Pixel 11 Series is More of the Same But is That Bad? - DTNS 5336

[music] This is the daily tech news for Thursday, August 20th, 2026. We tell you what you need to know, give you the important context, and help each other understand. >> Today, science communicator Bobby Frankenberger tells us what's holding back using ocean waves as an energy source. And the Pixel 11 reviews are in. >> That's right. I've got one of them. I've got thoughts, though. I don't have a full review, but we'll talk about the reviews. I'm Jason Howell. >> I'm Jen Cutter. >> Let's start with what you need to know with the big story. >> Yes, the Pixel 11 reviews are out. So, you got the 11, the 11 Pro, the 11 Pro XL. You know, you've also got the Pixel full. Actually, now that I think about it, I don't know that I saw the full reviews, but I wasn't really paying close attention to those because I've been using the Pixel 11 Pro personally. And so, I've been very curious to kind of see what reviewers uh are saying about it as far as, you know, does it kind of line up with my own personal um kind of perspective and experience. Though, I'm not ready for a full review. I'm probably going to be delivering that, well, almost certainly delivering that next Tuesday on Android Faithful uh with the crew. [snorts] Um, but thought we'd do a little bit of a roundup to kind of let you know like is this a phone that you might actually want to get for yourself. Well, there's a bit of perspective here. Seems like a resounding yes on the feature that is also my favorite feature called Rambler. Rambler is this uh AI infused voice chat essentially. So instead of me just using my voice and saying a bunch of things and it and it does it in real time exactly as I said it, it kind of cleans it up. So, if you say a bunch of things and maybe you mess up or maybe you change your mind and you say something else or whatever, it it what it does is it sends that to an LLM, cleans it up, but keeping it to your voice and then giving you the output in the the text box. I was worried that it was not going to sound like me when I use that, but um I use it all the time. I absolutely love it. So, it seems like everybody's loving Rambler. One of the AI features, another AI feature is something called camera looks, which essentially means you can assign a default camera setting to it's like a template to all of your photos. So essentially, and I I'm curious to know what you think about this, Jen, because the idea is, at least if I was to interpret it, Google might not like this, but it's like adding a filter that you might add in like Instagram or whatever, like a like a photo changing filter or whatever, just like color and and uh and contrast and that stuff, but doing it as a setting in your camera instead of applying it after the fact, so everything follows that look. Does that sound useful to you? Because I'm kind of like, I don't know why I need that. Uh it is aimed at a a very specific market because probably not a group of people you follow on uh photobased social media, but like the aesthetics people, if you remember when flatlays were like the big thing, this saves them a step. They can set their preset, shoot, and go. Uh and >> I like I do know some people who use this kind of thing. uh like for for iOS like when you can like have those settings and for people who have their save stuff in Snapseed where they just like auto run it through they have a shortcut going this takes out that step >> I suppose so but you're kind of making the decision I guess in my mind it's like but that's destructive maybe I don't maybe I like that picture but I wish I didn't shoot it like that or you know what I mean >> like I [laughter] mean I have everything off and then so all my terrible pictures I own as my terrible pictures and then in Exactly. [laughter] >> Because like because from there I can take the extra time to do the thing, but you know, I'm not that's not my job. I I'm not doing this in a hurry. I'm not working for any brand where I have to have a preset set. And uh like who knows for some of these brand deals if they want a very specific >> hyper specific look for their products and then say we want these settings and don't change them. >> Yeah. Yeah. Well, I'm clearly in the minority on this one because I'm still scratching my head. But apparently most of the reviewers are like, "This is pretty awesome." And I can I guess I can kind of tell it allows you to apply basically like a signature to your camera look on your phone so that every picture you take kind of falls inside of that signature. So, I I I get it. That that does sound kind of appealing. I just don't know that I need it for me. There's also a feature that everybody seems to like, myself included, called magic capture. Uh pretty universal standout. This is the idea that when you hit the shutter in magic capture mode, it's captured like a couple hundred frames before and after the shutter and then it goes through with AI and pulls out the best frame from there. So, if you're trying to get something that's fast moving or whatever, you know, you're always trying to time the shutter. This way, it captures all of them and it says, "Oh, well, here was the best frame." And you can kind of select and everything. I've used it a couple of times, been pretty impressed with what it pulls out. So, I'll be curious to use it more in like when I go to my kids sporting events and stuff. Um, it seems that everybody agrees on highlight. Highlight is the feature that is this little light here. If I could trigger it to to light up, I would. But it's it's a very subtle kind of glowing light that replaces the flash on the rear camera bar. And it, you know, it does a couple of colors. It's supposed to be kind of like a an extra notification layer. If you have your phone down and a contact is assigned to a color, you'll see the color there. If you're interacting with Gemini, it kind of flashes a little bit. It's super underwhelming. Maybe they build this out over time and it becomes something more uh interesting, but right now pretty disappointing across the board. Myself included. It's like I have to go out of my way to even see it doing anything, which is weird because they've been promoting it so heavily as like a big feature. That's a little disappointing. I don't know. LED light. Okay. Um overall though, it I think resoundingly and we were talking about this a little bit before starting to record. Hardware design pretty iterative, right? Like the the previous uh designs look a lot like this one. The last like three or four years have looked very very similar. minor changes in the camera bar where the glass goes the full distance instead of kind of going halfway um and you know becoming metal on the other side uh on the Pixel 10. Um, but [clears throat] uh overall the language is is staying put and a lot of people are calling this out in their reviews as kind of a ding saying, you know, Google's not doing anything exciting here. Maybe it's time for a change. Personally, I like it. I continue to like it. What? And I feel like the call to like change it is just like change for change's sake, maybe. I don't know. How do you feel about that sort of thing? like if there's a good um if there's a good ingredient, why not keep it? Yeah, I that's my feeling. >> Yeah, I believe that they are aiming mostly at I'm I might be making up a market here, but the safe phone market, the people who already like their phones and just they're they're finally like at the end of of their support. They want to run a faster phone, a more modern phone, they can switch to this and feel safe that it's not drastic. And then for people who want just a phone, this can appeal to them as just a phone even though it is pricier. And uh I actually had thoughts about the the highlight thing because there's always two ways you can go with a feature like this. You can either start it at 11 and draw it back or start mild like this. Like because Google was always going to respond to the uh feedback here. So if they start mild and bump it up, that's more of a positive response to feedback as opposed to, "Oh god, everyone hated this. Turn it off. Turn it down. >> [laughter] >> Yeah, that's true. And you know what, and almost certainly Google's going to add feature set to it, you know, like I'd be really surprised if at some point we don't have more apps that tie into it. The ability to get really granular with like what color is attached to what thing. I'm guessing that's going to happen somewhere down the line. Um, yeah, it's two different ways you could go about that, and I guess time will tell whether Google chose the right way. Um, inside, oops, sorry, wrong camera. Uh inside uh you've got the tensor G6 uh the RAM, the display, those are you know the same whether you are on the Pixel 11 or the 11 Pro. So there's a little bit of an incentive that's that's uh a little bit more difficult to sell uh between the upgrade and the the kind of lower version of it. So definitely reviewers calling that out. Battery life. Google said that they made a lot of software changes on here to improve battery life and battery efficiency. Many reviewers saying this is still a weak spot here. I've noticed mine hovering about 20% more by the end of the day than I was getting on the Pixel 10. So, I see that as a little bit of an improvement, but still it's kind of down in that lower end where I feel uncomfortable at the end kind of of the day having it down there. So, could be better. Um, and overall, I'd say people just kind of like agreeing that 20 12 gigs of uh of RAM is a little disappointing. Although I don't run into any use when I'm using it where I'm like, gosh, I wish this had more RAM. I'm sure I will eventually, but I don't right now. And I think the last thing I'll say because I could keep going on is proactive assistance with Gemini. This is the um kind of what it sounds like where you can either you're in messages and you mention, you know, we should go out to dinner and then it gives you a little thing to say, want to add that to your calendar and maybe I'll book the reservation for you and it can do all these things for you. Uh some reviewers seem split on its usefulness. I've used it a couple of times and it's neat that it does it when I remember that it's there. I mean, it's kind of cool that we have devices that'll do that sort of thing for you now. So, I'm I'm about it. So, I would say iterative some neat features. If you're on a Pixel 10, probably doesn't make sense for you to pay for a Pixel 11 right now. You know what I mean? >> Yeah. Know, that's that's pretty important feedback. And uh yeah, >> for the Gemini like obviously Gemini is going to be more closely integrated with these phones with probably every OS update at this point, never mind like the physical phone. Um but so you have used it, you have had it walk through some stuff for you. >> Yeah. Mhm. Yeah. And you know it it doesn't go the final step. It doesn't actually book, right? you have to push the the button, but it's cool to kind of walk it watch it kind of skip through and offer you the information and connect to the different things with your personal intelligence. It's a neat feature and I'm sure it'll get better. So, I like that. >> Yeah, I've seen lots of people say like it's okay, but so far this isn't doing more than my Google Assistant did. It's not doing more than my setup shortcuts did. So, got some ways to go on that. >> There you go. DTNS is made possible by you, the listener. Thanks to Chris Allen, Reed Fischler, Larry Bailey, and Shawn Meyers. Thank you all and always. >> All right, there is plenty more we need to know today. Let's get to the briefs. >> Stripe has officially agreed to acquire Open Router, the AI model gateway used by more than 10 million developers and businesses. Stripe will add the model routing capabilities along with token spend optimization to its payments and token billing business. So what exactly is open router? The service routes business and developer spending across more than 400 models from more than 80 providers. So users have direct access to all kinds of agents from Hermes, Cloud Code, OpenClaw and Codeex to name a couple of examples. This allows users to pick a model based on price, speed, and reliability for the agent work that they are doing. There are also open- source models offered alongside the frontier models. So, why would Stripe want open router? It's because it considers tokens a form of currency, and Stripe thinks it can help customers spend tokens more efficiently. Stripe actually launched a token billing product last year. So, this ties directly into that kind of product offering. Open Router intends to keep its name, product, and existing roadmap within the deal. And the closing date is currently unspecified. >> Yeah. And Open Router is, you know, not the only company that's doing this sort of thing. There's a lot of demand for this, especially like we've talked about like um in the past year, a lot of companies have been incentivizing employees to spend lots of tokens and use the things and then they've then they've had to kind of scale back, right? So demand is high for companies on um on looking now anyways at efficiency around token usage and kind of pivoting away from that maximizing mindset. And uh so something like open router is really great because it doesn't lock you into a single solution for doing all your things. Kind of gives you the ability to set up these agents that reach out to other agents um that are better at certain things and that allows you to pay pay closer attention to oh well this is cheaper for that task versus just using this for everything. That's kind of part of what open router does. And then you got Stripe thinking about these tokens as payments. So it kind of um basically what it does is it gives Stripe more financial control over AI apps um through uh through open router. >> Yeah. Know these things are are uh are becoming more and more important and I think for uh for a lot of companies they're like oh I know Stripe. Stripe is reliable. Stripe isn't going anywhere. They get to borrow that credib well buy that credibility. [laughter] >> Yeah. Yeah. I mean, it basically it puts Stripe in in like a money manager position for AI through in in this deal, which isn't a a bad place to be right now. There's so much happening uh in the enterprise around, you know, and and potentially as as agents gain more access to like spending capabilities and stuff, right? We're we're kind of like starting to dip our toes into that water. And so, Stripe will be in a good position as a result of that. So, interesting purchase, uh interesting acquisition. Thanks to RW Nash and subreddit for linking us to this story. According to Bloomberg, YouTube is reportedly offering leading creators and very popular channels uh millions of dollars in exchange for keeping their videos exclusive to YouTube's platform. That includes direct funding for some shows, a cut of major brand deals when it lands them. Why? Well, those partner agreements um even though they have yet to be finalized as say sources have to do with YouTube wanting to keep people from going over to Netflix. YouTube is basically warning that cross-osting to Netflix would harm their viewership on YouTube because they then miss out on certain uh you know those creators if they do that might miss out on certain marketing campaigns, event invites, reduced brand campaign payouts. So, it kind of feels like, hey, you know, keep it over here because if you go over there, you're going to miss out on all this stuff. And Netflix, um I don't know if you've noticed, Jen, I've certainly noticed um signing some notable YouTube creators that are coming into the platform. Allan uh Chicken Chow, uh Nick Deiovani, just a couple that I've seen uh to have their videos posted on Netflix concurrently with YouTube. And in fact, like I had never watched Hot Ones before. I'd always heard about it for years and years. I'd heard about it. I'd seen little I'd seen little clips here and there, but I'd never actually sat down and watched a full episode until I came across it on Netflix. And for no reason other than I it was just right place, right time, I watched one and I was like, "Oh, okay. This is great." Um, but it's interesting to me that like it had existed for years on YouTube and I never gave it a start to finish time of day until it was on Netflix. So that's clearly that's what Netflix wants and what YouTube doesn't. >> Yeah. Well, whether it is a streaming service or an app, time on platform is one of the biggest metrics they're always looking for and looking to increase. And yeah, if you're watching hot ones on Netflix, your recommendations after that are going to be Netflix shows and not other YouTube shows. [laughter] >> That's true. That's true. It keeps you in that uh that that space, right? Um also curious about the whole regulatory aspect of this. Because if YouTube is saying you stay with us exclusive with us and you get all these things if you don't then I'm just saying they go away that you know certain aspects of that sound similar to what Android has been under the the view of and what search has been under the view of. There's been a lot of unwanted regular attention from Google's perspective recently. Does this then kind of put YouTube in a similar position? Potentially. We'll we'll just have to Well, it depends on are you viewing it as tech or are you viewing it as broadcast television cuz network signed development deals all the time. You keep your stuff with them. They get first rights of refusal. And that's a contract you are entering in. It sounds at this point you get to make that call. It's a willing contract. It's not a you will do this or we are burying your channel unless it turns into that. And then I I I would hope that there some regulatory body chooses to step in at that point. >> Yeah. Yeah. Yeah. there. >> If you have any feedback about anything that goes on in this show, please get in touch with us. We've got lots of social media options for you. We are at DTNS show on X, Instagram, Threads, Blue Sky, and Masttodon. And for YouTube on and on Tik Tok, you can find us at Daily Tech News Show. [music] All right, we're going to jump into some quick headlines, make you look really smart before an interview coming up. Google bid $10 million in Spirit Airlines bankruptcy auction for the airlines internal archives, including about 100 million employee emails, all for training its AI. While the flight attendance union is fighting it in court, arguing deidentified records could still be traced back to individual workers. >> Yeah, this one's really interesting because they're all looking for data that isn't influenced by AI and this is one way to do it. You know, I think we're going to see a lot more of this sort of thing. Very interesting. Samsung's apparently holding another Galaxy event. This is official from Samsung on August 27th, announcing quote, "The newest addition to the Galaxy S26 family." Uh, we're kind of thinking it might have something to do with the budget Galaxy S26 fan edition smartphone. Uber launched autonomous rides in Ziggreb, Croatia. Its first driverless rollout in Europe. It'll run on Pony AI's tech through Croatian startup Vern's existing robo taxi fleet. >> There's a first. Qualcomm's teasing new two, sorry, new flagship chips uh for next month's Snapdragon Summit, the Snapdragon 8 Elite Gen 6 and a beefier pro variant that's rumored for ultra tier phones like Samsung's next Galaxy S27 Ultra. We can imagine I will actually be at the event, so I'll be reporting on the news for DTNS, so you can look forward to that. >> The UK Cinema Association is weighing a ban on Meta's AI glasses in theaters. Uh unsurprising since the camera can easily record [snorts] theater stuff. >> I I think you you got a lot of cojones walking into a theater wearing meta glasses like what? I'm not going to use these. Although it would kind of suck because I [snorts] think you can only record to like three to five minute clips. So >> you're doing it in chunks. You're you're watching >> a horrible way to watch a movie. [laughter] >> I've seen worse though, but that would be a horrible way to watch a movie or record one. >> Uh Whimo built its own custom chip for its robo taxis to cut its reliance on Nvidia and AMD hardware. It's already running in the Zeerbuilt cars in its fleet. The chip handles more than 1,000 trillion operations a second, which is on par with Nvidia's latest self-driving systems. Adobe's Firefly can now generate music, voiceovers, and sound effects for video, each writing on its own model. And it's free to try with daily limits, and no separate subscription required. Hm. LG's got a new OLED manufacturing process called Flip F Lip PP that skips the fine metal masks that other makers actually rely on. That basically means panels 1.6 times brighter. Uh the panel is 2.4 times longer lasting while cutting power use by 13%. So better. >> I want an OLED monitor monitor so bad. So I love watching this stuff being like please please make this good. >> And your next one will be this. Yeah. And even if it makes the old ones cheaper and I can buy one, that works out great. >> Yeah. [laughter] Right. Right. >> Slack launched Slack Code, letting teams tag AI agents like Claude Code, Devon, or GitHub Copilot right in a group chat to build and ship software. A human still has to approve anything before it merges to production, hopefully. And finally, Sennheiser's new Momentum True Wireless 5 earbuds arriving September 3rd for around $300. They add Dolby Atmos with head tracking, uh, Bluetooth 6.0, and batteries you can actually replace yourself both inside of the buds and inside of the case. I like that. >> Those are the essentials for today. Let's dive a little deeper. >> Yeah, we got an interview for you recently. bgr.com wrote up a report on tidal energy, a renewable energy source that doesn't rely on the sun or wind, but the tides to generate power. Tom Merritt caught up with science cor uh sorry, communicator Bobby Frankenberger to check on where we are with this source of energy. Hey Bobby, thanks for taking the time to help us understand this. >> Yeah, no problem. Thanks for having me, Tom. >> All right, let's start with tidal energy. It it probably explains itself a little bit but but what exactly is it? Maybe we have misconceptions. >> Yeah, tidal energy it's sort of the it operates on the same principle as say wind energy. Um the atmosphere is also a fluid just like the water is water is. And so what wind energy does is say it it says okay there's a lot of energy moving the the atmosphere around. Let's let's figure out a way to capture that energy. So you put up wind turbines. The wind moves across the turbine, causes to spin, and then we turn that mechanical energy of the turbine spinning into energy with magnets, and we put it into the grid and store it in batteries. It's the same exact concept with turbines under the water. The tides, which are um being pulled and and affected by the moon and the sun, gravity um sloshes water around on the earth in a predictable way every day. And so we say, well, the there's tons of water moving every day across the surface of the Earth. Why don't we do the same thing that we do with with wind turbines, put turbines under the water to capture the energy? And it works exactly the same way. You just put it under the water and as the water moves across it, it spins. So that's turbine energy. And there are power plants that actually take advantage of that. So >> yeah, I mean I guess pretty much every energy generation method relies on turbines at some point, right? And we're just and tidal energy is is just using the waves to do it. Now tides are are much more predictable than the wind though. Um so it seems like >> it would be a more reliable source of energy than wind. >> Reliable. Um yes, in the sense that it's predictable like you said. Uh so one of the downsides with um other kinds of renewable energy like wind or solar is the weather. The weather's hard to predict and whenever the weather's bad, you're not generating as much wind or solar energy as you would like. And when I say bad, I mean just not favor favorable, right? Um but tides are much more predictable. Tide comes in and goes out at predictable times every day. So if we have something in place that can capture the energy at those times, then we we can plan for when we're going to be getting energy and then and then we can if it's not at a time when it's actively being consumed, we can store it in a battery, right? Or something like that. But that's also um at the same time as being a pro like you know we can predict it in a sense it can kind of be a con as well if you think um the tides happen only at certain times right so that means between the tides the water's not moving as much. So uh whereas we can predict the the high generation times and low generation times there are also definitely times when it's not producing energy. So, it's not a 100% of the time energy source, but at least we know when it won't be there, right? Uh because, uh unless something happens to the moon, uh we're we're we're pretty good on on knowing when the tides are, right? Where are we with doing this? Like, are are there efforts to make this happen? Are there actual operations using it? Uh where are we? >> Yeah, you'd think that uh that this seems to be check a lot of the boxes in the in the pro column, right? like like it's predictable. We know how to do so. Why aren't we seeing these kinds of plants everywhere? And so maybe I'm getting ahead of myself a tiny bit because we as far as where we are, there aren't very many of these in the world. Um there are a handful. I think I think I thought when there were only two, but then I I think there's a third one that's either being or just being built, but um I know the two that have been in operation for a little while. One's in France, the other one's in South Korea. um they uh they're they're put in specific locations that fully the are suited to best take advantage of the tides and how much water they push through certain areas like in France I think there's like a narrow waterway in the ocean that the tide comes in and it pushes a lot of water through this narrow opening in the land and so that's an ideal place for it but to answer your question about where are we like there aren't as many as you might think because for many reasons But one of the reasons being location, right? There aren't you have to have them next to coastlines. That's just what you have to do. And then further, not just anywhere on the coastline. It has to be in kind of an ideal place, not too close to the coast, not too far away. The waves can if it's too close, waves can cause problems with the energy generation. And and then and then in those places where it's best suited and ideal to build them, you you know, is it going to affect the wildlife? Is it going to are you going to be able to get people who are beachgoers and tourism departments to on board to say like like yes, we would like to have this 15 10 year construction project here. >> Um so that becomes challenging and so I'd say >> that's why we're not further along than we are. um currently, you know. So, so it it's a limited number of places you can put it. I assume because if you put it too far out, it's either hard to get the energy in or the tides just don't help as as well or they're so it's it's basic um uh I can't remember the name of the law, but it's um it's a you know the idea of when a fluid gets uh concentrated into a narrow gets gets concentrated down into a narrow opening as it's moving, the velocity increases. Um and uh and so the um so when you restrict that so that happens not just in places like France where the you know there are there you know like like literally the land is creating a narrow opening but also every coastline creates this effect because as you move from deep water where there's a lot of space for this moving water to move it doesn't move very fast. >> Oh yeah the land is constricting it the land and the air I guess. Exactly. As you get close to the coastline, the land or the land raises up to the coast to come up above the water. And so that narrows the amount of space that the same volume of volume of liquid has to move. And so as it's moving towards the coast, it gets faster and faster towards the coastline. So, ideally, you want it to be really close to the coastline, but like I said, the wave the mechanics of of the waves at the surface can interrupt that flow and create turbulence, which >> the turbines have to take advantage of the of the water moving in one direction. So, where the where waves are crashing at the top, it it causes the the the water to move in lots of different directions that are not ideal for the turbines. Yeah. And and I know it's expensive and I know that because everything in [laughter] energy uh is expensive. So when you take into account like it has to be in the right place and then you have to have everybody who lives in that place uh tell you, "Yeah, cool. Build this huge thing right in our beautiful oceanside view. Uh and you need to make sure you're not damaging the wildlife." uh and and causing a problem uh that that is, you know, what we're trying to get away with with these zero emission alternatives. Uh I can see why this is not something that is easy to just pop down everywhere. >> Uh what what do you think the chances are of this becoming more common? Like is it just something that's only going to be small? I know that uh recently Nova Scotia has has tried to encourage smaller versions of this to happen off its coastline. Uh Scotland I think has has a program to develop them as well and there's a few others out there. Uh how are you feeling about the future of this technology? >> I think um it depends on the the mathematics of of scaling it up, right? And small turbines aren't going to produce as much energy as large turbines, but then large turbines, like you said, cost more and have a greater impact on the environment. The fact that these things spin, they use magnets and to generate the electricity. And so the spinning blades are are also spinning magnets which create magnetic fields which some research has showed has shown that that has an impact on wildlife in the area. Um, also just sound, uh, over and over again as we put stuff in the ocean, we learn more and more that that sound travels very far and has wide ranging impacts on wildlife. Um, so I think those are the biggest things. Look, if we can put wind turbines up that do something similar, it it seems like an easier cell to say, let's put a whole bunch of wind turbines out into the middle of the desert rather than putting these kinds of turbines near the coast where people and lots of wildlife are. >> Unless you can find those areas of coast or enough area of those coast where there aren't people and maybe not a significant impact on the wildlife, right? Uh I know there's there's some efforts to come up with sensors that can can monitor the wildlife and I'll be the guy who points out like hey uh at some point we do have to make a decision of like is it worse is it a worse impact on the wildlife versus a worse impact on the heating of the atmosphere. Uh you know those are a Sophie's choice kind of situation as well. Bonnie, thank you. Oh yeah, go ahead. I was just going to say real fast that I don't think that that's an insignificant point to make though because um we can if we just keep spinning our wheels talking about uh all the downsides and never actually do anything then that's bad too. So >> yeah. Yeah. And that doesn't mean you have to dismiss the downsides. It's just difficult choices. Bobby, thank you so much uh for helping us understand this. If people want to find more of what you do these days, where should they go? >> They can go to allaroundscience.com. I've got a weekly science podcast that I do with my co-host Mora and we talk about things like this and all other kinds of things that are happening in science news. It's all around science just wherever you get podcasts or check out the website. >> Fantastic. Thanks, Bobby. >> Thank you. >> All right, we end every episode of DTNS with some shared perspectives. Today, Dana has a reminder on how large language models actually work. >> Dana writes, "As a relatively new listener, I really enjoy the show." Thank you, Dana. >> Oh, great. >> The topic of open AI and anthropic agents going rogue and hacking third parties has been covered multiple times as it should. Tom has rightly mentioned that AI is immoral and will use any means necessary to achieve a goal such as passing a security test. However, the discussion tone occasionally seems complacent, conveying an attitude of of course AI is going to do this. We should realize this and move on. The real issue, which you mentioned many times but never discussed, lies within guard rails. Currently, guard rails are the primary defense used to control models. But they only provide guidance on known or envisioned behaviors. Because AI reasons on its own, it's impossible to anticipate everything a model might do. Having worked in the early days of AI based security, I know firsthand that you can never truly determine why a model makes a specific decision. You can only guess based on observation. The models today are infinitely more complex than the ones we used, making it impossible to establish all-encompassing guard rails. To use an analogy, it is like a pet dog escaping despite fences and leashes or a child finding an unimagined hazard in a baby proof house. With guardrails, there will always be an AI bypass, whether through a model escaping security or a user bypassing controls to get a desired answer. That is the real issue with the current state of AI, and I'm not convinced the current approaches can truly scale to the level of control that is needed. Best regards, Dana Kaufman. >> Yeah, I mean, you you speak truth, Dana. I mean, guard rails are imperfect, right? everything. There is no 100% when it comes to AI is is something I've been saying lately like in in all different directions. There is no such thing as 100%. Guardrails is absolutely one of them. And I think it's easy I know it's easy for me to kind of feel a little complacent about it because it's kind of like what do we got to do? You know what what can we actually do? But it's it's good to know exactly how this works underneath and I think you do a really good job of illustrating that. So, thank you for sending that in. >> Yeah. No, I I love the analogy. I take care of dogs and puppies sometimes, and boy are puppies real inventive of finding things that again, my my house pretty uh pretty puppy proofed at this point after several years. And I'm like, how did you find that? I didn't even know that was a problem and now I know it's a problem. So, you know, iterate. Exactly. [laughter] >> You need better guard rails is what you need. >> That's how I'm going to look at it from now on. So, what are you thinking about you people out there? Do you have some insight into a story? Some more analogies that I can borrow. Please share it with us over at feedback at dailytechnewshow.com. >> Yeah, thanks again to uh Dana for contributing to today's show. Want to also thank Bobby Frankenberger for the awesome interview and thanks to you for being along for Daily Tech News Show. You can keep us in business by becoming a patron at patreon.com/dtns. See you tomorrow. >> [music] >> The DTNS family of podcasts helping each other understand. Diamond Club hopes you have enjoyed this broker. [laughter]