Microsoft Announces Its Most Powerful Surface at Build 2026 - DTNS 5281

0:00This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade-in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One.

0:33If you've used Babbel, you would. Babbel's conversation-based technique teaches you useful words and phrases to get you speaking quickly about the things you actually talk about in the real world. With lessons handcrafted by over 200 language experts and voiced by real native speakers, Babbel is like having a private tutor in your pocket. Start speaking with Babbel today. Get up to 55 % off your Babbel subscription right now at babbel.com slash acast. Spelled B-A-B-B-E-L dot com slash acast. Rules and restrictions may apply. When you're a maintenance engineer in a beverage manufacturing plant, you keep production lines moving and quality on track because there is no room for slowdowns.

1:18With Granger's vast selection of high-quality motors, sensors, belts, and hard-to-find parts, you can get what you need fast and all in one place so nothing gets in the way of getting the job done. Call 1-800-GRANGER, click granger.com, or just stop by. Granger, for the ones who get it done.

1:40I'll be honest, nothing tests my patience quite like sitting on hold, listening to the same four bars of music on repeat. Turns out, all that time adds up. The average person spends over 40 days listening to that dreaded hold music. 40 days! This episode is sponsored by Parloa, the AI platform built to make that a thing of the past. Parloa's AI agents are available 24 hours a day, 7 days a week, fluent in any language, and smart enough to remember every customer across every interaction, because no one should have to explain themselves twice. From payment processing, to roadside assistance, flight rebookings, to appointment scheduling, Parloa delivers the kind of customer experience that turns a one-time buyer into a customer for life.

2:23See Parloa's AI agents in action at parloa.com. That's P-A-R-L-O-A dot com.

2:35This is the Daily Tech News for Tuesday, June 2nd, 2026. We tell you what you need to know, give you the important context, and help each other understand. Ah, and it is Microsoft Build, Microsoft's developers conference. So we're going to talk about how they're trying to convince you to choose Microsoft as your AI co-pilot. It's so weird that they would start with AI at a developer conference in the year 2026. Yeah. What is this? I'm Jason Howell. 2026. I'm Tom Merritt. Let's start with what you need to know with a big story. And yes, Microsoft has kicked off its Build Developer Conference today.

3:15You know, on Tuesday, it's all happening. In fact, it's happening while we record this. So, you know, if there's some big major announcement that happens, you know, we've caught a little bit of it that we're going to tell you about, but there's plenty more and some of it we might miss. So we'll follow up on that tomorrow. We'll never miss, Jason. Don't tell them we're going to miss anything. We never miss. It's always exact. We're wizards. We arrive exactly when we meet. We know the future. Yeah. No, if there is some more important stuff that happens later, we'll get to it tomorrow. But we have the biggest announcements that came right off the top.

3:49And Microsoft Build is very developer-y. If you're a developer, there's lots of good stuff in there. But unlike Apple or Google, there's not as much for the mainstream audience. But a couple of things here I think are worth paying attention to. First of all, new hardware for developers called the Surface RTX Spark DevBox. very similar to what nvidia announced from itself yesterday because it uses nvidia's new arm-based rtx spark right it's just this one's from microsoft and is called a surface instead of being a dgx and from nvidia uh kind of i don't know what you thought did you think it looked like an xbox series s because seems like a lot of people did it's got a really interesting look i like i can't tell the scale of it um there you know what it reminds me of and probably no one else is early on when google was first announcing their android tv boxes they had this small little developer kit oh really yeah yeah and it looked kind of like this but probably i'm guessing smaller it's hard for me to get this this one's a little bigger than that was i think yeah yeah but it's not huge uh and It looks like it would hurt you.

4:58Well, it's because it's got that aluminum chassis. And that is also a heat sink. So that's kind of a nifty part of the design. This is 190 watt thermal envelope. So it can actually really, you know, power, powers some stuff away. It is quite capable of using up your electricity. 128 gigabytes of unified RAM, capable of 120 billion parameters locally. Microsoft is trying out a new buzzword, Jason, called unmetered intelligence. That's their way of referring to it's processed locally, so we can't charge you for it because it's not coming from our cloud. Well, that makes sense. And that sounds like something that I would kind of want as we do more and more of our work on these, you know, using these various models and everything.

5:50that can really add up. And I'm sure that is going to be and appears to be a major push as we go forward. It's like, hey, maybe unbuckle your reliance on the cloud, go local, and go whole hog. Yeah. And we don't know how much you're going to pay for this yet, but it's not going to be cheap. However, Andrew Hill, corporate vice president of Surface for Microsoft, told The Verge, the Surface RTX Spark dev box ships with Windows 11 Pro pre-configured at the image level for the developer. So you're in your dark theme, taskbar is simplified, widgets removed, do not disturb is on when you turn it on, developer mode is enabled, PowerShell 7 is the default shell.

6:37So this really is not meant for you or I, Jason. It's meant for Huenzweidao. It's meant for somebody if she were doing Microsoft development to be a developer, be able to do inferencing right on the device. This will come later this year from the Microsoft store. You can get on the waitlist now. Sachin Adela joked that he is also on the waitlist that he doesn't have one of these yet, which makes me think, oh, so you haven't actually made any yet because Sachin would have one. Something tells me he would have access to it. That's so interesting. So you've got a couple of prototypes down in the lab, but you haven't manufactured any of these yet.

7:13And as I said, no price on that. Also, the other big announcement has to do with the operating system. Windows chief Pavan Davaluri told The Verge, we've optimized the Windows 11 experience for developers, bringing frequently used command line utilities, a familiar comfort shell, faster setup experience, a built-in way to create and interact with Linux containers on Windows, as well as something they are calling the Intelligent Terminal. which is experimental. But the takeaway of this for folks who care is they have optimized Windows with a lot of Linux access. So for example, Core Utils for Windows is an adaptation of UUtils, the open source project.

7:59It's a cross-platform re-implementation of GNU Core Utils in Rust. That means something to people who need to know that. But you can think of it basically as a Linux-like command line that runs natively on Windows. And speaking of which, the Windows subsystem for Linux, which was open source last year, now has WSL containers. These are CLI, command line interface and API for WSL containers that lets you run Linux containers inside native Windows apps. So you can have a little WSL container, put it inside a Windows app. And then that intelligent terminal lets you give context to your favorite agents via agent communication protocol or ACP.

8:43So you can stay in the terminal and do your querying and your debugging complete tasks. General availability of the Windows developer configurations is now coming. A quick way to set up new machines with developer optimized settings. So you can image those for your developers. That's all the kind of thing it's like for most of us, probably not a huge deal, But for developers, this is a very helpful and useful tweak to the operating system, several tweaks to the operating system. The other thing that I want to point out is the new, I think it was VentureBeat. Yeah, VentureBeat pointed out Microsoft Execution Containers, or MXC.

9:24You might not see this written up as much, but this is an execution layer built into that Windows subsystem for Linux, WSL. It's also built into Windows, and it lets you declare exactly what an agent can and cannot access. So it's basically a sandbox for agents that gives the agent an ID, either a local ID or you can provision it through Microsoft Entra. And OpenAI, NVIDIA, Manus, Nu Research, maker of the Hermes agent, or Hermes agent, N-O-U-S research, and OpenClaw, the open source project, are all integrating MXC. And I feel like this is going to be a topic for AI inside for you, Jason, is this ability to give an agent a sandbox where you can say, go nuts, you don't have permission at the operating system level to do anything like delete files that I don't want you to do.

10:18Yeah, and this is reminding me of actually an interview that we had back in November with Charles LaManna from Microsoft. at the time talking about Agent 365 and this idea of giving your agents NID, like thinking of your agents, like, you know, you often hear, like, think of your agents as an employee or someone who you're tasking with a job or whatever, kind of going a step further. And in the Microsoft world, I wonder if these two things are connected, actually, giving them that identity so that when those agents appear inside of the tools that you're using, you're kind of also kind of working with them as if they are an employee, even though it is an AI agent and you're kind of seeing them on that same level.

11:00So yes, this seems pretty important. I mean, as a user of these systems, you want to know when that entity that is there that maybe presents itself as a worker or something is an agent, right? Like we're going to treat those things differently. So that's good. Yeah. Well, and I think it's the security level that's important here too. And not so much security as in attackers and malware and prompt injection, although it's a defense against that. Security in that I don't trust the agent to do certain things, but I do trust it to do other things. If I just want to have it do some low-level coding, I just give it access to the coding tools and nothing else.

11:42That way it can't go in and suddenly turn off things in AWS that causes an outage. Uh, you know, it can't, it can't get out of, uh, get out of its head, the idea that it needs to do something that it shouldn't do. It just won't have the permissions to that. And if you're treating it like an employee, that makes sense. If you and I are sharing a machine, you and I are going to have different permissions because I shouldn't be able to access your personal files. You shouldn't be able to access mine, but there are certain shared resources we do want to access. And that's true of your agent as well.

12:14I think. Yeah. This is a story that's actually going to come up a little bit later in the show as well related to meta but yeah i i kind of think of it as like driving a giving someone my keys to the car it's like if i give you the keys to the car and then tell you don't drive my car i shouldn't be entirely unsurprised when in the very off chance that it happens you decide to drive my car anyways if i just don't give you those keys you're probably not going to drive my car but you want the agent to drive your car so let me let me go one further with you, this is valet mode. You want the agent to drive your car, but you don't want it to be able to do everything.

12:53You don't want to be able to change all the settings. You don't want to take it on a joyride. Exactly. You just want it to park it. That's all I love. I love that. MXC is valet mode for agents. And then this final bit was not at build, but it's somewhat related. Copilot just switched, and they had said months ago they were going to do this, but they switched from request-based billing to usage-based billing, which has quite a bit increased the amount some developers are paying, especially if they weren't properly estimating the change, which Microsoft did give you tools to estimate. Previously, a short answer to a quick question or a multi-hour autonomous coding session cost the same if you only put in one prompt that was about the same length.

13:39Now, the rate is charged by the number of input and output tokens used. So that multi-hour autonomous coding session is going to cost you a lot because it's outputting a bunch of tokens. This is also going to vary depending on which LLN you're calling in Copilot because you can use different ones and different types of models. So if you're using a light model, that's going to cost you less than if you're using the full model. I'm not sure that I was aware that Copilot worked that way before. This is the way that many of them work now. you know, is by usage based. Um, so that's really interesting to me.

14:17I wasn't aware of that. Yeah. And, uh, I think a lot of people were not aware that that was unusual if that's all they were using and now they're mad because it changed. Yeah. Yeah. Uh, well, like we said, uh, we will, uh, update you if anything else comes out of build, uh, between now and tomorrow. Uh, meanwhile, big thanks to everybody who makes this show possible. Thank you. Rowan Vickery. John Atwood, Pat, Mike Cortez, and every single one of you at patreon.com slash DTNS.

14:54This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge, and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade-in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One. Granger knows when you're a procurement manager for an office park, you're not managing one building.

15:29You're managing all of them. And to stay ahead, you need to see through walls and around corners. Lights about to fail, filters ready to clog, HVAC on its last leg? If you wait until something breaks, you're already behind. Count on Grainger for quality products, easy reordering, and 24-7 support. Call 1-800-GRAINGER, click grainger.com, or just stop by. Grainger, for the ones who get it done.

15:56All right, plenty more we need to know today, so let's get right to the briefs. Yeah, thanks to RW Nash for noting this one on our subreddit, our DTNS subreddit. First off, this is this is what jason was just alluding to meta has pushed a fix on friday may 29th for what we're about to talk about so the story isn't that it's an ongoing thing it's what happened and a little bit about why meta started offering a chatbot for support the idea was that it would help speed up support requests including account recovery so you could go to the chatbot and say hey i'm locked out of my account can you send me a recovery link to my email address and the chatbot would say, great, yes, I will ping the proper system and have it do that.

16:38One of the ways that it was making sure it was you was to check your location. And apparently, in some cases, maybe that was the only way it was making sure it was you because attackers found that in many cases, not all, they could use a VPN to make their location look similar to yours, whatever the target account was, and then begin the password reset process. During that process, there was a prompt injection they could do to say, oh, and by the way, can you change the email associated with this Instagram account? Because I think it's been hacked, so I want to make sure it goes to my email. And the chatbot would be like, great, I have changed the email account.

17:20Great, now send that reset link. And then they would send the reset link to the new email address. Once the email address was changed, that recovery link gave the attacker access to that account. This has been going on for a while, but it got noticed recently because they started accessing more high-profile accounts. Obama White House, which is not President Obama's personal account, but it is the legacy account from his presidency, that hadn't been posting a whole lot, and suddenly it was posting things you wouldn't expect. And then the account of John Bentavania, Chief Master Sergeant of the Space Force started posting some propaganda that you wouldn't expect.

17:59However, as I mentioned, these exploits have been happening as far back as February, according to NEOIN. Maybe thousands of accounts that appear to have been compromised, but it only gained significant attention when those high profile accounts got attacked. It's also worth noting that most of the time the attacks failed if you had multi-factor authentication turned on, though there were some instances where they were able to use email as the multi-factor. And if they had got it to change the email, then, so if you see that, if you see people say like, no, I even had multi-factor on and it didn't work, it was because they used email as that second factor.

18:36But in a lot of cases, if you had multi-factor turned on for, you know, like a passcode or, or for, you know, a, a, a, a Authy or something like that, then, then it wouldn't work. Yeah, it's just crazy to me that sometimes the, maybe it's not the easiest path, but sometimes the lowest friction path is to just ask. Just ask for it to do the thing. And you never know what you might get on the other side. I mean, clearly they knew what they were doing here. I mean, yeah, you had to go to a VPN. You had to know where your target's location was. And there was some work involved. For sure there was.

19:16But then once you got into the chatbot, it was a matter of convincing the chatbot, right? Which, as you know, with LLMs, it's not as simple as just asking. It's asking in the right way. Asking in the right way or asking in a really strange roundabout way that most people wouldn't even consider. But when you do it, yes, totally. These things are definitely susceptible to that, clearly. Yeah. So they say they have mitigated it and that this is no longer possible. and yes, it is perfectly reasonable for you to say, well, it should have never been possible, but there you go. So what's interesting also to me is the two-factor thing via email because often people will say, you will hear advice saying, do two-factor, but don't do it in the way where you get texted the code because someone can intercept that through some SIM interaction and everything.

20:11So we shouldn't also do that via email because then something like this could happen. So is the best two-factor way either a hardware key or an app? Email is still the best way over SMS. Absolutely. This is a vulnerability that is unusual in that it was allowing attackers to change your email. And there's no defense against that. It would be as if the attacker... You could also have an attack conceivably, right? Where somebody is able to say, hey, I need to change my multi-factor account. I'm going to use a different... I'm going to use Authy now instead of Google Authenticator. And if you could trick the chatbot into saying, great, let's change that.

20:50Well, then suddenly the multi-factor protection isn't any good. So it's not that email is a weak form of multi-factor protection. It was that this system did not protect your email address, which is the crown jewels, which is almost unforgivable, but not a common thing that is going to happen. Whereas SMS, it's less likely than it used to be, but click jacking and things can make it so that you can call the cellular company and say, hey, switch my phone number from that device to this device, which is actually similar to what they were doing with email here. It's just usually you can't do that with email.

21:30They weren't getting into your email. The account provider shouldn't have changed the email on the account, right? So the email wasn't the vulnerability. It was which email is associated with the account. Yeah, you're right. You're right. You're right. Yeah. It's all done on the meta side anyway. Right. So it has nothing to do with your email address or your email inbox. Whereas with the phone number thing, with the text message, they're actually getting a hold of your phone number. It would be as if they got into your email. Yeah. Interesting. Well, speaking of security topics, Anthropic is expanding access to Claude Mythos.

22:05They are inviting more than no, sorry, 150 more organizations into Project Glasswing. That is on top of the 50 partners that were first announced back in April. The new partners cover a number of countries, 15 of them cover a number of high risk sectors. So you got power, water, telecom, health care and anthropic shared, quote, what each partner has in common is that a successful attack on their code base could be catastrophic. catastrophic, end quote. So they're basically saying it would impact more than 100 million people if it were to happen in this particular place. And so that's why they're doing this.

22:46Anthropics' own security requirements, by the way, must be met by each of them in order to actually use Mythos. Financial Times is reporting a few of those organizations to be Okta, Samsung, NATO, EU's cybersecurity agency, ANISA, and SK Telecom, just to name a few. Anthropik still plans to keep Mythos out of general release until it can add even stronger safeguards. They also say that they expect competing companies to develop models with similar capabilities, which is not that surprising. They're all kind of - I mean, OpenAI already has one. So, oh, guess what? You're right. And, you know, so that's driving their own kind of continued and rapid development of safeguards for the model.

23:30And all of this happens the day after Anthropik's announcement that it filed confidentially for an IPO, an initial public offering, putting Anthropik ahead of OpenAI, ahead of SpaceX, both of which are expected to announce their own IPO listings. That could happen as soon as this week. So, you know, yet another big week for Anthropik. Like, Anthropic just continues to stay in the news with these big announcements. And Mythos continues to drive awareness for what they're doing. Everybody wants a piece of Mythos. SpaceX has filed its paperwork, though, right, for its IPO? Is that right? Yeah. They did that last week.

24:08Oh, that's right. No, that's right. The news was last week. My apologies. Right. But they're still ahead of OpenAI. And it may be that Anthropic filed more paperwork than SpaceX did on this one. But yeah, it's definitely part of the hot IPO summer that is coming to technology. And what interesting timing to get Mythos back in the news, right, as you're filing your IPO and hoping that people will be interested in it. I also still think that while Anthropic definitely does have concerns about Mythos being misused, also has concerns about compute. And this certainly doesn't hurt to limit how this is done and also move it into a different region.

24:55But yeah, clearly OpenAI is doing this already. Other companies are doing this as well. and Anthropic is going to continue to benefit as being perceived as the more reliable version of Bugfinder right now because it is an impressive tool. Yeah, and like I said, it continues to capitalize on this, I don't know, this in the zeitgeist momentum that it has. Mythos is just flying around all over the place. Everybody wants a piece of it, is very curious about what they're missing. I don't know. It's going to be interesting to see what it actually means if and when they get to a point to where they're like, all right, we're nearing some sort of a public release.

25:40What does that even look like for a model like this? My guess is they have a more advanced model that the people in Project Glasswing get. Yeah. And then all of us get access to the older model. I'm curious if we ever have a story of the organization that wasn't one of the 150. It was the 151st on the list and didn't get access, and then it gets hacked because somebody has found an open source version of a bug creator that it used to develop a bug. Yeah, it is quite a dynamic situation with security these days. Speaking of mitigating against dangers, Chinese EV maker BYD announced it will cover liability for crashes with its cars for Chinese drivers.

26:29This is only in mainland China. If you choose to use Urban Navigate on autopilot. It's a self-driving feature on the new God's Eye 5.0 system. So if you put your faith in the car's hands and something goes wrong, BYD will be on the hook for being responsible for it. Guarantee covers repairs to the vehicle, third-party property damage, personal injury, no payout cap, no need for separate insurance, and no premium level hikes if you do have a crash because it's all on them, not on your personal insurance. coverage applies for the first year of ownership after you get delivery, as well as for existing owners if you upgrade to that God's Eye 5.0 system.

27:13Yeah. Super interesting. I mean, that's a bold move. Tesla offloads responsibility to the driver. I think they pretty much all do for the most part. These are different kinds of systems, though. Tesla's is not a system where you are allowed to not pay attention. So it makes sense that they would say, no, you're on the hook because you're supposed to be paying attention. This is just an assistant. God's eye is like, we'll drive for you. And so this is a level four system. You can lean back. And then if something bad does happen, we'll take responsibility for it. Wow. Okay. All right. That does change things.

27:47Because when I was reading through this, I was like, my general feeling about this, where does the responsibility lie when you're talking about agents or when you're talking about self-driving and everything is like I'm the operator at the beginning of the day so I'm the person that chooses to take on the risk or the responsibility if I say yes do this thing for me versus no don't and so in general I've kind of felt like you know does the vehicle maker have some sort of responsibility if they have like software that just makes a really horrible bad decision repeatedly or whatever I yeah sure of course but I think as a driver it's kind of up to me to say, you know what?

28:25I'm wrong. I just double check this and God's eye is not level four. It is level two plus, which is basically the same as Tesla's autopilot. So I take it back. They are going a little bit further than Tesla would go. And that makes it more impressive now that I've doubled. So then is this meant to, well, I mean, clearly it's meant to get people more, you know, more trusting of the BYD systems and to buy into them and everything. But I also, I'm very curious to know the impact that this has on other manufacturers. Does it kind of push or motivate them to match that? Is that going to be something new?

29:04Especially in China, right? Because this is only China, but Tesla's operating in China. Is Tesla's autopilot going to have to do this in China as well if they want to maintain competition? Very interesting. Yeah, yeah. Sorry for getting that wrong. I saw God's eye and I know that BYD is developing level four, but it hasn't deployed it yet, obviously, or that would be bigger news, I guess. Joining the conversation in our Discord to say, Tom, why did you get that wrong? By linking your Patreon account. We've been organizing meetups around the globe. We've got one coming in July in London and also in Munich.

29:39we're in there just talking about Linux subsystems in one chat. So go join the Discord. You can link your Patreon account to get access at patreon.com slash DTNS.

29:58This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One. Grainger knows when you're a procurement manager for an office park, you're not managing one building.

30:33You're managing all of them. And to stay ahead, you need to see through walls and around corners. Lights about to fail, filters ready to clog, HVAC on its last leg. If you wait until something breaks, you're already behind. Count on Grainger for quality products, easy reordering, and 24-7 support. Call 1-800-GRAINGER, click grainger.com, or just stop by. Grainger, for the ones who get it done.

31:00All right, we got a gaggle of quick headlines to throw your way to make you look smarter at the dinner table tonight. Yeah, Motorola announcing the Edge 2026, a super thin mid-range phone for$600. Has 120Hz AMOLED display, 50MP triple camera system. Runs on the MediaTek Dimensity 7450 chip and has a 5 ,000 mAh battery. You can get it in the U.S. starting June 11th. I wonder if that battery is silicon carbon. I didn't check on that. They've been doing that in the U.S., so I wouldn't be surprised. Meta is rolling out its 13-plus teen safety setting globally as default across Instagram, Facebook, and Messenger.

31:42And they also shared that 9 out of 10 teens stayed on that default setting during their trial run. So there's that. meta is also testing a new series feature for reels on instagram and facebook so that you can bundle episodic videos into a dedicated hub for sequential viewing so basically you could turn your reels into uh mini series yeah which is kind of those micro series are pretty hot right now yeah they are that's cool sony's data own data shows first party playstation game sales have fallen from a 58.4 million peak in fiscal year 2020 to 32.1 million in 2025. As you can see, that's a pretty sizable drop in the performance of their big budget exclusives.

32:28So definitely notable. Some interesting news from Blue Origin and encouraging if you were saddened by the explosion of their rocket. CEO Dave Limp is assuring that New Glenn will return to flight by the end of the year. This is after the May 28th explosion at Cape Canaveral. NASA administrators, however, are saying that repairs might drag on until 2028 for the launch pad. So, you know, the CEO is being a little more optimistic than NASA. Yes, indeed. MSI's CLAW 8X AI Plus is a powerful Intel Arc G3 extreme powered gaming handheld. It was actually shown off this week at Computex and it's going to set you back a pretty sizable amount,$1 ,500 for that handheld device.

33:13Also at Computex, Asus launched the ExpertBook B5 Flip G2. That is a 2.9-pound, 360-degree touchscreen convertible for business users. That alongside a refreshed ZenBook 14 lineup that has Intel, AMD, Ryzen AI, and Snapdragon X options. Whatever chip you want, they'll put it in a ZenBook. Whatever you want, you get it. A group of mathematicians and researchers have issued formal guidelines calling for transparency, peer review, and public interest safeguards so that powerful AI systems like those behind OpenAI's recent unit distance claim don't actually end up overwhelming human mathematics. Yeah, this is like the patch problem or the bug finder problem.

33:56It's exactly that. Mathematics. This past weekend, several Dashlane password manager customers were locked out of their accounts as attackers were trying brute forward access. Now, Dashlane saying the attacks were successful on about 20 accounts. Those 20 accounts gave access to the encrypted vault to the attackers, though it did not expose the master password. So whether the attackers can access your password vault will depend on how strong the master password was. But you should probably change all those passwords anyway. Yeah, it probably should. SK Hynix says it will double its memory chip wafer capacity over the next five years in an effort to ease global memory chip shortages.

34:40The company expects those shortages to last until at least 2030. So that's something they're doing about it. Takes a while to build new capacity, but it's always good to hear new capacity is coming. Withings launched a cheaper version of the BodyFit. This is its scale, a$280 mid-range version of the$600 BodyScan smart scale. This one targets GLP-1 users, You know, Zempick, Wigowi, all that stuff. If you need segment-level tracking of muscle versus fat loss during weight reduction, you can do it for cheaper with the body fit. Not bad. NVIDIA's new H200 AI chips being actively sought by at least seven Chinese universities that support the country's military and defense industry.

35:25This is according to documents reviewed by Bloomberg. And Uber continues to announce new partnerships with autonomous car companies. This is where Uber provides the hailing system so you can get a ride. Uh, they're going to launch a pilot in Madrid with bookable rides coming later this year. That one is in partnership with China's WeRide for the software and moves Avomo for the fleet management. And then Uber is going to partner with Israel's auto brains. That's an interesting company because they don't use all the sensors that a Waymo does. They use the onboard sensors that your regular car has to be able to do autonomous driving.

35:59And it runs on NVIDIA's drive Hyperion platform that is going to launch in Munich. And then finally, Alphabet has its$80 billion in equity raise. That includes, by the way,$10 billion investment from Berkshire Hathaway as they kind of focus more of their spending on, you guessed it, their artificial intelligence plans. Alphabet borrowing a lot of money. They also have a lot of money. A lot of money. Yeah. A lot of AI. Yeah. We end every episode of DTNS with some shared perspectives. Today, Andrew has thoughts on the value of cheaper machines. Yeah, so that's one thing we've been seeing out of Computex, like the Dell XPS 13, really taking on the MacBook Neo.

36:45And then, of course, Microsoft's Surface laptop, based on the new NVIDIA chip, taking off on the MacBook Pro. Andrew writes, I'm in the midst of rehabbing a 2012 MacBook Pro into a terminal-only box. I can hook it up to an AI CLI, that's a command line interface, and do a lot of work on it despite its age. And the lightweight setup lets me use it unplugged for a couple of hours at least. Friday's show talking about the value of cheaper machines prompted this thought. Even though they can't do on-device AI, any cheap machine that will let you get to a terminal can run the AI CLIs. They all seem very lightweight.

37:22And you could do a lot with those. Yeah. I mean, it's just happening like in the cloud. and you're just passing your information off. Be careful of your token charges, especially if you're using Microsoft. See earlier in the show when we were talking about Google, which is cloud, but yes. No, that's good. 2012, that's a 14-year-old laptop putting it to good use. Pretty cool. What are you thinking about? You got a 14-year-old laptop you're putting to good use? Something else like that? Share it with us. Feedback at DailyTechNewsShow.com. Yes, please. Thank you again, Andrew, for contributing to today's show.

37:53And thank you out there for being along for Daily Tech News Show. you can keep us in business. All you got to do is become a patron. That's patreon.com slash DTNS. We'll see you tomorrow.

38:10The DTNS family of podcasts, helping each other understand. Diamond Club hopes you have enjoyed this program.

38:22This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One. When you're a maintenance engineer in a beverage manufacturing plant, you keep production lines moving and quality on track because there is no room for slowdowns.

39:01with Grainger's vast selection of high quality motors, sensors, belts, and hard to find parts. You can get what you need fast and all in one place. So nothing gets in the way of getting the job done. Call 1-800-GRAINGER, click grainger.com, or just stop by. Grainger, for the ones who get it done.