Inside Google DeepMind's Genie 3 world model - DTNSB 5075

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. Your team just added its 67th AI tool and also your 67th security blind spot. The good news? The Vanta agent works like a GRC engineer in the background, finding every app your team uses, scoring the risk, and drafting fixes for you.

0:43Vanta is the platform used by over 16 ,000 fast-moving companies like Ramp, Cursor, and Harvey, who are shaping the future with AI and staying ahead of AI risk. Get started at Vanta.com. We all belong outside. We're drawn to nature. Whether it's the recorded sounds of the ocean we doze off to or the succulents that adorn our homes, nature makes all of our lives, well, better. Despite all this, we often go about our busy lives removed from it. But the outdoors is closer than we realize. With AllTrails, you can discover trails nearby and explore confidently with offline maps and on-trail navigation.

1:22Download the free app today and make the most of your summer with AllTrails. This episode is sponsored by Morning Brew Daily. If you're anything like me, you want to stay informed about what's happening in business and finance, but you also don't have an hour every morning to sift through headlines. Morning Brew Daily is the go-to business news podcast for millions of people. It's a daily business news podcast that's a quick, fun, and easy way to start your day. Whether it's the latest in tech, markets, or the trends shaping the economy, the show breaks it all down in a way that's clear, engaging, and actually enjoyable to listen to.

1:59Hosts Neil Freiman and Toby Howell make the show feel like you're catching up with two friends, while giving you the context you need to feel smarter and more confident throughout your day. New episodes drop every weekday wherever you get your podcasts. Tune into Morning Brew Daily wherever you get your podcasts.

2:18This is the Daily Tech News for Tuesday, August 5th, 2025. We tell you what you need to know, follow up in the context of those stories, and help each other understand. Oh, and today we got a good one. Ryan Whitlam from Ars Technica joining us to talk about Google's DeepMind Genie 3 world model. And Cloudflare says perplexity crawlers are sidestepping site restrictions. Hey, don't do that. Don't do that. I'm Jason Howell. I'm Tom Merritt. Let's start with what you need to know with the big story. All right, so Google DeepMind has announced Genie 3. This is an updated world model that can do some really cool things.

3:01It can generate entire virtual and interactive environments. It's all done in real time, has the ability to dynamically change and shift based on requests that can be put in in real time. And Ryan Witwam is senior technology reporter at Ars Technica, wrote a fantastic piece all about Genie 3 and is here to talk to us about it. Welcome to the show, Ryan. It's good to see you again. Hey, good to see you too. Yeah, awesome. Let's dive right in. So I guess tell, why don't we start with kind of like what makes this so special? Obviously, this is the third generation Genie 3. what are the big advancements that are going on here and what really stands out to you in this announcement what gets you excited about this so i mean i guess the most exciting thing is that it's so much better than genie 2 which they revealed six or seven months ago uh and like the comparisons are pretty staggeringly different uh the visual fidelity is is much much higher it runs at 720p 24 frames per second and it happens in real time so you can actually interact with this essentially it's an ai generated video that's just happening on the fly and you can interact with it um and it's it's you know it's it's sort of uh it feels like a video game that you just sort of invent you know as you go um and it's uh it's it's it's not just for for games Like it's tempting to think it's like, oh, this is just like going to be the next generation of how video games work.

4:34But DeepMind talks about this mainly as a research tool, because, you know, while you can interact with the video, like with a keyboard, you can also have an AI agent doing it. And this is a good way to train AI because video games have sort of a goal and you can measure progress. So it's a way that you can you can train an AI agent. um and and you know and because this is happening live and it and it's generative ai you can make it do whatever you want uh deep mind has this this uh thing now they call promptable events so while while you're creating this video you can just say i would now like this other thing to happen i would like the conditions of the video to change i would like to add another character or like you know uh change the way the world works man in the chicken suit literally ran through one the video.

5:16So the problem though is that these are just, these are demo videos. We're seeing what I assume is a very impressive application of this technology, but none of us get to use it. Even if you're paying for Google's$250 a month AI Ultra subscription, you do not have access to this yet. Gini3 is just a research tool for DeepMind. And they sort of, they talk about like, eventually they would like this to be something that people can have, but it's not really available to any of us yet, probably because it is very, very expensive. And DeepMind neglected to elaborate on how expensive it is, but just based on the fact that even if you're paying for the most expensive AI subscription, you only get a couple of eight second VO3 videos per day.

6:06And they talk about how how like the the quality level is very similar to what you get with vo3 so uh we're basically we're talking about an ai that is creating video on the fly real time and it has a couple minutes of memory sort of like the context window in a chatbot you know if you talk to a chatbot for too long it runs out of context it forgets the beginning of the conversation uh genie two only had a few seconds of memory genie three has a few minutes and if you're talking about making a video that is a few minutes long, uh, that's gotta, I mean, if they were charging a person for that, that's thousands of dollars and who, who even knows how much it costs Google to make probably more than that.

6:43Uh, so there's, there's, there's no access for, for those of us who are just sort of in awe of, of these, these demo videos. Uh, they said they're going to roll it out to, you know, trusted testers, which is a phrase we hear a lot with these AI models. Uh, and, you know, and they'll get feedback and maybe, maybe, you know, Genie 4 will happen in a few more months and, you know, and maybe we'll get that one. I don't know. But if we do, it's probably still going to be extraordinarily limited just because all this stuff is so, so expensive. Yeah. I mean, it seems like the research they need to do is figure out how to make it cheaper or who would be willing to pay that much to use it, right?

7:16Is there some enterprise level system that this can help? Yeah. I mean, there are definitely, I think with all this, this AI video stuff, there are definitely applications. There are industries where, you know they're paying human beings a lot of money to create video content and it you know at the end of the day it might be cheaper to just pay you know you know objectively a lot of money for an ai to do that but you're not paying humans um you know and a lot of you know and people in the games industry know there was a lot of backlash when genie 2 came out and people like oh this is gonna be great for game design and a lot of designers like that's not really the way you know things work that's not that's not how we approach you know level design but i have to think that even just being able to give this model some text or photo and say, okay, create a world out of this, like that's got to be at least useful for brainstorming or to like, you know, analyze an idea a little bit more.

8:04It's sort of like the way a lot of us use text-based models to just to brainstorm things. But I mean, there is definitely the danger that people who don't really understand game design are going to look at this and say, oh, we can just fire half of the designers and we'll just have AIs that do this. And, you know, and that's, that could lead to, you know, some, some bad outcomes in the quality of what you know comes out the other end but um i mean i think like right now google seems mainly interested in using this to to improve ai agents which is where a lot of the a lot of these companies are looking right now to create more more capable ais that you can task with things and let them go do do their thing um you know because there's there's only so much training data in the world but with something like genie 3 you can just keep creating new worlds and training AI to do different things.

8:47Yeah. An endless number of scenarios for it. That makes sense. Yeah. Yeah. I think that's, what's really interesting to me about this kind of a couple of different directions is, you know, the, just the term world model when I hear and read world model and granted, I'm, I'm constantly learning about this stuff. So, you know, I get curious and I don't understand it until I do. But when I hear world model, the question that always comes up for me is, does world model define the world as we know it here? So in other words, this is a model that adheres to the physics of reality that we live and experience here on planet Earth, let's say, or is it just world model is to define whatever the world is that has been created for this particular system to live within?

9:35Does that make sense? I think it's mostly the latter. But that said, Google does talk a lot about how its video models adhere to real physics, that they can simulate things that look right. And I think that's mostly true. I mean, of the AI video models I've seen, the video stuff, and I guess the Genie demos seem to do that better. but there's still an uncanny valley in the way things move, especially people. The team talked about one of the... Essentially, it's a hallucination. This is still generative AI. It makes mistakes. One of the things they've seen go wrong most often is that people don't move right.

10:15Sometimes they look like they're walking backwards. In some of the demos, you can see things sort of glide along surfaces in ways that don't look quite right. Like they're not moving like something should move when it's, you know, walking or crawling or whatever the thing is doing. Yeah. So and then I think the other aspect of this that's fascinating to me is this whole idea of like synthetic data generation. Like it's easy and you said this, but it's easy to look at something like this and immediately jump to the kind of games application. Oh, wow. Instead of needing a team to create this entire world, it's just created on the fly.

10:54And who knows whether it actually creates an enjoyable world, but it's creating something that didn't exist, that somebody didn't specifically direct. And you mentioned the problem of just the sheer fact that there's only so much data that we have that we can feed into these systems. Can synthetic data from a system like this kind of be trusted to be the gap filler in order to develop other AI models? Like as a world model coming from a Genie 3, how is it determined that that's usable information on which to base the training of other AI models? Yeah, I think the jury is still kind of out on that.

11:36It depends a lot what you're doing with the generative AI system. I think for things like, you know, chatbots, there's good reason to think that synthetic data will help. I don't know if it's going to be enough. I mean, we've already, you know, they've scraped the entire Internet, basically, to make these things work. But when you look at some other applications of generative AI like robotics, I think synthetic data, from what I understand, isn't as good. The Gemini robotics team has explained to me that the way you generate new training data for a robot is you have somebody teleoperate the robot.

12:08And that's how it gains new insights into what it's supposed to do when you say, pick up this glass of water and set it down in front of me. You have somebody manipulate the robot remotely and it learns over time. You can generate synthetic data to do that. Essentially, pretend that the robot learned this 20 ,000 times in a short period. But that data, they say, is not as high value as having physically moved the robot. Yeah, that's really interesting. And so we're talking a lot about robotics, obviously. That's probably a really big kind of potential application for something like this longer term, right?

12:48It isn't just video games as we've repeated, but what are some of those other kinds of applications that this could be applied to outside of entertainment, let's say? I mean, just about anything you can do with AI. I mean, if you were trying to train an AI that understands how to drive a car. You could show it infinite roads. You can change the conditions. And presumably if Google is simulating physics correctly, it would gain some useful insights from that. But again, this stuff is all very early. When you think about it, the tech industry has only been obsessed with this stuff for two, three years now.

13:26And it's not clear where it's going to go just in the next few. But I mean, if it doesn't get cheaper very quickly, I don't know how much, how much longer most of these companies are going to be able to keep doing it. I mean, Google has got essentially unlimited money right now, uh, which is why they're, I mean, you know, they're, they're giving away so many of these, these video generation things, you know, even if you're paying a lot of money for it every month, you're costing Google money probably when you, you know, go through all your video credits. Um, so, you know, if, if Google can make this stuff cheaper and make it better than like, you know, you can do almost anything with it, but you know, I don't, I don't know how many industries are going to want to predicate their entire product line on having access to like something like genie 3 when they don't know if they'll be able to afford it you know yeah it feels like like one outlet might be a high level training you know of of industrial uh people or search and rescue people or safety people something where you can get a big budget to throw against it uh would would be something you could possibly see them do now but the cost is going to have to come down even for video game companies to want to integrate it, I think.

14:32Yeah. Well, it's fascinating stuff. I love seeing just how rapidly this stuff is advancing. And to think that this is all happening in real time to the quality that it is, it's just, it's mind boggling to me that this is even possible. But you know what? Like in a couple of months, we're going to be saying the same thing about the next step, whatever. Yeah, like Ryan was already pointing to Genie 4 in the Ars Technica article. I was like, yep, maybe then, you know. Maybe then we'll be amazed. The team themselves, they seem kind of shocked that they had gone as far as they had in just a few months.

15:05Oh, that's cool. Yeah, it's pretty remarkable. Well, Ryan, thank you so much for joining us. Folks, of course, can find all of your awesome work at ArsTechnica.com. Thanks for joining us, Ryan. It was a pleasure to see you again. Take it easy. All right. DT &S is made possible by you, the listener. Big thanks to ThatCharlieDude, Justin Zellers, Carmine Bailey, and we got some brand new patrons to shout out. Welcome Rami, Lutopi, Breanne, Sebi, and Tonic. Yeah!

15:40Episode prep gets messy when research, Slack threads, and script trackers are all in different places. That's where ChatGPT work comes in. Give it the goal and connect to the tools where work happens, like apps and files, and it can help pull the pieces into a draft you can review without you having to re-explain the project at every step. For example, ChatGPT work can gather team feedback from Slack and use it to suggest updates to a project brief in Google Docs. You stay in control reviewing the draft and deciding what moves forward. Put ChatGPT to work on your most ambitious ideas and projects.

16:15Get started at chatgpt.com by selecting work mode. Available on Plus and Pro plans. This 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. Summer's supposed to be the easy season.

16:56So why are so many people quietly Googling a therapist between summer Fridays? Because more daylight doesn't fix the hard stuff. Sometimes it just turns the volume up. Grow does therapy differently. Grow connects you with thousands of high-quality licensed therapists across the U.S., offering both virtual and in-person sessions, nights and weekends. There are no subscriptions, no long-term commitments. You just pay per session. Grow helps you find therapy on your time. The therapist you want takes your insurance on Grow. Grow accepts over 125 insurance plans. Sessions average$21 with insurance and some pay as little as$0 depending on their plan.

17:37Visit growtherapy.com slash ACAST today to get started. That's growtherapy.com slash ACAST. growtherapy.com slash ACAST. Availability and coverage vary by state and insurance plan. Hi, I'm Angie Hicks, co-founder of Angie. And one thing I've learned is that you buy a house, but you make it a home. Because with every fix, update, and renovation, it becomes a little more your own. So you need all your jobs done well. For nearly 30 years, Angie has helped millions of homeowners hire skilled pros for the projects that matter. From plumbing to electrical, roof repair to deck upgrades. So leave it to the pros who will get your jobs done well.

18:17Angie, the one you trust to find the ones you trust. Find a pro for your project at Angie.com. You're a pro at running your life, at committing to your workout, at showing up every day. At Bombas, we're pros too. Pros at making socks. Our Sport Assortment has specialized socks for whatever sport you're committed to. Running, hiking, golf, Pilates, and so much more. Made with sweat-wicking yarns, blister-fighting details, and targeted art support. Bombas Sport is pro-level socks from the pros of socks. For another pro, you. Go to bombas.com slash audio and use code audio for 20 % off your first purchase.

18:51That's bombas.com and use code audio.

18:57All right, there's more. we need to know today so let's get to the briefs yeah cloud flare says it caught perplexities ai bots accessing web pages that have been set to block its bots cloud flare says it created in response to customer complaints uh some a set of fresh web domains that had not been indexed by search they were so new even google hadn't got to them and they set robots.txt files to prohibit scraping both of search indexes and of bots it then asked perplexity a question about that site which it shouldn't be able to know anything about and perplexity answered with detailed information from the content on those sites cloudflare then investigated and found that what perplexity was doing is when its agent was denied access and i use the word agent on purpose it changed the user agent string to not be perplexity anymore because it was denied, but to appear as a Chrome browser.

20:02I believe a Chrome browser for macOS. And if the IP address it was coming from was one that Cloudflare was blocking because it's only used for scraping, it would rotate IP addresses to avoid the blocks. Cloudflare said this happened across tens of thousands of domains. So when I saw this story, I was like, this is too bad to be true. like what is perplexity's explanation of this and i gotta say jason it's it's both good and bad as far as i'm concerned about how much i believe this perplexity said in a blog post that cloud flare identified behavior from browser base which is a third-party cloud browser service that perplexity says it only occasionally uses for highly specialized tasks and that would be less than 45 ,000 requests a day.

20:49So Cloudflare saying tens of thousands matches up 45 ,000 requests a day. Cloudflare has delisted Perplexity's bot, but it sounds like there's a little bit of a mixing of message from Cloudflare where they're implying that this is scraping and Perplexity saying, no, it's just the user agent. And a lot of people have been defending perplexity saying, well, yeah, if I am in ChatGPT and I say, go to this site and tell me what's on it, it should be able to do that. It's not scraping it for training. It's looking at it for me. And in fact, ChatGPT can do that because ChatGPT follows Cloudflare's standard for user agent identification is when it says like, I'm an agent, I'm not scraping, whereas perplexity does not.

21:39So I don't think perplexity is blameless here. They could be following the rules. I think perplexity is always quick to shout, third party, not me, because it farms stuff out so that it doesn't have to take responsibility. And I don't love that. But this probably isn't as bad as it sounds. I'm happy you explained that because I don't know in my head that I have organized around the difference between scraping for, and scraping could be a bad word depending on who you're talking to, but scraping for these systems versus using an AI agent to help me with this particular page that I happen to be on.

22:19And how is that traffic seen differently? I'm sure it's seen differently on the other side. But yeah, what you explain, Cloudflare kind of confabulating or whatever the word is, kind of combining these two things into the same thing when potentially maybe they're a little different. I hadn't really considered that. When I use perplexity to summarize a page, is that different than perplexity having access to all these things to slurp up or whatever for use on the back end? Are those two different use cases, and should they be treated differently by the website? Some absolutists are going to say, absolutely not 100 if we say no nothing that has anything to do with perplexity should ever touch the site but that might be a little unrealistic based on how the internet works well no it's a reasonable question of like well what if i don't even want an ai agent to act like a browser and come here i only want humans to do it and cloudflare says uh there's a standard for that it's called web bot off and it's why when you go to chat gpt uh and do this we don't see the behavior perplexity is doing where it pretends to be something else chat gpt can identify itself as chat gpt but it's using cloud flares web bot auth proposed standard to say like not scraping you just being an agent and then the site can say well in that case that's okay uh or it could also say no i don't care whether you're an agent or not and then chat gpt will respect that most websites will be fine with that because they want people to get the information from the site uh that but there might be situations where you don't.

23:55And so Cloudflare has proposed that standard and perplexity ought to follow it. I'm sure perplexity has its reasons why it's not. Maybe there needs to be some more work done on the standard, whatever. But yeah, this isn't as clear cut as perplexity bad in this situation, just perplexity being a little sus and they could behave better and this wouldn't be an issue. Yeah, indeed. Indeed. Well, authorities in Taiwan arrested three people accused of stealing trade secrets from TSMC, the semiconductor manufacturer that, of course, supplies companies like Apple and Qualcomm with its chips. The information was reportedly related to the company's proprietary two nanometer chip technology.

24:35That's its most advanced technology in terms of density and energy efficiency. I think 2026 is going big time going to be the year of the two nanometer chip for mobile. I think that's on the horizon, according to a lot of rumors that i've read the identities of the individuals not released investigators believe of course that they had plans to sell or use that stolen information in the marketplace and uh yeah like i said two nanometer right on the right on the cusp of that so this kind of ties into that uh i think it was 2018 or roundabouts there that that they busted some people trying to take trade secrets from tsmc i think it was 28 nanometer so that's how long ago it was crazy and uh and i and i think they were trying to sell them to china presumably uh that could be a place where they might be trying to sell two nanometer chip technology information but maybe intel wanted it i don't know yeah who knows who knows i'm not saying that intel broke any laws when i say that it's just they they would have liked it i'm sure yeah 11 labs has released a new ai powered music generator tool that it says is cleared for commercial use company says content creators advertisers developers they They can all use the music that they create in this tool in their own content and not worry about legal liability or copyright strikes.

25:55Eleven Labs says that the data set, training methods, and output design avoid copyrighted sources and risky audio samples. So it's this new trend of like, this is ethically sourced training data. Yeah. And I mean, I've heard... So I went on their site and I listened to some of the stuff. And, you know, of course, as with everything music gen related, it has that kind of like glossy kind of quality to it. If you know what you're listening for, it's almost immediately like, yeah, this is the I generated. But if this is like true background stuff, like music bed underneath, does it really matter?

26:32Whatever. But I do think that it is a trend right now, and I think it's a worthy trend of making sure that the companies that are creating the music gen, that they are getting this ethically because it's such a big deal for the creatives who want to know that this system isn't built off of the use of their content. Or if it was, can we get a little cut of the revenue for helping you train up on it, right? Yeah. And I mean, I've heard a couple of systems like this that are done at least report purportedly, you know, self-reported done with ethical kind of approaches in mind. And in general, as you may expect, maybe not, but the quality of the music that it generates is often a step down from the ones that we're pretty certain get their music illegitimately.

27:21Well, yeah, if you get it from the very best musicians in the world, your training data is going to be better than if you have to pay the very best musicians of the world and couldn't afford it, right? Totally, totally. But yeah, it sounded good to my ears for what it is. So there you go. And I appreciate the approach and the reason behind it. How about a few smartphone rumor threads that we can pull out? We can tug at them a little bit. But first up, we have Apple carrier documents pointing to a Tuesday, September 9th launch event for Apple's upcoming iPhone 17 lineup. If this comes to pass, just kind of taking a look at how history tells us what to expect beyond that.

28:05If it's September 9th for the announcement, you could expect pre-orders to open on Friday, September 12th. Retail sales and availability might open on Friday, September 19th. All of this is speculative based on historical precedent from the past, but enough for you to start getting excited, I guess. If you want to get the iPhone 17, that's something to look forward to. And then we've got Google, as it's made by Google event that's legitimately, we know of this. Not a rumor. Not a rumor, set for August 20th. So that should give the company a head start in front of Apple on its big Pixel 10 lineup release, right?

28:42Well, a report by WinFuture claims that supply chain problems might actually push back the Pixel 10 Pro Fold, the Pixel Watch 4, and the Pixel Buds 2a all the way to October 9th, 2025. That's nearly pretty close to a two-month delay from the announcement to availability if this rumor or this report is to be trusted, which I guess puts Apple in the driver's seat to get ahead with their device. So anyways, right around the corner, you got some devices. You just might have to wait a little bit if you're a Google fan. Yeah, I mean, I'm not shocked that September 9th, the day that is consistent with when Apple always does iPhone announcements might be that day.

29:26Like that's not like huge earth shattering news. But the idea that Google wouldn't be able to release things till October and still do an announcement in August has me puzzled. So I don't expect the entire Pixel 10 lineup to be delayed till October. but maybe some of it, you know, and that could be the source of some of these supply chain issues. But a lot of times the stuff, you know, like Digitimes and folks talk about supply chain stuff and they're just misreading something or a date was set the farthest out just to be safe. And then it gets moved back up. So I'll be curious what we actually hear on August 20th.

30:01As will I. U.S. state of Illinois, my home state, passed a law that forbids anyone from providing advertising or otherwise offering therapy or psychotherapy to the public unless it is conducted by, quote, an individual who is a licensed professional. So no AI psychotherapy bots can be marketed in Illinois. Licensed professionals may use AI for administrative and supplementary support. So as a therapist, you could use AI with your clients. However, However, the therapist has to take responsibility for everything related to the AI's used. You can't later go like, oh, what was the AI that did it?

30:39You're responsible for everything. And you cannot use it in sessions where you're recording or transcribing without detailing and writing how the AI tool is going to be used and getting consent from the patient or their authorized representative if it's a child or someone else. The penalty for violation of those rules is up to$10 ,000 per incident. Wow. Yeah. Yeah, I mean, I'm sort of not surprised. And I think initially when I kind of read the headline and kind of started looking at this, it was like, oh, wait a minute. You can't use AI for therapeutic use in the state of Illinois. But it's less about that.

31:18It's more about specifically marketing it as such, right? If you're using ChatGPT to give you guidance on how you live your life, there's a lot of gray area. It's a little bit vaguely written because it says you can't provide, advertise, or otherwise offer therapy or psychotherapy or otherwise offer. So it would be up to a judge to decide, like, is just ChatGPT not blocking therapy otherwise offering it? Or is that like, well, that's a use that is contemplated by the user and not being offered. It'd be up to a judge to decide that. And I'm not going to be shocked if we have somebody bring the case at some point.

32:01Yeah, at some point. You're probably right. Oh, that'll be fascinating. Well, Spotify has launched Audiobooks Plus in the U.S. as an add-on to its existing streaming plans. This is something that the company had trialed with users in Canada, Australia, New Zealand, and a few other European markets prior to this. Spotify offers two distinct plans here. So one is meant for premium individual subscribers as well as plan managers on family or duo plans. So if you're a plan manager or an individual, you can get an extra 15 hours of audiobook listening per month in your account. The other is for plan members.

32:36So if you're like a member of a family plan or whatever, the person managing that plan could grant you 15 hours. And then you also have the option of topping up an additional 10 hours on demand if you read a lot of audiobooks or listen to a lot of audiobooks. The new feature launches in the U.S. on August 5th. Well, it's about time the U.S. caught up with Canada, Australia, and New Zealand on Spotify. Yeah, seriously. We were behind in the one place that matters most, audiobooks. Roku, really known for selling ads. It's making more of its revenue on ads than devices these days. That's where the growth is.

33:12However, it is launching an ad-free streaming service called Howdy for$2.99 a month. It will include Roku Originals. So it's a way for you to pay to watch Roku Originals without ads. And it will also have library content from Warner Brothers Discovery, Lionsgate, and a few other sources. But the big deal is it's only$3 a month. So next to all the other$20 a month services, it looks pretty cheap. Yeah, super affordable. You could almost sneeze that. Maybe not in the year 2025. Well, you could, Jason. Your sneezes are quite valuable. And loud, my kids will attest. That's why they're so valuable.

33:50Yeah, I scare them when I sneeze. And finally, speaking of scaring them, and speaking of my kids, love them or hate them, Crocs are a thing that people enjoy wearing, including my children. Microsoft is reportedly bringing its Windows XP grassy hill and blue sky bliss to a 50th anniversary edition of Crocs, complete with a clippy shoe charm that you can attach to it if you want. Microsoft employees get first dibs ahead of a worldwide launch. So if you can't get it once they're available, you can blame those Microsoft employees for putting these on their feet. These stylish shoes will set you back around$80 once Microsoft makes its official announcement that is expected sometime in the coming days.

34:35Yeah, it is sad that when we look at a blue sky these days, we think, oh, it's like Microsoft.

34:42Folks, if you have thoughts on Crocs or Clippy Shoe Charms or why Microsoft Bob has been left behind, join our conversation in our Discord, which you can join by linking your Patreon account. Become a Patreon at patreon.com slash DTNS.

35:01This 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 sack. That's technology at Capital One. Summer's supposed to be the easy season. So why are so many people quietly Googling a therapist between summer Fridays?

35:38Because more daylight doesn't fix the hard stuff. Sometimes it just turns the volume up. Grow does therapy differently. Grow connects you with thousands of high-quality licensed therapists across the U.S., offering both virtual and in-person sessions nights and weekends. There are no subscriptions, no long-term commitments. You just pay per session. Grow helps you find therapy on your time. The therapist you want takes your insurance on Grow. Grow accepts over 125 insurance plans. Sessions average$21 with insurance and some pay as little as$0 depending on their plan. Visit growtherapy.com slash ACAST today to get started.

36:19That's growtherapy.com slash ACAST. growtherapy.com slash ACAST. Availability and coverage vary by state and insurance plan.

36:33We end every episode of DTNS with some shared wisdom. Today, James has a reaction to Justin Robert Young's appearance about political messaging on iOS. Yeah, James writes, Hi, DTNS folks, longtime listener, medium-time patron, and love the show. I don't normally write in, but when I was listening to the August 1st DTNS briefing about the changes to the message app on iOS 26 preview, my blood was boiling. To me, this is basically the same thing as Gmail auto-enabling the categories a few years back to make email more usable than before. I remember there were doomsayers about, no, our emails won't work anymore.

37:10Hearing the numbers from jury helps me understand why it was so much worse before the election last year. 200 plus text per person. Jeez, that sounds about right. But damn, I'm really glad to hear iOS is doing more about this. I hope they give the user more controls like email filters. Personally, any text from unknown sender with certain words can go straight to the trash. As far as I'm concerned, I don't even care which side they are coming from. I hope Apple turns this on by default. Same with Gmail. The industry will just have to deal with it. Anyway, thanks for listening to my rants. Quack on, a.k.a.

37:44James. I can get behind that. I mean, oh, God, I am so inundated, not just in email, but like James says in text as well. It's overwhelming. As soon as the elections come back, the midterm elections, it's going to ramp up again. And the next presidential election in the US, it's going to ramp up again. So it is something that I think a lot of people are paying attention to. I'm actually going to be on Politics, Politics, Politics with Justin Robert Young this week, talking with him and the money man, Michael D. Cohen, who follows campaign finance, to kind of talk about all the different angles on this.

38:21I'm looking forward to that. Excellent. Everybody should check that out. And if you have a comment about anything that we're talking about here, if you have insight into a story, whatever it is, share it with us. Feedback at DailyTechNewsShow.com. Big thanks to Ryan Whitwam. Thanks to James for contributing to today's show. And thank you 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. And thank you for doing that. We'll see you next time.

38:52the dtns family of podcasts helping each other understand diamond club hopes you have enjoyed this program this 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 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.

39:34Your team just added its 67th AI tool and also your 67th security blind spot. The good news? the Vanta agent works like a GRC engineer in the background, finding every app your team uses, scoring the risk, and drafting fixes for you. Vanta is the platform used by over 16 ,000 fast-moving companies like Ramp, Cursor, and Harvey, who are shaping the future with AI and staying ahead of AI risk. Get started at Vanta.com.