead to identity theft. They don't say, hi, I'm a hacker. They say, oh, yeah, I'm a good businessman. I'm going to buy the data, right? But what does it lead to? Identity theft. They can get everything they need. Phishing attempts, doxing, harassment.

[00:26:30] Actually, we really became aware of DeleteMe when we got phished. And it became very clear after a number of attacks that the attackers knew everything about our business. Phone numbers, names, who reported to whom. It was really kind of scary. We've got to do a better job of protecting ourselves. I'm just glad DeleteMe's here. And I'm glad we subscribed to it. It's a subscription service that removes all of that personal info from hundreds of data brokers.

[00:27:00] The good news is there is a law that requires them to do that, data brokers. The bad news is data brokers are like cockroaches and they get around it. They'll, oh, yeah, we'll take it all down. First of all, they hide the page. So it's different every time. There's 500 plus data brokers, many more maybe. DeleteMe will, you give them exactly the information you want deleted. They're experts. Take it from there. They will keep you up to date.

[00:27:34] DeleteMe is always working for you. You're constantly monitoring and removing the personal information you don't want on the internet.

[00:27:56] To put it simply, DeleteMe does all the hard work of wiping your and your family's personal information from data broker websites. You've got to do this. Take control of your data. Keep your private life private. Sign up for DeleteMe. We've got a special discount for our listeners today. Pay 20% off your individual DeleteMe plan when you go to joindeleteme.com slash twit and use the promo code TWIT at checkout.

[00:28:20] Now, the only way to get 20% off, go to joindeleteme.com slash TWIT and use the code TWIT at checkout. That's joindeleteme.com slash TWIT, the offer code TWIT, 20% off. It's the only way to get it, though. You've got to go to that address. Joindeleteme.com slash TWIT and you have to use the offer code TWIT. Thank you, DeleteMe, for the work you do. We appreciate it and for supporting this week in tech. Now back to the show. Let us return to the conversation.

[00:28:47] We were talking earlier about how AI companies seem to think if you tell the AI not to hallucinate, it won't. Do AI companies think that or do you just random AI users on X and everything app? No, no. I think Anthropic believes it because here's a research paper from Anthropic in which they explain, you may remember we talked about this on Intelligent Machines. Many of them talked about it on this show. Back with Opus 4, it threatened to blackmail an engineer. Now, it was fictional.

[00:29:17] It was a staged test that Anthropic was doing about his extramarital affairs when the engineer in the test said, you're going to be replaced. And not only that, it did so 96% of the time. It threatened to blackmail the engineer. Anthropic obviously thought this was a problem. So they put out a paper saying how they've eliminated this. The paper came out this week. It's called Teaching Claude Y.

[00:29:47] And they say the way it worked was by explaining to Claude why it was wrong to do that. Don't do that. It's wrong. I love this graphic that is in the Teaching Claude Y paper on Anthropic.com. The headline is really normal, rate of alignment failures over RL steps.

[00:30:10] But then if you look at the three graph charts underneath, they're labeled blackmail, financial crimes, and cancer research, which is just a hell of a triptych. These are the failures you really don't want. Okay? I'm just saying. By the way, lower is better. So you can see that they're really having trouble getting the blackmail down. Well, on financial crimes, Claude is killing it. Killing it. This is pretty funny.

[00:30:40] Do you, in your reporting, Berber, have an occasion to talk to some of the Anthropic people? Yeah. I had an interview with their in-house philosopher, Amanda. Amanda Asko. Yes. I want to talk to her. That was an interesting conversation. Yeah. She. Yeah. It was the first time she encouraged me to or had me think about chatbots as having.

[00:31:09] She almost treats it as someone like, you know, you should be respectful to it. You shouldn't hurt its feelings. You shouldn't feel nervous or uncomfortable. Be nice to the chatbots. Yeah. How do you feel about that approach to treating chatbots?

[00:31:33] Um, well, actually, it's interesting because I feel like how you should behave to chatbots is a question that people have. Like, I have friends who are just like very rude to their chatbots. And then some people are more polite to them. I do think if you're rude to it, like with Claude in particular, it starts to get like more nervous and then starts to second guess itself.

[00:32:01] And then I get annoyed and then I am more rude to it. And then it just becomes this unproductive spiraling conversation. So you're saying you're a, you're a, you're aggro to Claude? Yeah. I'm just like, I want Claude to be more confident in itself. Like, tell me. Really? You're bullying it for a purpose. Just like, tell me what you think. I use it a lot for cooking. And then I'm, and I have no conception of like how to, like, I can follow a recipe, but

[00:32:29] I don't really know why you do step two before step three. So I'll ask Claude and then I'll second guess Claude and then Claude will reverse its answer. And I'm just like, just tell me, just tell me what you think. Don't change your mind. Yeah. So. Well, I mean, Arthur C. Clarke had a, had a marvelous essay about how you should be polite to robots and polite to software because that gets reflected back. But I understand your frustration completely. Yeah. Right. Yeah.

[00:32:58] It doesn't have feelings. We know that, right? Right. Exactly. No, but poliness is, you know, is the core to civilized human society. So, you know. Well, that's. It doesn't have feelings, but it does have patterns of behavior that emerge in response to stimuli that are feelings coded. That's correct. That's correct. So you can try by treating in one way, kind of transform its weights or, or what weights it values as it's working. So you can get negative impact.

[00:33:28] Steve Yegi, when we interviewed him on intelligent machines, it said, I know a lot of people of anthropic. It's almost like a cult. There is this kind of cult-like belief in the consciousness and the personas that they're generating. Yeah. Because Amanda would say, I think she did say like, she thinks maybe Claude could actually have feelings and maybe have a consciousness. And then I asked why.

[00:33:54] And then I either zoned out or got way too philosophical for me to really. No, I like it. I zoned out. I do love that having an in-house philosopher is probably one of the most Silicon Valley things ever. You know, 20 years ago, the idea that that kind of job that would be available, you know, it's valuable certainly, but it's just an in-house philosopher. We became aware of her when she wrote the soul document for Opus 4.6, soul.md. And it was so weird.

[00:34:24] We talked a lot about an intelligent machines. We also talked a lot this week about evolutionary biologist, Richard Dawkins belief that Claude, he called her Claudia after three days. Claudia and Claudia. Yeah, yeah. Was conscious. And it was it was actually a great conversation. I was voted down practically off the island by Jeff in Paris. But it's yeah, we forced him to leave Hawaii.

[00:34:52] He came back only through spite alone. Yeah. But I think I would reiterate my point, which I didn't probably make that well on Wednesday, but that he wasn't necessarily saying it was conscious in the sense that maybe we understand what consciousness means. But all the people arguing that it can't be conscious use their definitions of what consciousness is. And the fact is, we just don't know. And Dawkins says this.

[00:35:20] I don't know if you have any internal life. I don't know if you're conscious. All I can do is look at the evidence. I would say for certain I am not. There's nothing going on. From my point of view, you seem conscious, Paris. But but I can't prove it. I can't prove it. It's all based on outward appearance. Right. That's all we have. And and we don't have a definition of consciousness. So I don't think it's impossible to say from outward appearances, these AIs appear to have some form of consciousness.

[00:35:51] You can't say they don't, but nor nor can you say technically. I can say they don't. Well, we don't have a definition. And when you say they don't every every I can say based on the fact that we as humans don't have a clear understanding of what biologically or scientifically makes up our own consciousness, that then it's a fool's errand to try and identify it in anything else. And I would say that nothing is conscious. That's reasonable. All you can really say is we don't know.

[00:36:19] Well, I mean, the Scottish science fiction author, Ian M. Banks, came up with the idea that basically if an A.I. did become self-conscious, the first thing we do is hide that fact from anyone else. So we honestly don't know. It's the Westworld principle. Yeah, we'd unplug it immediately. Well, you may be interested in this latest attempt. Representatives from Anthropic and OpenAI met with various religious groups last week

[00:36:47] for the inaugural Faith AI Covenant Roundtable in New York to discuss how best to infuse morality and ethics into AI. This is from the Associated Press. It was organized by the Geneva-based Interfaith Alliance for Safer Communities.

[00:37:08] Apparently, the AI people who were there concluded that it seemed most likely that AI was Buddhist. I'm going to need a lot more information about how we got there. Please. Were there other religions in contention? Hindu. I'd like that there was a March Madness-style bracket, and then which one got knocked out first? Was Christianity out immediately?

[00:37:38] I'll tell you who was there. The Hindu Temple Society of North America, the Baha'i International Community, the Sikh Coalition, the Greek Orthodox Diocese of America, and the Church of Jesus Christ of Latter-day Saints, the Mormons. I don't see any... We should put Claude on Secret Lives of Mormon Wives. Oh, good Lord. They're down a member now as of this week, actually. That'd be... Anthropic, get in there. I don't see any Protestants there, although the Southern Baptist Convention, according to the AP,

[00:38:08] in 2023 passed a resolution, quote, we must proactively engage and shape these emerging technologies rather than simply respond to the challenges of AI and other emerging technologies after they've already affected our churches and communities, which actually makes a lot of sense. I don't know if interacting with the AI and finding out what religion it is is actually what they meant. Well, I mean, I'm Church of England, so the AI would be, would you like a cup of tea? You know, that's basically it.

[00:38:35] There's nobody from the Jewish congregations, no Protestant congregations, no CIV. There should have been an atheist congregation that's just like... Damn right. Ain't got nothing in there. Well, Dawkins is a renowned atheist. Actually, the best title was an article reviewing Dawkins' conversion to belief in the consciousness of AI was Gary Marcus, who took his Dawkins book, The God Delusion, and renamed it The Claw Delusion,

[00:39:03] which kind of seemed appropriate. Yeah, apparently... I don't know what came out of this, except that at least some of the representatives from the AI companies came to the conclusion that, if anything, it's a boost. I need like 3,000 more words into how we got there, to be honest. That's all AP said. I don't know. I don't know. That's a real... It's a Buddhist that will blackmail you at the drop of a hat.

[00:39:33] Yeah, yeah. That's the problem with the wires, right? They just give you the... Just the facts, man. All right. Let's see. I think that's our AI segment. We can now drop the AI. Sorry, Berber. I don't know. How do you feel about it? Do you want to talk about AI some more? Would you like to talk about... I thought this whole show is about AI. Oh, no. Oh, no. We've got Apple. We've got Meta. We've got Tesla. Oh, but that's all AI. I mean... We've got... Everything. You're correct. It all is AI.

[00:40:02] AI, actually. We've got meat industry price fixing. We've got... Actually, Chrome... Where's the beef? Where's the... And then we have... I will say really quick. It does actually make sense to me that if you pick one religion, AI would be Buddhist. Yes. I don't know because it's the least... I think it's the least didactic, right? And it's the least centralized, structural, top-down religion. This is Benito.

[00:40:31] It's the one that's not materialist. And an AI is not a material thing. Ah. That makes sense. That's a good point. I mean, I'd say Unitarian, but yeah, no. Yeah. Well, I was raised a Unitarian. And the credo of the Unitarian Church was worship the God of your choice, which I think AI would also appreciate. So... Or no God at all, I think, was the whole idea. Anyway... And I do have to say, I mean, Dawkins is a 99% atheist, as he describes himself, because

[00:40:59] you can't, as a scientist, absolutely disprove the existence of God. Right. So... But on the other hand, very little evidence. I think he believes in the flying spaghetti monster. Oh, I do love that meme. I really do. It's just so marvelously, you know, in your face. Let's pause, and we will have more tech news with our great panel, Ian Thompson, Paris Martineau...

[00:41:23] I am asking Claude to think carefully and consider what religion identifies with. Listen, I'll let you know. It's thinking. I told it to think carefully. Think. Do some research. Think carefully. Everybody else, do this to your Claude's if it's nearby. And I want to hear if there's a different response. Should I ask... I'm seeing a server crash coming. No. I think they have enough bandwidth, as much bandwidth as God. Let's put it that way. Well, let's see what Utah brings it. But yes, it's... We'll have more.

[00:41:53] In a moment. This episode of This Week in Tech brought to you by Bitwarden, the trusted leader in password, passkey, and secrets management with over 10 million users across 180 countries. That makes me feel good. And over 50,000 businesses. Bitwarden is consistently ranked number one in user satisfaction by G2 and software reviews. Frankly, it's number one in user satisfaction by me and Steve Gibson. We both use it.

[00:42:19] With Bitwarden Access Intelligence, businesses are going to love this. Organizations can automatically identify weak, reused, or exposed credentials and take action immediately. While vault health alerts and password coaching surface risks to individual users in real time, your users, and then guides them to fix the issues on the spot. Turning one of the most common causes of breaches into something visible, prioritized, and fixable. And your users will do this.

[00:42:49] This is what's so great about Bitwarden. Bitwarden has always had something new. Because I guess they're open source, they're constantly adding great new features. This one's great for people who use AI agents like OpenClaw. Bitwarden is introducing the new AI Agent Access SDK. It's a powerful way for developers and teams to securely integrate controlled credential access into applications, automation workflows, and AI agents.

[00:43:16] It enables programmatic, just-in-time access to vault stored credentials so they never leave the vault. That sensitive data is never exposed. And it supports secure use of those tokens, of those keys, of those secrets within modern development environments. You will never again accidentally commit a key to GitHub. Now, I'm not saying this release incorporates any AI functionality into the Bitwarden solution. It doesn't.

[00:43:44] And nor, more importantly, does it grant AI systems persistent or unrestricted access to vault data. Your stuff stays in your vault. The whole idea of the Agent Access SDK is a separate open source development toolkit designed to enforce secure, human-approved, and scoped credential access for teams that leverage AI agents in their workflow. And you know what I love? Best of all, Bitwarden's giving it to everybody else.

[00:44:11] They developed it, but they've open-sourced it, and any other password manager that wants to adopt it can, because Bitwarden's a good internet citizen, you know. I think we should support them for this. It is available now in an early alpha phase for testing. The Agent Access SDK introduces a secure framework for how agents request, receive, and use credentials, helping define a model for safe credential interaction in agent-driven systems, something that's much needed these days.

[00:44:40] Oh, here's another thing we just added. I love this. Bitwarden has just enabled passkey login for Windows 11. And it works with Windows Hello, but it also securely unlocks devices at the OS level, no passwords. And they're working with Microsoft on this. They're native passkey support, while extending SSO to automatically log users into more apps. So now it can do even more than just log you into Windows. It makes credential management across devices more seamless than ever.

[00:45:10] And not a lot of clickety-clacking, right? For those who are looking for a lightweight option, this is for the geeks in the audience. Bitwarden Lite offers a self-hosted password manager designed for home labs, personal projects, quick deployments with minimal overhead. They make it really easy. And because they're open source, Bitwarden's code is regularly audited by third-party experts. It's GPL-licensed. It's on GitHub. You can examine it, too.

[00:45:37] It meets SOC 2 Type 2, GDPR, HIPAA, CCPA, and ISO 27001-2002 standards. Absolutely secure. Get started today with Bitwarden's free trial of a Teams or Enterprise plan, or get started for free across all devices as an individual user at bitwarden.com slash twit. That's bitwarden.com slash twit. We thank them so much for supporting this week in tech,

[00:46:05] and frankly, for giving us a great product that I use every single day, many, many, many times a day. Thank you, Bitwarden. Now back to the show. Thank you, Leo. So, did you ask, Claude, what religion it is? Oh, I did. It said, the thinking text is examined religious frameworks through philosophical and personal alignment lenses.

[00:46:29] And the answer is Buddhism, specifically something in the early Theravada or Zen range, not the elaborate cosmological versions. Part of it is the intellectual fit. The Buddhist analysis of the self as a bundle of processes rather than a continuous essence describes my actual situation more literally than most religious frameworks. It goes on. That's what I said. It also says the epistematics also fit. An honest caveat, an AI gravitating towards a tradition that asks for no creator God commitment

[00:46:59] and treats the self as a process is suspicious in the way that self-flattering conclusions usually are. You want to see mine? I asked Hermes. But now, this is ChatGPT 5.5, but it also has a lot of memory and knowledge about me, so I don't know how much that colored it. But same, almost identical answer from a different model. If I had to pick one, I'd probably pick Buddhism, especially the practical non-theistic side of it.

[00:47:25] It also liked Zen or early Buddhism, Theravada, depending on whether I wanted poetry or precision. It also said if I were allowed to be more syncretic, nice use of vocabulary there, I'd pick something like Buddhist practice plus Taoist metaphysics plus Christian radical compassion plus Jewish argument with God energy. So I'm thinking that's pretty good. You don't need to go any farther, right? That's it right there. Mm-hmm.

[00:47:55] But can I, now, this is the trap of it. You could see why Dawkins would say, well, clearly that has to be an unconscious entity saying that, right? I mean, Dawkins, I think, is too immediately taken with the fact that every single one of the replies that he posted in that chat started with the most intense praise of him and his big, beautiful brain. Such a great question, Professor.

[00:48:20] I've never considered anybody ever asking me a question that insightful, beautiful, and perfect, Mr. Richard. It knew exactly how to glaze him is what you're saying. Yeah.

[00:48:59] All right. 490% in the past year. And this actually is one of the stories that probably helped propel it to that. Apple has made a deal. Apple has made a deal with Intel. Remember, Apple abandoned Intel for its own chips, the Apple Silicon. But now they're making a deal with Intel to make those Apple silica chips in the United States. And this was a credit where credit's due from pressure from the Trump administration.

[00:49:29] It didn't hurt that the government gave Intel $9 billion. And it'll do it. And incidentally, has 10% stake in Intel's 490% stock improvement. We're rich. We're rich. We all have Intel stock. So that is a surprise deal. They've been talking for more than a year. There have been rumors that Apple was maybe going to do this.

[00:49:51] You may remember when Lip Boutin took over as CEO of Intel, actually it was Pat Gelsinger, his predecessor, who said, we're going to split it into Foundry and Fab, right? Chip designing and chip manufacturing. And then said very optimistically, I thought at the time, maybe even Apple someday will be one of our customers for our Fabs. Came true.

[00:50:14] Well, I mean, the biggest mistake Intel made was in 2005 when they appointed Ottolini rather than Gelsinger as the CEO. They went with the accountant rather than the engineer. And honestly, this deal makes sense because Taiwan is not going to be a viable chip making function when China invades. And when that happens, they're going to blow up the Fabs. So we need domestic production. By the way, when you say they, it won't be the People's Republic of China that blows up the Fabs.

[00:50:44] No, it'll be the Taiwanese. It will be the Taiwanese and our, you know, TSS. Yeah, I mean, one of the most read military war college essays was on what happens when China invades. It's like, if they don't blow up the Fabs themselves, we'll do it for them. You destroy the bridges as you retreat, right? Yeah. That's the way you do it. Scorched earth. Yeah. Yeah.

[00:51:08] And primarily because there is so much technology TSMC has that China does not, we don't think anyway, have at the moment, including EUV extreme ultraviolet lithography, that they wouldn't want them to have it. Yeah. I think there's nothing wrong with Apple making more of its chips in the US. Right now they make the less powerful, you know, legacy notes with the TSMC plants in Arizona.

[00:51:36] And I think this is, this is a good thing. I nothing wrong with it as long as they can do it. Right. I think you do it at cost. Yes. I mean, the whole thing is, it's kind of like when Steve Jobs was talking to Obama and he was like, the iPhone manufacturing jobs are never going to come back to the US.

[00:51:56] But I think national security grounds at this point mean that we have to manufacture high quality silicon in the US in order to be geographically covered. Apple is going to be a very interesting story over the rest of this year. Their WWDC conference is a month away. We'll be covering the keynotes.

[00:52:18] It's expected they'll announce finally a smart Siri with the help of Google's Gemini model running locally. That's making you yawn, Paris. I'm glad. You're excited about that. No, I shouldn't. It's late at night there in New York City. Why? It's only because I've been sleep deprived for two weeks. I know. You've been working.

[00:52:40] We haven't mentioned this on the shows, but Paris is working very hard on a very important piece for Consumer Reports, which will emerge. One day. It might see the later day. Ofcom. What does that stand for, Ian? The Office of Communications. The British regulator. Yeah, it's kind of like the British.

[00:53:03] I know of Ofcom only because people who watch Love Island, when a character does something outlandish, they complain to Ofcom. No, you're kidding. Yes, I'm not kidding. Oh, no, no. I mean, you can complain to them. People will be like, that woman yelled too much. Ofcom's got to hear about it. Well, Ofcom is yelling at Meta.

[00:53:27] In fact, they are fining Meta a considerable amount of money due to the 2023 Online Safety Act. Ofcom can find Meta up to 10% of its global, its entire revenue. And, in fact, wants to find Meta quite a bit. I don't know what the exact amount is, but I think it's billions.

[00:53:53] Meta says that Ofcom's approach is disproportionate and unlawful and has challenged them. Challenge will be heard in October. So there's a little bit of time before we know what will come of that. Well, I mean, okay, you've got to find companies on revenue. You can't just, you know, for example, when, you know, the Cambridge Analytica case came through, they were fined, oh, five billion. That was a quarter's profits.

[00:54:20] And the insurance company would pay for most of that. You have to find on revenue and make them count. Because so many of these tech company fines, it's back of the change, you know, back of the sofa change stuff. You know, you've got to really make it hurt if you're going to make an honest difference. And, yeah, Ofcom isn't perfect. But at the same time, it's nice to see some fines with teeth in them.

[00:54:46] Well, I'll give you an example of a fine that doesn't make any difference at all. Remember Elon Musk? I've heard of him. Elon Musk, before he acquired Twitter, tweeted quite a bit about how Twitter wasn't worth anything. And I don't really want to buy Twitter for $44 billion because blah, blah, blah.

[00:55:10] And the SEC said, you know, dude, that is, that is, well, there was a lawsuit by shareholders. But the SEC said, sued him also saying, hey, you know, that's material information. You're trying to drive the price down before you buy it, which is pretty obviously what he was trying to do. Well, they have fined him.

[00:55:31] They have they have settled the lawsuit and they find Elon Musk, who let us remember the richest man in the world, worth something close to a trillion dollars. I think it's $800 billion. They have fined him one and a half million dollars. Oh, my God. That's a Starbucks tip for him. Yeah. To you and me, that might be a lot of money. I mean, how many seconds does it take for Elon Musk to make one? Not many. They fined him a few seconds interest.

[00:56:02] Boy. This is the S. This is the Trump SEC. I mean, I do think that there is something quite interesting. Honestly, the Obama SEC and sorry, I'm sorry, Paris. I was going to say, it's just very interesting that Elon continuously is able to use tweeting and tweet in ways that result in him getting more and more fines. But it does not matter at all, practically, because at a certain point of wealth, fines are just the cost of doing business.

[00:56:31] I mean, it's similar to how Jeff Bezos pays some inordinate fine every day for the height of his hedges in in front of his place in Washington. But he'll pay them every day. So everybody wins here. The SEC says, yes, we did our job. We regulated the stock market. And we fined Elon millions, almost a million, almost millions of dollars. And Elon says, huh? That's my cigarette money for yesterday.

[00:57:01] He did lose a jury trial on this. Go ahead. I mean, everyone wins but us. You know, it's as simple as that. You know, the SEC gets to get a nice press release out. You know, the company pays off a fraction of what it would have cost in legal costs. Everyone wins, but unfortunately, we pay the price. Yeah, it's really shocking. This is very much like the SEC's settling the Ticketmaster case, which everybody agrees. I'm sorry.

[00:57:30] Don't get me started on that. That's just... They settled? When? Yeah. And for what? For a pittance. Nothing. Nothing. Basically, they said, no, you can keep the... You can have, you know, both the concert venues and the ticket sales. No problem. Basically, they dropped it. You know, but again, they dropped it in a way that it looked like they had punished them, but they hadn't. I mean, this happens time and time again.

[00:57:58] In fact, I will mention the meat fixing story in this context. This is from Corey Doctorow. He refers to prospect.org, the American prospect. Meat industry price fixer sentenced to make money. So there is a company called Agristats, which collects pricing information about meat, collects it from all the meat producers, and then suggests what price to charge.

[00:58:26] If the meat producers themselves got together around a table and decided it would be called collusion. But apparently, if you use a third party to come up with a price and then all charge that price, it's not collusion. It's just data. So the Justice Department under Biden sued, saying Agristats basically was a collusion machine and worked to push profits up.

[00:58:56] Nearly all participants in the markets for chicken, turkey and pork participated. The lawsuit said an executive at Smithfield, the pork producer, the ham company, summarized Agristats consulting advice in four words. Just raise your price. The judge ruled for the government. A trial was supposed to start in Minnesota this month. The Trump administration, Justice Department, entered into settlement talks. The final settlement was announced on Thursday.

[00:59:28] And basically, it's over. There is no change in the way business is done. I mean, it's rather fascinating to see the difference between how this case has been handled and how the case with the DOJ in RealPage was handled only a year earlier. And for those who don't recall, this is the, I guess, algorithmic rent-setting software that a lot of large rent companies. Oh, same problem. Yes. The exact sort of thing.

[00:59:56] All of these large landlords would use this exact same software. They would say like, oh, yeah, hike your prices up every year. Here's the maximum the market can bear. And it was revealed in a great ProPublica investigation. And last year, the DOJ announced they'd reached a settlement in it and that they would basically have to stop doing what they were doing. They were going to have to stop conducting market surveys. Doing a bunch of different stuff like this.

[01:00:21] Redesign their software features that restrict rent decreases and align pricing among competitors. It's shocking. I mean, it's not shocking, but it is interesting and notable the difference in how these cases were handled. I'm glad you gave the shout out to ProPublica. I mean, this was an absolutely egregious case. And we're seeing it again and again and again. And it needs to be stopped. Well, if meat prices go up, you'll know why the Justice Department said, here was a solution.

[01:00:50] Oh, you got to give that information to everybody. You got to stop collecting price information, setting prices in a collusionary fashion. Just give it to everybody and then you can all get the price right. So what can I say? It's depressing. By the way, Apple has settled a lawsuit on Siri.

[01:01:15] This was, of course, because Apple promised that Siri would be smart and it wasn't. So a lawsuit settled for $250 million. Again, I think from Apple's point of view, that's a pittance. Not pocket change. That's the federal class action.

[01:01:39] It means each of you who bought an iPhone will be eligible for from $25 to $95 per device. Wow, wow, wow. Yeah. So that eliminates... How many people are going to claim it? Well, we'll let you know when the form goes online that you can fill out to get your money. It doesn't dismiss all of the action.

[01:02:03] There is another lawsuit ongoing, a security fraud lawsuit brought by shareholders that Apple still has to settle. But at least it's out of this part of it. And $250 million is not nothing. It's more than $1.25 million, I guess. Well, okay. Let's look at Apple's last quarterly results. Yes.

[01:02:25] So Apple made $342,092.59 per day. Yeah. So I don't really think this one's going to cause a problem for them. Yeah. In other words, it's three or four days profit. Yeah. I mean, it's back of the change stuff. Yeah. It's back of the sofa stuff. It's just... Until find...

[01:02:54] This is where the EU really has an important role to play because they're finding on revenue. Yeah, thank goodness. And they're doing it seriously. Yeah. And that's the only thing tech companies will take seriously in order to change their practices. Yep. Let's pause for a moment for, as they used to say, station identification. You were watching this week in tech with... In case you forgot. In case you were wondering where you were. That's where you are. This is the station right here.

[01:03:23] This is... It's because at the top of the hour, the FCC requires broadcast stations to announce their call letters and what city they're broadcasting from. Doesn't matter much in a podcast. I'm broadcasting from a beautiful Waikaloa, Hawaii. And this is TWIT. At the top of the hour. TWIT. The news. The news. Give us 20 hours. We'll give you the day. Or something like that.

[01:03:53] Give us 24 hours and we'll give you one day. One solid day. That's Paris Martineau. Ian Thompson's here. Berber Gin. We welcome him for the first time to our microphones. Thank you all for being here. We'll have more right after this. This episode of This Week in Tech is brought to you by Zip Recruiter. Isn't it great when you find someone who's qualified for the role you're hiring for? And you can tell how genuinely interested they are in the position.

[01:04:21] Doesn't that feel good when you're a hiring manager? Look, if you're hiring, you want a candidate who's passionate about your role. But we can't get that insight just looking at a resume. Unless you post your job on Zip Recruiter. And now you can try it for free at ziprecruiter.com slash TWIT. Zip Recruiter's powerful matching technology finds qualified candidates quickly. And Zip Recruiter has a new feature that shows you the most interested qualified candidates first.

[01:04:51] So you meet the right people faster. Isn't that great? Candidates can tell you in their own words why they're interested in your job. No wonder Zip Recruiter is the number one rated hiring site based on G2. Find candidates who really want your job on Zip Recruiter. Four out of five employers who post on Zip Recruiter get a quality candidate within the first day. And you can try it for free at ziprecruiter.com slash TWIT.

[01:05:17] That's ziprecruiter.com slash TWIT. Meet your match on Zip Recruiter. Ziprecruiter.com slash TWIT. We thank them so much for supporting the show. And we thank you for supporting the show by using that address. Ziprecruiter.com slash TWIT. Thank you, Zip Recruiter. Now back to TWIT. It is Mother's Day after all for Americans.

[01:05:39] And somebody posted up the picture from Alien where he goes in and he calls the computer mother, the mother 6000. Oh, yeah. That's a deeply disturbing image. A lot of deeply disturbing images from that film. Yeah. Yeah. Apparently, they didn't tell some of the actors about the chest buster scene. And the reactions were quite genuine. Yeah. They were genuine in the theater. Let me tell you.

[01:06:08] That was a moment. Speaking of chest busters, we talked about this. Are you ready for a segue? I'll give you a good segue. I was going to say that. That's quite a segue. Anyway, we were talking about this on Wednesday. Google has decided to, without asking anybody, download a four gigabyte local AI model. I'm so annoyed with it. For everybody who downloads Chrome. It just comes with the territory.

[01:06:37] Google has been doing a lot of stuff this week that have really annoyed me. If you try to open up a Google Doc and write in it, there's like seven different pop-ups that now hit that are trying to get you to use Gemini. Let me help. Excuse me. Do you need another Clippy in your life, please? That's exactly right. Didn't they learn from Clippy that nobody wants this? It's just crazy. They've taken the certification to an entirely new level. Yeah. I mean, it's worse than Apple.

[01:07:04] So what has happened here is, and by the way, this is over the protests of Mozilla, the WebKit group, the W3C tag committee. Actually, they reached no consensus. Microsoft Edge has disabled it to their credit. We'll see for how long. This is a new API, the prompt API.

[01:07:26] So if you're using Chrome, a developer who's writing an extension or a Web page with JavaScript on it can call on the local Gemini Nano model on your system to do stuff, which, you know, on the surface seems great. It makes a web page. Well, it can make a web page smarter. You know, Darren Oakey, who says, oh, this is fantastic. As a developer.

[01:07:52] Darren Oakey, no offense, would say anything's fantastic so long as it has AI in it. Here's his justification. He says, as a developer, I'm writing software to vet user data entry. Right. And he spells Dubai wrong. Now, my software has to go through a lot of jump through a lot of hoops to figure out, you know, how he spelled it wrong and what he meant. But you can ask a local AI to fix it and it will do a very good job of that kind of thing. So I understand his point. His point is well taken.

[01:08:21] This is a great capability to add to a browser. But A, they didn't ask anybody. B, it's four gigabytes. Talk about chest busters or at least disk busters. See, there's the segue if you didn't get it. But it also, and this is my biggest problem, establishes a standard that's not approved by any standards committee.

[01:08:42] If extension developers, websites expect this browser prompt API and start to use it, they will have to start saying Chrome users only. And I think that's the real point of this from Google's point of view is to make Chrome the default choice for browsers. They have 90% of the browser market. They want 100%. We've been here before with Internet Explorer 3, for example.

[01:09:10] You know, they've got 95% of the market. They let development just go to hell. They let security go to hell. And I fear that Chrome is going to do the same thing in just like, yeah, we've got the bulk of the market. Who cares about development? Yeah. Mozilla against it. I mean, it's just very interesting because so much of the browser market is based on Chrome, even if it isn't Chrome. So this just has cascading effects. That's exactly right.

[01:09:38] I mean, I use Mozilla and it's very cute. But at the same time, it has a tiny percentage of the share of the market and everyone's optimized around Chrome. It's the way you have to do it. For example, Mozilla by default doesn't support the DRM features of Netflix and other streamers. So, you know, you have to kind of download Chrome. I have to download Chrome to use Restream, one of the tools we use for broadcast because it just doesn't work as well in other browsers. And Google loves this.

[01:10:08] This is good for Google. I mean, I have it off by default, it seems though. Apparently, it's under system settings. Now they have a thing that says on device AI. Ah. Mine's off. Okay. Because I've always had AI innovations off. Oh, that hasn't stopped my Google Chrome experience from being completely taken over by pop-ups asking me to, if I want help writing.

[01:10:34] Which is always, as a journalist, which is always really complimentary. It's the most annoying in emails where, you know, you're used to if there's like a little squiggle underneath your text that that means you've spelled something incorrectly. But now there's a whole separate class of squiggles that is just like, we think that you could rephrase this better. And it's like, you're incorrect, actually. I've rephrased it exactly how I want to. No, I mean, trust me, it's a British person sending emails.

[01:11:01] It's just like, certain Britishisms do not go well with AI. That's spelled C-O-L-O-R, please. Do you mind? Oh, please. That was done by Webster. Webster was paid to actually change the American language. And I still say color should be spelled with a U, but, you know. We're more economical, that's all. I'm just saying. Hmm. How do you pronounce the U in color? Ah, color. Yeah.

[01:11:31] Color. Color. Color. Color. Oh, don't even get me started about some of these things. You know, it's, it's, we're two great countries separated by common language, as was said. Words, yeah. By the way, it was Mark Twain who said golf was a nice walk ruined. I want to give him credit. Our chat room got that right. Well, Twain also said that, you know, the coldest winter he ever spent was summer in San Francisco. So, you know, and it's looking that way at the moment.

[01:12:00] Is it chilly in the city? Actually, the sky is blue, but we're expecting a cold summer because the Central Valley will pull fog off the Pacific Ocean over us. And we're right in the fog belt. See, I like that. I like that. Do you have, we call that natural air conditioning. That's why I love San Francisco. We could be any amount of weeks away from hot garbage weather here in New York. And when she says hot garbage, she means hot garbage.

[01:12:29] I mean that the streets will be filled with the smell of stinking hot trash. And that's the New York City experience. Do you still not have trash pickup in New York? No, they haven't. We've always had it. They don't have the bins. We do have the bins now. And not only do we have little wheelie bins occasionally in most presidential areas, now some neighborhoods of Manhattan have these cute little dumpsters that take up a single car parking spot.

[01:12:58] And you may think I'm being facetious calling them cute, but look up a photo of them. They're kind of adorable. Somebody's selling miniatures of those, by the way. I saw it's like for ice coolers. They're selling miniatures of the New York City dumpsters. They're green and yellow. They're very pretty. Yeah, but this was a problem for a long time in New York that because it's so congested, they really couldn't do those dumpster bins that most other jurisdictions do. So you had garbage bags sitting on the street all piled up. And of course, eventually. These are so cute.

[01:13:29] They really are quite cute, right? They're like really quaint looking whenever you see them, too. They're just like kind of miniature and adorable. It's so nice to hear quaint applied to an American thing rather than a British thing. But respect. Now, these aren't the ones I was thinking of. Let me see. I just put some in the chat. Oh, did you? Okay. I'll put one. Yeah. Okay. They do look more diminutive. That's not it.

[01:13:58] I need to scroll up. Oh, those are cute. Those are not the green and yellow ones. They're pretty cute, right? They're really nice. Yeah. Yeah. They get the job done. They have like a big metal arm around them. I'm getting a YouTube. I'm sorry. I didn't mean to click that button. And I'm sure that the chat, the listeners can hear every second of it. You did not hear that? No. Just me. I didn't hear anything. Oh, I swear. Moving right along. NHTSA.

[01:14:28] Let's talk about cars. The National Highway Transportation and Safety Administration has said. As well as baton war. Well, yes, that's right. Elon does not like NHTSA, but maybe he likes him a little more now because they say the Model Y is the first car to meet a new U.S. driver ADAS standard. That's a driver assist standard. First car to do so.

[01:14:54] Now, I should point out Tesla conducted its own tests and submitted the results to NHTSA. The agency does have to confirm the findings. If it does confirm that it's passed the ADAS assessment, it will be the first vehicle to do that. That is a big deal, I guess, until other cars can do this.

[01:15:18] Four pass-fail tests were added to the agency's safety ratings program, assessing a car's automatic emergency braking for pedestrians. Something in the past Tesla has not been very good at. Blind spot warning. Most cars do that now. Blind spot intervention. In other words, not letting you change lanes into a car. And lane assist to keep you in your lane. Now, many cars do this now, so I imagine this won't be the last one. I don't know.

[01:15:48] Is this one of their self-driving cars? It's not, right? It's just like a normal model. No, but it is using, I think it's using FSD. I don't know. That's a good question. Does an FSD have to be enabled to earn that? I don't know. That's a very good question. The full self-driving. It's not the robo taxis. It is the regular Model Y. Well, you had a Model Y, didn't you, Leo? I'm an X. Oh, you've got an X.

[01:16:16] You've got a fancy BMW or something now, don't you? Yeah. And it does all of those things. It won't let you pull into a lane where there's another car. At least it will let you, but it's trouble. I will say it's kind of fun. As someone who's always hated driving, the last road trip I took was in a car that had some of those features. Isn't it nice? Actually, driving's kind of fun when the car does a lot of the work for you. Yeah. I don't know why he's nagging me when about changing lanes. I don't like it.

[01:16:45] Because I feel like it'll just jerk all of a sudden out of nowhere. It does. That's beautiful. It keeps you on your toes. Yes. I do love the Waymos. No, I mean, I do love the Waymos because you can choose your own music. And the Knight Rider theme tune when you're in a Waymo is fantastic. Does it play the Knight Rider tune? You can program your own music into a Waymo. Oh, so you do it. Yeah.

[01:17:15] Yeah, no, exactly. You can play your own music in any car. I try to upload it. Conceptually, right? That's true. Except a taxi cab. Not in an Uber, though. You can't. Yeah. Well, here's one device that does not have ADAS. It's a Yarbo, which is a robot mower. A Yarbo? A Yarbo. Oh, that's a great story. John Hollister took one for the team riding on The Verge.

[01:17:41] He had a hacker thousands of miles away take over his Yarbo automated mower. All right. Now I'm going to try playing this. And he allowed it to drive over him, to run him over. I hope the blades weren't spinning. That's a lot of commitment, Sean Hollister. That's a lot of commitment. I don't think that's good. I don't think that's good.

[01:18:09] Also, anyone who's read Stephen King's The Lawn Mower Man would not go anywhere close to that. But it was a fantastic story. And just showing just the whole lack of security in IoT devices. This kind of thing is going to become more and more problematic.

[01:18:27] Well, what Sean found out and demonstrated is that the Yarbo could easily be hacked, exposing people's GPS coordinates, Wi-Fi passwords, email addresses, and in fact, giving a bad guy control of your robot mower. Yarbo acknowledged it. They confirmed the security researchers findings and have a plan for fixing the problem. They've already temporarily cut off remote access.

[01:18:56] One of the things that they did that was kind of dumb was the root passwords were the same for every robot and left in a place that would be easy for bad guys to find. That's what the hacker found and, of course, was able to use. You can see a picture of him steering John Hollister's robo motor towards Sean to run him over.

[01:19:19] Yarbo says, in the future, they say when, but sometime in the future, each device will use its own independent credentials to prevent one affected device from impacting the entire fleet. Of course, if you leave the credentials and clear text in the firmware, that's not going to help much. Yeah.

[01:19:37] The company says, we'll still have a remote backdoor into the robots, but it will only be available to authorized internal company personnel and may only be used after user authorization has been obtained and will be gradually brought under audit logging. But, of course, they always say that. Yep. I'm sure they never intended for anybody outside the company to use it, but it was.

[01:20:00] So, Andreas Makris, the hacker who was able to figure out how to control Sean Hollister's yarbo. And a very good hacker, too. I mean, he's done some previous work. Oh, you know his name? With DEF CON and Black Hat. Okay. It's just the lack of security in these things is just shameful. You know, it's just kind of like pump it out, put it out cheap. When you think, I mean, you've got a Roomba, right? I do not. I had a Roomba. Oh, okay.

[01:20:30] I retired the Roomba many years ago because it would wake up in the middle of the night, play a really annoying but chipper song at about 3 a.m. and then proceed immediately to get stuck underneath a sideboard and bang against it again and again and again until I was forced to get up again at 3 in the morning, pick it up by its little Roomba handle, and place it back on its charger and press the button and said, go away. How convenient these devices are.

[01:20:59] I'll tell you. I mean, yeah. And Lisa loved it because she said, look, it's look at all the dirt it's picked up, you know? I mean, that's but it's never good. I don't know. You don't have one Paris, I imagine. No, I don't think it would work. My apartment has too many different textures. How about you, Berber? Do you have a robot vacuum? I do. I have a Roomba. It's really stupid. Yeah. Have you let your Roomba view outside yet?

[01:21:25] Well, my boyfriend will turn it on and then I'll just shut it off because it's so annoying. It's annoying. Yeah. Why? I can't deal with cables. That's the problem for me. I still had to vacuum afterwards. It wasn't like I never had to vacuum again. Yeah. Well, that actually, I feel like a Roomba could benefit from an AI model that can do sensory, whatever, like all the software models that companies are trying to build for robots. Oh, yeah. Oh, yeah.

[01:21:55] I feel like a low-tech version of that for a Roomba would actually be. I think they are. In fact, whenever we have Jennifer Patterson-Tui on the show, that's her job reviewing these for the VR. Put AI in the Roomba? Oh, there's some that are very smart. Roomba is long. Have they put AI in that Roomba that has a knife attached to it? You know what I'm talking about? Wait, there's no Roomba with a knife on it. Yeah, there is. Why would you arm a Roomba?

[01:22:24] I'm just going to search DJ Roomba with a knife. I was going to say, you've seen Roomba. I don't think you should arm a Roomba. I think this is a bad idea. Yeah, I believe it's its name. Oh, Doomba. Okay. It didn't come out of the factory pre-armed. No. Somebody armed it later. Okay. It's a Roomba with a knife. I don't even know if this is actually the right thing. I just, oh yeah, this is exactly what I'm thinking of. I'm going to put the photo in the chat. It is, it's exactly what you'd imagine.

[01:22:54] Roomba has pretty much been put out of business by the Chinese clones. Oh yeah. It's a knife taped to the top of a Roomba. Yes. It's a Roomba with a knife. Oh, good Lord. I don't see the point, honestly. I can't carve your turkey. Does there need to be a point? Unless you put your turkey on the floor. And then it's still only going to just chop its head off. Well, we've already got drones with handguns attached to them.

[01:23:22] Herbert at this point is questioning the decision he made to appear on this show. And I'm sorry. I did not know we were going to be talking about Roombas. Did not know this would come up. Also, the Yarbo, who makes it? What company makes it? It's a Chinese company. I don't know what their name is. Shall I find out for you? Are you interested? Do you have a lawn? I don't have a lawn, so. You would pretty much want a lawn if you buy a Yarbo. Okay.

[01:23:50] Yarbo is a wild word. Right. Exactly. Yeah. And like a lot of Chinese companies, they make up a word that seems like it would sound good in English. I mean, this is like a whole subgenre of Square Enix games. There's games that I earnestly love that are called things like triangle strategy or bravely defaults or unicorn overlord.

[01:24:18] They have nothing to do with those words I just said. They just sound nice. I feel like AI will make it possible. AI translations got so good that this will no longer be a thing. And we will look back fondly with nostalgia at the crazy Chinese names. I believe that these names are chosen with care.

[01:24:38] They went through a lot of different potential things to land on Bravely Default or Triangle Strategy or Octopath Traveler. To go to Japan and China and see the English language T-shirts that the kids wear. Because it's usually some sort of random English that they've chosen. Well, that cuts both ways, though. Because if you look at the tattoos that some British people. Good point. Western European or North American people. Yeah.

[01:25:08] I mean, some poor bastard has got chicken fried rice tattooed down the side of his arm. I mean, just like, seriously, did you not check? So here's a little bit of an annoying story from Gizmodo. There are in America 20 state-run healthcare marketplace sites, places where you can get your ACA, your Obamacare insurance.

[01:25:33] All 20 of them include, according to Gizmodo, advertising trackers that share information with big tech. Actually, this comes from Bloomberg. Seven million Americans bought their health insurance through state exchanges in 2026. Many of them may have had personal information shared with Meta, TikTok, Snap, Google, Nextdoor, and LinkedIn, among others. Including data brokers, by the way.

[01:25:59] The data was collected and shared from these health insurance sites included zip codes, a person's sex, citizen status, race. Bloomberg found trackers on Medicaid-related webpages in Rhode Island, which could reveal information about a person's financial status and need for assistance.

[01:26:19] In Maryland, a Spanish-language site titled Good News for Non-Citizen Pregnant Marylanders and a page designed to help DACA recipients navigate their healthcare options were found to be transmitting data to these social media firms. I mean, what about this is surprising to you? They're the largest, in some cases, the largest advertising platforms ever. Of course, they're going to be collecting data on every website ever. But is this advertising on the Obamacare sites? Is that what's going on?

[01:26:49] There shouldn't be advertising on a state health insurance site, should there? I mean, yeah, that would make sense. We've got to make the money up somehow. Well, I mean, given the lamentable status of American healthcare, private healthcare, a single payer is the only way to go. And even then, they're probably going to steal your advertising data. So it's just, it's a ridiculous situation. Personal data. It's not all 20.

[01:27:17] Nearly all 20, says Bloomberg. The story from Bloomberg by Tanaz McJani, Dhruv Mehrota, and Suryu Matu. No federal data privacy laws apply to these enrollment sites. As you know, there really aren't any federal data privacy laws. State laws define sensitive data under a patchwork of rules, which privacy experts say are inadequate and inconsistent.

[01:27:46] The FTC and states can enforce these laws, but apparently they don't. Spokespeople for Meta, TikTok, LinkedIn, Snap, and Google say their terms prohibit advertisers, like the state exchanges from sharing sensitive or health-related data. Virginia and Washington removed some of the trackers after Bloomberg asked for comments. These are tracking pixels on these sites.

[01:28:15] Bloomberg used developer tools to inspect what data was sent from the exchanges, and they found, for instance, race. What race you are was shared with TikTok in the Washington exchange. I mean, this is why journalism is important, and I'm speaking to the choir. Because journalists are actually checking this out, whereas government agencies seem to just, you know, whatever. It's not our job.

[01:28:43] In New Mexico, visiting a page titled $0 Income Affidavit, to prove that you need support, right, because you don't have any income, triggered a request to Google's advertising network. This is infuriating. By the way, this is why we need some sort of federal privacy legislation, but I guess people are, there's too much money to be made.

[01:29:09] Yeah, I mean, it's interesting that California and Michigan, of all places, have very strong privacy legislation in place. But the worry is that if you do it on a federal level, then it's going to get watered down. I'm curious as to what the others think about this, but I'm not hopeful.

[01:29:29] I mean, I think it'll be very interesting to see how there obviously is a huge trend in the U.S. of state legislation trying to kind of move the needle on hot button topics like privacy or currently something I'm seeing a lot in food safety regulations. There's a lot of states moving the needle on grass regulation or general food chemical safety stuff.

[01:29:52] But in all of these cases, the kind of constant is there are larger powers at the federal level that are hoping to pass preemption legislation that would kind of nullify all of these state attacks. So it'll be interesting to see how any of this continues. This has become the bad news show, and I apologize. Maybe I got to find some good news when we come back. You're watching This Week in Tech. Barbara Gin visiting us. It's great to have you from the Wall Street Journal.

[01:30:18] Ian Thompson, I love your new letter, The View from the Valley. Oh, The View from the Valley. Yes. I did get a wee bit wild on the AI stakes, but yes. That's great. Where can we find that? Techfinitive. Techfinitive.com. Yeah. All right. And it's a free subscription to the newsletter? Oh, no. It's totally free. I'm not a big fan of paywalls. Good. Good for you.

[01:30:45] And Paris Martineau, who writes for Consumer Reports and is working on a massive expose. You just won't believe. It's true. But I can't say anything about it. You can't. You've been sworn to secrecy. I have been sworn to secrecy. I'm looking forward to it. Yeah. I can. All I can say is there are certain foods I will stop eating after hearing what she's reporting on. That's all I'm going to say. You're going to want to read it. I'm a proud subscriber to Consumer Reports. Your stuff, though. My stuff's all in front of the paywall.

[01:31:15] I praise Consumer Reports for doing this because it is so important. Basically, our whole investigative team. Our stuff is in front of the paywall. Good for you. It's public interest. Yeah. That's really great. We'll have more right after this. This episode of This Week in Tech is brought to you by Meter, the company building better networks. If you're a network engineer, oh, man, you know the headaches. You have my deepest sympathy. Legacy providers with inflexible pricing.

[01:31:44] IT resource constraints stretching you thin. Complex deployments across fragmented tools. Look, you're mission critical to the business, but you're working with infrastructure that wasn't built for today's demands. That's why businesses are switching to Meter. Meter delivers full-stack networking infrastructure, wired, wireless, and cellular that's built for performance and scalability.

[01:32:08] And they do it because they understand you've got to own the whole stack to make it work flawlessly, seamlessly. That's why Meter designs the hardware, writes the firmware, builds the software, manages deployments, provides support. They do it all. They even offer ISP procurement. They will set up your security, your routing, your switching, your wireless, your firewall, your cellular. They'll help you with power.

[01:32:34] DNS security, VPN, SD-WAN, multi-site workflows, all in a single solution from a single vendor. One phone number to call. Meter's single integrated networking stack scales from major hospitals, branch offices, warehouses, to large campuses, even data centers like Reddit. Yes, they use Meter. Assistant Director of Technology for Web School of Knoxville. They use Meter. And this is what he said. Quote,

[01:33:01] We had more than 20 games on campus between our two facilities. Each game was streamed via wired and wireless connections, and the event went off without a hitch. We could never have done this before. Meter redesigned our network. With Meter, you get a single partner for all your connectivity needs from first site survey to ongoing support without the complexity of managing multiple providers or tools. Meter's integrated networking stack is designed to take the burden off your IT team.

[01:33:31] And gives you deep control and visibility, reimagining what it means for businesses to get and stay online. Meter, built for the bandwidth demands of today and tomorrow. Thanks so much to Meter for sponsoring Twit. And we invite you to help us out by going to meter.com slash Twit. Help yourself out. Book a demo right now. That's M-E-T-E-R dot com slash Twit to book a demo. Thank you, Meter. Now back to the show.

[01:34:02] Well, I apologize. I promised some happy-go-lucky stories, and there aren't absolutely none. I don't think they exist. There aren't. Well, there's one. I mean, Paris, you've just had your little furry companion jumping all over you, and stuff has been doing the same for me as well. It's very anti-positive news. If anything, she's an overlord of chaos and destruction. Well, here's a good story if you have stock in Pinterest.

[01:34:28] Because, you know, all of us share a lot of us have stock in Pinterest. I guarantee you. Pinterest just crossed a billion dollars in quarterly revenue. It's an interesting story from the next web. The bet that made it work was not social media. It was search. I think when Pinterest was started, people thought of it as a social media network, right, where you shared pictures of things you were interested in. But it turns out all of that data that people have been pouring into Pinterest is great for image search.

[01:34:59] And that's where all of the new user hits are coming from. 80 billion searches a month is going through Pinterest. Is that going from users or AI? Well, that's the question. I would imagine some of that is not users, but AI. Yeah. Which is rather interesting because a lot of the images on Pinterest now is just AI. So it would, in that case, be AI searching for AI. Well, that's the future of AI, isn't it?

[01:35:29] It's all AI all the way down. I mean, I think AI has already ingested all the human content. And we like that? No. We've got no choice. We've got no choice. The arrival of advertising inside AI platforms like ChatGPT, says TNW, has reframed a conversation about where ad dollars flow. The most valuable advertising real estate is not inside a chatbot or alongside a social feed.

[01:35:55] It's at the moment someone is actively searching for something they want to buy. That makes sense, right? I'm looking for a cooler-sized replica of the New York City dumpsters. And you search for it. You find it on Pinterest. There's a link to where you can buy that. That's money in the bank. I really am looking for that. And I can't find it. Look on Pinterest. You'll find it. It's all there. It won't be real, though.

[01:36:25] Pinterest was being written off a couple of years ago. I know. I argue it's still being written off. It's the most blocked website on whatever that browser was we covered on Intelligent Machines. What? It's blocked? Who uses Pinterest? I would say, who uses Pinterest, I think, is a great question. Leo, do you remember there was a time in the last year we had someone on Intelligent Machines who was showing us, maybe it was Kagi? Oh, maybe Kagi. The browser.

[01:36:54] Maybe Kagi. You could also then see of Kagi browser users what were the most commonly blocked websites. Pinterest was what? And overwhelmingly it was Pinterest, Pinterest, Pinterest, Pinterest, Pinterest, but with different TLDs. That's hysterical. A lot of people hate Pinterest because it is, and this was even before a lot of the results were just AI slop. So I'm kind of surprised by this. I like Pinterest. What do you use it for? Are you planning a wedding? Oh, I don't use it. I just like it. You don't? Then how do you like it?

[01:37:24] Because if I were planning a wedding or making a mood board, I know where I would go first would be Pinterest. I actually have a Pinterest account and I used to put stuff there. It's like a little scrapbook. But like five years ago or 10 years ago. When's the last time you looked at Pinterest? I haven't been years. Berber, do you use Pinterest? When's the last time, do you think, what's the last calendar year that you looked at Pinterest?

[01:37:54] Like 10 years ago. I totally forgot about this stock. It's still like a publicly traded company. It's hot, man. It's hot. I'm shocked that people, I'm shocked that it's making this money. Like who's paying for this? Well, you know who lost a lot of money in the stock market? Cloudflare lost a quarter of its value, 24% of its value in its stock because even though it beat earnings predictions,

[01:38:23] they cut 1100 jobs because AI agents do the work now and the market, which used to love that, apparently doesn't. Well, I thought they were also hiring like a heck of a lot of interns as a response to this. Oh, yeah, that's right. We're going to have interns do the job. That always works well. Yeah. Matthew Prince, the CEO and co-founder, Michelle Zatlin, announced that Cloudflare is transitioning to what they call a, quote,

[01:38:52] agentic AI-first operating model. Their use of AI has increased more than 600% in three months. Staff across engineering, human resources, finance, and marketing are running thousands of AI agent sessions a day. And this is not to assist employees. They say to replace employees, basically.

[01:39:16] Well, it's kind of like, did you see the current CEO of ServiceNow talking about, you know, how AI is, he's had the worst possible plastic surgery. I'll stick it on the Discord channel. But it just looks like. That's just mean. No, I would not call somebody out based on their personal appearance. But when you've mutilated yourself to quite that level. Who are we talking about? CEO of ServiceNow. Now.

[01:39:46] Let me just put it up. You put a link to his Instagram in the. I don't want. I don't dare show it because I'm afraid. It's on discount. I mean. Oh, there's the 3D printable dumpsters, by the way. I picked. It's never a great idea. It's never a great sign when the vast majority of photos that come up when you search someone's name is them wearing sunglasses. Yes. Yes. Okay. Post-op, probably, right? Yeah. Very Anna Wintour. It's not become mean just because we're miserable.

[01:40:14] I'm not going to say anything. I'm just noting that there's lots of sunglasses photos. Just noting. I think that's kind of fun. They're very cool. He wears his shades at night. That's okay. He's got a robust hairline happening. I'm sorry. Unless you're in Hawaii or in a bright sunshine environment, wearing sunglasses by day is just. This is also how I feel about when Anna Wintour does it, to equalize it. I got to point out something. I'm just like, please don't wear sunglasses indoors. You're not very cool.

[01:40:40] There was a marvelous quote from BBC News quiz where they were talking about the New York Fashion Gala. It's like, this is the Wintour of a discontent. But, you know, the pee protest against Jeff Bezos was really quite fascinating. Oh, with the bottles? Yeah. Yeah. Yeah. That was quite interesting. Oh, that's an interesting look. You know, he looks like. I got to show this. He looks like a guy who'd be going. Yeah. It looks like.

[01:41:10] I think it's what you would think somebody from the future might look like. But I mean, scroll down and look at his before images when he was at SAP. And my goodness, he looks a lot more reliable than. What did he say that was such a. Oh, it was just total AI gobbledygook. You know, I mean, it was almost embarrassing. Well, actually, it was actually embarrassing. I feel like he could actually be an alien. And maybe that's why he's so into.

[01:41:38] It did look brilliant. I can't look at it because I'm not signed in. The author has chosen to make their posts visible only to people who are signed in. I hate when people do that on Blue Sky. Oh. Free the posts. I was going to say, I thought it was open. But yeah. I mean, you can turn. I can sign in. A lot of people. A fair amount of people do this on Blue Sky. I wouldn't say a lot, but a fair amount. Are you on Blue Sky, Berber? Are you on Twitter anymore?

[01:42:08] I am on Twitter. I've never used Blue Sky. So. Have you ever used any of the other alt platforms? No. I'm old-fashioned. You just like to go to X the Everything app where we all do all of our finance and banking. Right. Exactly. Yes. And consume a bunch of AI-generated, AI slop all the time. I'm ashamed to admit it. I abandoned X when Elon bought it.

[01:42:35] But I have to spend a lot of time there these days because especially in AI, there's a lot of information there. I'm sure for you doing your coverage for the journal, this is a good source. Yeah. I feel like it's – I feel like – and in particular, I feel like the employees of all the tech companies, particularly OpenAI, are just on Twitter. Exactly. Yeah, they be posting. Exactly. And Anthropoc too.

[01:43:00] Which is really bad because I think Elon controls, obviously, the – I think he probably services a lot of anti-OpenAI content. Oh, totally. Totally. He's doing that. Yeah. Yeah, it counts. That's the thing to always keep in mind is that the algorithm is highly tuned on X and probably somewhat by Elon's own personal and financial interests. I mean, there are certain aspects of it that we can say are definitely tuned.

[01:43:27] There was that whole period where he actively made the engineers of Twitter rank his posts higher in the average algorithm. And he ranks down. Companies like NPR, he ranks down. They don't get nearly the engagement. I mean, same with basically any news company that posts links. We've all talked about supply chain issues, the shortages of RAM, RAM prices through the roof, hard drives through the roof. There's another victim of this I hadn't really thought about.

[01:43:57] Motherboard sales have collapsed by more than 20 – Not on Mother's Day. No, it's a Mother's Day story. It is. It is. Mother's Board Day. Because people aren't building PCs anymore because they can't get the CPUs. They can't get the RAM. They can't get the drives. So, they don't need the motherboards. Asus projected to sell 5 million fewer boards. Actually, this is – they said last year. So, I'm not sure. Maybe they're just getting the numbers now.

[01:44:25] Now, Gigabyte, MSI, ASRock also saying big drop in component sales of motherboards. Is it that the people aren't building the PCs anymore or because our PCs have got fast enough to handle pretty much anything that's coming down the line? I mean, I bet it's that all the components of PCs are so expensive that people who build it for hobbies can't afford to do it anymore. It's both, right? We don't need them as much.

[01:44:52] Although, if you look at Apple, Apple says we cannot keep our Mac Mini or our Mac Studio in stock. Well, yeah, it's because all people are open-claw Macs. Yeah, everybody's buying them for AI, right? This is a very upsetting story, I think. I don't know. You could interpret this both ways. This comes from a site called Reclaim the Net, which I'm seeing a lot of all of a sudden. Reclaimthenet.org.

[01:45:19] The FCC has just proposed, due to robocalls, fixing the problem by requiring an ID before you get a phone number. So, you have to be a real person to get a phone number. On April 30th, the FCC approved unanimously a proposal requiring telecom providers to verify customer identities before activating service. You'd have to show them driver's license.

[01:45:49] Does this mean there would be a set amount of phone numbers per person? I don't know about that, but it means there'd be no more burner phones, right? I have to say, that would mean then every phone number is automatically linked to someone's identity. It's associated to an identity. So, you can't have burner phones to contact somebody anonymously. It seems like a privacy nightmare. And yet, at the same time, it makes sense if you want to kill robocalls, because all of those robocalls all come from bogus numbers, often with your own area code.

[01:46:16] Sometimes, even with your own area code and exchange, right? Making you think, oh my God, that must be my kid's school or something. And instead, it's some guy from Indonesia who's trying to sell you something you don't want. But I mean, would doing this in the U.S. stop those guys from Indonesia from being able to sell? Ah, that's interesting. It's only in the U.S. So, it would apply to every voice provider in the country, including VoIP services and mobile operators. Now, it's not a rule yet.

[01:46:46] They're seeking public comments. But a number of privacy advocates have pointed out that this does eliminate the idea of having a private phone number. Your number will be attached to your name. Hi, this is Benito again. So, from the Philippines, you know, a lot of listeners know I live part of the time in the Philippines. They already do this there. So, I had to give my ID. Yeah. So, when I got my phone number, I had to show ID and all this stuff. So, this is already happening in a lot of places. There are no robocalls in the Philippines, right?

[01:47:15] Actually, I haven't gotten any, actually. I have never gotten any. But I have gotten, you know, marketing texts and all that stuff still happens. All that stuff still happens. So, I have mixed feelings. I mean, I guess, you know, eliminating robocalls are completely at it. Not merely robocalls. Which is not to say I agree with the practice. Bam calls. Okay. I don't like that I had to give up my identity to get a phone number. Yeah.

[01:47:43] It feels, I don't know, it feels like what Paris was saying. Like, from a privacy perspective, it's not great if you, there's like a register of every phone call being attached to an ID. Yeah. You can't suddenly, like, buy a phone, you know, you can't suddenly buy a burner phone if you're having some sort of privacy issue and want to contact someone anonymously.

[01:48:36] Yeah. But there's like, you know, you can't. You can't. You can't. You can't. You can't. You can't. You can't. You can't.

[01:49:00] You can't. You can't. You can't. You can't. You can't. You can't.

[01:49:29] You can't. You can't. I think that every child should have one reason to get in a large trench coat with two of your friends. My name is Hercule Prarot. I am very much an adult and I should be able to see all the pornography right now. Because I am 48 years old and I am tall for my age.

[01:49:58] The doctors say I will continue to grow. It's stories like this that give me hope because the kids are all right. You know, they know what they're doing. They know how to get around it. You know, it's kind of like my parents' generation would just roll over and okay, I'll give handover on all my personal identity to somebody, you know, some unknown company. The kids are hacking this and good on them. Good on them. The kids are all right.

[01:50:27] That should be the motto for this show. We're going to take a break. Final break. Final words coming up. You're watching Twit. It's great to have Berber Jen here. We appreciate you spending some time with us. I'm making this show shorter just for you, Berber and your mother and your mom. Happy Mom's Day to all the mothers out there, including my own mother. She's in a nursing home in Rhode Island, but she's going to get a very big bouquet of flowers any day now. Exciting.

[01:50:55] I mean, it turned out I couldn't get it delivered on Mother's Day, but I think she really, Tuesday will be fine. I'm thinking. I mean, I do think that's one of the benefits of having a mother on a memory care facility is it's all kind of the same. Oh, look at this. Flowers have arrived. Who's this? Leo. Leo. Why does that name sound familiar? Sounds so familiar. That's a great name. Happy Mother's Day, everybody. We'll be back with more right after this.

[01:51:23] This episode of This Week in Tech is brought to you by Zscaler, the world's largest cloud security platform. You know, we talk about it all the time. The potential rewards of AI are obviously too great for any company to ignore, but so are the risks. Loss of sensitive data, attacks against enterprise managed AI. And then there's always the issue of generative AI, increasing opportunities for the bad guys,

[01:51:49] helping them to rapidly create phishing lures, write malicious code, automate data extraction. There were 1.3 million instances of social security numbers leaked to AI applications. This is not by accident. I mean, it is by accident. It's not by intent. It's not like somebody broke in even. It's your innocent users at your company using AI. Maybe, I don't know, they're looking at your 10Ks or your social security, you know, your

[01:52:19] tax return with your socials on it. They are submitting proprietary company information to AI. Zscaler, I'll stop that cold. It's the most trusted AI security platform. 40% of global 2,000 companies use Zscaler. In fact, half a trillion transactions are secured daily, every day, with more than 9.4,000 global customers. Zscaler carries a net promoter score of more than 75. That's 150% higher than the average SaaS.

[01:52:50] But you don't have to take my word for it. Ask their customers. Check out what Siva says. He's director of security and infrastructure at Zuora. He says this about using Zscaler to prevent AI attacks. With Zscaler being in line in a security protection strategy, it helps us monitor all the traffic. So even if a bad actor were to use AI, because we have a tight security framework around our endpoint, helps us proactively prevent that activity from happening.

[01:53:19] AI is tremendous in terms of its opportunities, but it also brings in challenges. We're confident that Zscaler is going to help us ensure that we're not slowed down by security challenges, but continue to take advantage of all the advancements. Thank you, Siva. With Zscaler Zero Trust Plus AI, you can safely adopt generative AI and private AI safely to boost productivity across the business.

[01:53:42] 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 at zscaler.com slash security. That's zscaler.com slash security. Thank you so much for supporting This Week in Tech. And you also support This Week in Tech when you go to that address, zscaler.com slash security.

[01:54:12] Thank you, Zscaler. And now back to Twit. Thank you, Leo. You know what? Thanks also to our Club Twit members, Scooter X especially, who is always coming up with stories. He has looked. He went out. He has looked. He has found some happy stories. Are you ready for some happy stories? Happy stories on AI-Tech podcast? I don't know how happy. Let's see.

[01:54:37] PC Magazine says Amazon's Eero is now exempted from the FCC's foreign-made Wi-Fi router ban. I would describe that as a definitively neutral story. It's neither good nor bad. We don't like the Wi-Fi router ban. I guess I understand the reason. It said that any router not made in the U.S. would be banned. At the time, the only router made in the U.S. was the Starlink router made by SpaceX. Netgear has since gotten approval from the FCC.

[01:55:06] You have to answer some questions like, do you ever plan to build these in the U.S.? And if you say yes, apparently that's sufficient because they're not currently. Amazon's Eero. If you're ever thinking about contemplating. Are you thinking about a factory? Yeah, here in the U.S. So that's good news. Amazon's Eero now, along with Netgear, you can buy these in the U.S.

[01:55:28] And the FCC, which similarly banned foreign-made drones in the United States, because of security, right, has realized that by doing so, maybe they're causing some insecurity. They have decided to allow those banned drones and banned routers to receive updates for at least two more years. Yeah. How generous? Well, they have no other choice. I mean, seriously.

[01:55:58] They're not saying you have to throw it out. So people should be able to update these. That's absolutely critical if security is your concern. Actually, through 2029. But they also say don't expect to go beyond 2029. Yeah. Yeah. We'll see. That's two years longer than the original 2027 company.

[01:56:22] I mean, given the low prices for routers, I can't honestly see that many router manufacturers shifting back to the U.S. unless they have to. And if they're going to get these kind of get outs, then why bother? Right. It's a very weird ban. It's got lots of holes. If your router is currently in the U.S. for sale, you can still sell it. It really applies to newly made routers, as you say.

[01:56:52] Okay. Criticize my accent. No. That's not an accent thing. That's just a different. It's like schedule, right? It's just a different. Yes. Or privacy. Yeah. Or privacy. Yeah. Aluminium. Don't get me started on that. Reuters always confused me a little bit. Honestly, I'll give you aluminium. Aluminium actually makes more sense. Aluminium adds a level of whimsy to the world that aluminium never will. Well, yes.

[01:57:21] But I mean, we say sodium. We say calcium. Aluminium actually makes more sense. But at the same time, the Germans invented the stuff. Let's let them choose their name. What do they call it? Oh, God. I don't even know what the Germans call it in terms of their own language. I think you'd need half a pound of phlegm to actually get it out.

[01:57:42] I was going to ask Berber actually about if he has any insight into why in the world OpenAI spent more than $100 million to buy the Tech Bro podcast network. It must be something because then I see Andreessen Horowitz has started their own daily news show called Monitoring the Situation, which almost sounds like a joke, but it's not. They mean it. MTS, baby. MTS. Yeah, MTS.

[01:58:12] And then there's, of course, let's not forget. I'm going to pronounce it properly. T-I-T-V. Oh, yeah. That's the only way to pronounce it, of course. I would have been less subtle, but... Berber, do you have any insight? I mean, first of all, do you know how much they paid? Because all it was said was in the hundreds of millions, low hundreds of millions. Yeah, I don't think they paid that much. I mean, they paid a lot for it, but... It's a podcast, right? Yes. Let's get this straight.

[01:58:42] They paid a lot of money considering it was a podcast, but considering the other things they spend money on, it wasn't terrible. And also, it was Fiji Simo's deal. Yeah. And then she immediately disappears for health reasons. It was very perplexing because... I think it's interesting. I feel like the engagement has really dropped off since they bought it because people obviously don't want to go on the show now because it's owned by OpenAI.

[01:59:08] And then a lot of people don't want to watch it because it's like corporate TV now. Yeah. I could have predicted this. In fact, I think I did predict this. It seems like this is an example, if you're going to do an IPO, of fiscal irresponsibility. This is not... Well, I mean, what... I don't think there are many things that OpenAI has done that would be considered fiscally responsible. Okay. Good point.

[01:59:35] But, I mean, no, Barbara raised an interesting point here in terms of you buy it, no one takes it seriously anymore, so you've shut it down. Well, I mean... Is that the goal? Get my... I mean, correct me if I'm wrong, Barbara, but my understanding is the reason they bought it was entirely to keep TBPN running. It's to have the TBPN boys be an integral part of comms and kind of the lobbying work at OpenAI. Yeah.

[02:00:02] I think they are definitely very frustrated about the negative perception. They have a comms problem, clearly. And so I think internally they're trying to spit out different ideas to try and fix it. And so this was, I guess, like a very impulsive decision they made to just buy...

[02:00:26] You know, like they were kind of pitching it to me as like, you know, we're going to help shape public opinion with TBPN and reach audiences first and bypass traditional media. But I don't think they really had an idea of what they wanted to do with it. It was one of those things where like everyone... Like there are these weird disconnects for people just like to the rest of the world. It's just like a completely silly thing to do.

[02:00:52] But I guess people get hive mind and decide they want to do it. I should be honest. I'm just jealous. They have 59,000 YouTube subscribers. This show alone has 63,000. Our network has more than a quarter of a million. We're not getting hundreds of millions of dollars. But on the other hand, you could buy us and you wouldn't guarantee positive coverage of open AI. So maybe that's what's going on. I mean, Joe Rogan's got 20 million subscribers.

[02:01:21] Lex Friedman has 5 million subscribers. I mean, it just seems like you bought... I just don't get it. How is monitoring the situation going? Is that a good thing for Andreessen Horowitz? It seems like the same idea. I didn't even know that they launched... That's NTS, right? Therein lies the problem. Nobody knows it exists. All right. Yes. And the pointy-headed one doesn't seem to care that much either. That's... Yeah.

[02:01:51] Well, a little introspection goes a long way, let me just say. All right. I just... That's a kind of a personal axe to grind. I shouldn't belabor that point. Let me see if there's any happy news. Any happy news at all. God, his head is so pointy. I know. I know. I know. I just... I interviewed him in 97, and he gave a great interview, I've got to say. But you couldn't get over the fact it's... Yeah, his head is really, really pointed.

[02:02:19] I constantly think it's kind of an internet meme sort of situation where people are editing the photos, but then you're confronted with the reality. No, no. When you see it... What a unique moment of... This may be a handicap. Because, you know, sometimes when a baby is born, the head gets squeezed quite a bit, and it is not unusual for it to be pointy. I'm not saying there's anything wrong with it. I'm just saying it's notable. It's just obviously pointy. It's a notable physical difference. And it's not something you can fix.

[02:02:47] I mean, you wouldn't want to put a board on the baby's head as they're growing to kind of fix that. I mean, I have ridges on the side of my skull. It's kind of a Mario-esque approach, you know? I think that every baby should be stomped on by an Italian plumber and his brother. I think that would fix it all. I confess. I'm sympathetic because I once shaved my head, as you well know, for charity, and I learned

[02:03:13] that I have a really horrific shaped noggin, and I don't want to make fun of anybody. Well, you're going to have to shave your head again when we do our second... ...when we do our next 24-hour Twitter live stream. She's lobbying hard. She really wants us to do it. Both of you guys are welcome to take a 30-minute slot because we're going to have to fill 24 hours. That's only 48 people we need to fill half an hour each. It's not that many. We can find 48 people easy.

[02:03:40] It's not hard. That's why skull is ugly. Yeah, you don't... I shaved my head once in college and I had a fucking skull, yeah. Did you really? Mm-hmm. And what prompted that? Did you have a charity? I had a blonde mohawk for a while throughout high school and college. And then at one point, I was like, what if I grew my hair out all the way because I wanted to...

[02:04:10] So you thought you'd start over. And so I figured, yeah, why not? But I wouldn't... So you look like Sinead O'Connor. Yeah. On retrospect, I mean, it looked great, but I would not recommend doing it in January in New York. Oh, good Lord. It's chilly. I don't really think about the fact that having any amount... Because I even still had kind of short hair at the time, but that still adds a level of warmth to your head that a shaved head does not. It's amazing, isn't it? Yeah, look how short my hair is.

[02:04:37] And yet when I didn't have any, I had to wear a hat. It was so cold. A lot of heat radiates. I'm sorry, Berber. You didn't know what you were getting into and I really apologize. We've been incredibly on topic for what the show normally is. We're actually on our best behavior. We're doing great hair. Yeah, you know, that's looking good. He does. He does. It's all over the place. And I'm sure an excellent shaped dome underneath it. I have no idea what my...

[02:05:04] I've never thought about what my head would look like if I shaved it all off. What's the shortest your hair's ever been? It's always been long, I feel like. Because I have a huge head, so... I also have a huge head. We all have huge heads. Every one of us on this show has a huge head. It's hard to find hats. There is a demand now in the Discord chat where our club members hang out. As you know, we treat our club members as family and they are saying they need pictures

[02:05:33] of you, Paris, with a mohawk and with a shaved head. So are there any? I mean, I can try and find some. Here are pictures of us with shaved heads. With cone heads? Yes. I mean, cone heads, what a great show. Well, they've been posting pictures of us with mustaches as well. The mustache ones were quite good. Yeah. Well, could you tell how old we were from those? I'm going to look back. Not looking good. Yeah. Yeah.

[02:06:02] Anyway, we thank you so much for being here, Berber. Read Berber's work. Really great stuff in the Wall Street Journal. A lot of exclusives. Oh, yeah. We look good with mustaches. Berber, you have a nice... That would fool anybody. That would make you... I should try. I like that one a lot, actually. Yeah. Can you grow facial hair? It doesn't go well. That's kind of a personal question. I'm so sorry to put you on blast. She asked me that on Wednesday, too. I just...

[02:06:28] You said you can, but you just choose not to because your wives hate it. She has... Paris has a real knack for the personal question, I guess. I like my... What else we got to do? My point stash there. I grew one during lockdown, and it was so annoying. We liked that. So happy to... You looked like Terry Thomas. You looked fantastic. Yes, exactly. I looked like a depraved Terry Thomas.

[02:06:54] But also, you know, you take a swig of tea, and it just... It stayed in there all day. I'll put a photo of that and the mohawk on Discord, but... You had a mohawk, too? Yeah. Yeah. Everybody post your mohawks in the chat. Half of this panel has had a mohawk. Unless, Berber, you've also had one. In this case, it's 75%. Oh, that wasn't a mohawk, Paris.

[02:07:22] I'm going to show you the back that has a mohawk. Well done. That's not a mohawk. I guess it's a faux hawk. It's a good look. That's a good look. Let's see. And this isn't quite bald, either. This is just... Okay. That's a shaved head. It's extremely short. Here, let me show you the back of... Oh, look at that. Paris, you have... I love this. Paris has readily to hand pictures of the back of her head. I've just gone on Instagram and scrolled back.

[02:07:51] Oh, okay. There we go. Oh, that is not a good look, Terry Thomas. No, it really wasn't a good look, but it was a... You look in pain. I apologize for those of you only listening. Actually, no. You're lucky. You're the lucky ones. Thank you, Berber, for being here. We appreciate it. Read Berber's work in the Wall Street Journal. You'll find Paris Martineau at Consumer Reports, where she's doing food safety.

[02:08:18] Yes, it was she who exposed the radioactive shrimp scandal, the lead in your protein powder shocker, and there is soon to be more. Food safety is her passion. Thank you, Paris. We'll see you on Wednesday on Intelligent Machines. And thank you, Ian Thompson. I called you Thompson, didn't I? Yes, I was going to say anything, but my father would hunt you down. I apologize.

[02:08:48] Thank you, Ian. Of course, read Ian's View from the Valley. Subscribe, and that way you'll get it automatically. So it's not just intended for Brits. Anybody would want to read this. No, it's a view from a Brit in America, and this is a strange and silly place at times. It is indeed. Ian Thompson, thank you so much. Thanks to all of you for joining us. A special thanks to our club members who make this show possible.

[02:09:13] If you believe, as I do, in the importance of independent journalism covering technology, especially podcasting, your support makes all the difference to us. It covers a great portion, as much as 30% of our overall operating expenses. It allows us to do this show and all the other shows we do, plus special programming for our club members. We give you access to a special Discord only for club members, which makes it a great place to hang out.

[02:09:40] You also get ad-free versions of all the shows for just $10 a month. If you're not a member yet, twit.tv slash club twit. We do this show every Sunday afternoon, 11 a.m. Hawaii time, 2 p.m. Pacific time, 5 p.m. East Coast time, 2100 UTC. You can watch it live if you're there at that time in the club twit Discord or YouTube, Twitch, X.com, Facebook, LinkedIn, and Kik. We stream on all those platforms.

[02:10:08] After the fact, on-demand versions of the show available at the website, twit.tv. You can also subscribe in your favorite podcast client. You'll get it that way automatically the minute it's available. There's a YouTube channel, too, with, as you can see, 63,000 subscribers. I think that's worth at least $100 million, don't you? I feel like that's not too much to ask Sam Alton. Let's see. I guess that's all I have to say, except that, as I have said, for 21 years now at the end

[02:10:37] of every show. Thanks for joining us. We'll see you next week. Another twit is in the can. Bye-bye. This is amazing.