The EU Regulates ChatGPT Like a Search Engine - DTNS 5343

[music] This is the daily tech news for Monday, August 31st, 2026. We've run plum out of August. We tell you what you need to know, give you important context, and help each other understand. >> Today, the EU turns its regulatory sites on OpenAI, Reddit, and Roblox. >> Yeah, that's good stuff for them. Also, it's goodbye to Tim Cook. I'm Tom Merritt >> and I'm Bob Dunwood. >> Let's start with what you need to know with the big story. >> Hat tip to RW Nash for noting this one on our DTNS subreddit. The European Commission announced Monday that it has deemed chat GPT, Reddit, and Roblox as very large online platforms in Europe and therefore subject to the digital services act. uh chat GPT actually not a very large online platform but its designation is a very large search engine uh very large online search engine platform uh so they're just falling under the search engine category whether it's a platform or a search engine it counts as very large under the European Union's DSA if it has more than 45 million monthly users this is pretty straightforward there's not a whole lot to appeal here I get they could if they wanted to but uh Chad GPT has reported 159 million monthly active users in the EU. So well beyond the 45. Uh Reddit has reported 57.2 million and Roblox has reported 48 million. These are all numbers that the companies have reported themselves. So if they were to appeal it, they'd have to be like, "We were lying. We were mistaken." They're not going to do that. Under the DSA, every business that is in in the social media or search engine space must have mechanisms to report illegal content, provide advertising transparency, and comply with special rules concerning minors. But very large online platforms must also file transparency reports, detail how they what they're doing to reduce risks on their platform, and pay an annual fee. Risks include the aforementioned illegal contact uh content, how to protect your minor users, uh physical and mental well-being of your users, political things like human rights, uh election processes, public security. Researchers and regulators must also receive access to data and the very large companies must submit to independent audits for search engines and in this case that means Chad GPT. The reports must also detail how responses to queries are produced and how the company assesses any risk to those responses. JPT, Reddit, and Roblox have four months to come into compliance. I I think it's significant that Roblox has gotten big enough for this, especially with all the child protections that are in the DSA. Rob, uh, Reddit, it'll be will be an interesting one, but the the most fascinating one to me is chat GPT. how are they going to comply with these requirements that were meant for search engines. They're meant for Google and and Bing and those kinds of things. Uh and you got kind of a black box when you're talking about Open AI. >> Yeah, [snorts] it it is interesting from the the the LLM standpoint, but these tools work very much like a like a search engine. You can go ask ChatGBT questions just like you can ask Google questions and it will give you answers. Therefore, people are using it. Now, you may it may not technically be a search engine, but people are clearly searching it to get answers to the everyday questions that they would have asked Google or or you know or you know you know Bing or or whatever the case is. So, and you're also right Tom they they're reporting their own numbers and their own numbers are saying that you have now reached this threshold. So, there's not a lot of wiggle room that they're going to have on that. However, we know that they are a company and they're going to try to fight it because it's more conducive for them to not be on this list than it is for them to be. So, I would imagine they'll try to appeal something um just, you know, just in case they get lucky. But I I think that, you know, it's not just them. And I think any of these LLMs that get this big, they are going to be looked at like a search engine because if if you remember back to when they were putting the rules in for the DSA, part of part of the reason for this is that you're so big and you're so influential that you literally can shift how society operates based off of how you interact with it. So, I I do believe that even though sometimes it could be maybe a little bit overhanded that they're they're trying to say that if you get to this point, we've got to treat you different because if we don't, we've just been shown over and over again that you guys don't necessarily look out for, you know, you know, for society in the way that you should. >> Open AI is not going to uh appeal its classification, but I think you're right. It's going to try to modify what it's supposed to do. Uh, and this will be interesting because Google also has AI generated results in its search engine. So, Google has probably done some of this work that OpenAI can look to and OpenAI and Google may team up on like, well, when you're talking about a search index, we can show you the indexing algorithm, right? When you're talking about a large language model, you can't you cannot literally cannot show exactly how the large language model came up with something. What you can show is chain of thought reasoning. If that is sufficient for third party auditing and for the European uh Union's DSA regulators, uh then maybe that will be good enough. But it is a it's quite a different way of reporting versus the way this rule was put into place. This rule was put into place to say, okay, you've got a search index, you've got a spider, you go out and you do things. We're we want to see how your algorithm takes those page ranks and turns them into search results to make sure that you're doing something that isn't putting your foot on the scales or, you know, in some way biasing uh the populace or or giving them misinformation or something like that. Uh, I think I think JPT is going to have a much easier time with the misinformation side of things. Uh, because it's either it's either hallucinating or it's not and it's got a link to what it decided to do and it can say, "Yes, in these cases the link was wrong or in these cases we didn't have a link." And those have been getting less and less over time. Uh, I I think several independent studies have found you generally have less misinformation in an LLM's response than you have in your typical search engine where you've got a bunch of people gaming it with SEO. >> I'm not quite sure how governments are supposed to create laws because essentially what has kind of happened here, they they wrote a law for a tactic, a search engine. A search engine works in a very specific way. It can do these things. We now have LLMs that they can do some very very search engine type things but they're not search engines. The way that they're deriving to similar answers are a very very different way and the law doesn't necessarily allow for that. So I think that there will have to be a you a partnership between you know between the you know the government and between these large language model companies that ultimately will fall under this rule to here's how we're able to get to what you're ultimately getting to because once once again if you write a law for a tactic this is how it is done today um you know 10 years from today similar things are being done but they're not being done at all the way that you wrote the law for and that becomes problematic. Yeah, I I think the DSA isn't too bad in that respect in that it is very resultsoriented. Uh what are you doing to prevent misinformation? What are you doing to prevent manipulation of of elections? Uh and so then it's up to the company to say this is what we're doing. Uh but you're right. Uh a lot of that was was made with the envisioning of of how these tools worked. Uh and and the answer to your question, Rob, is what you're supposed to do is make a new law. Uh the problem is uh governments tend to move a lot slower than we would like uh in making new laws. So it's a lot easier to just take an existing law and say well what if what if we just kind of applied it? What if we tilt it at a little angle and make it work on this? >> Yeah. So I think ultimately they will figure uh they will figure this out. Um like I am I'm not mad at the DSA. I think that some good has come out of it. I do I do also think that in some cases it has been you know a bit overbearing but the truth is usually somewhere in the middle. So if the goal here is that, you know, that, you know, Open AI works with the government, the government works with Open AI, I think they'll get to some type of resolution that >> Yeah. I'd like to see some some studies of the effectiveness, you know, what what what changes, what what uh positive effects of the DSA and the DMA for for that matter, which is kind of a similar law in the what have they had? If if anyone knows of studies that have been done like that, I'd be curious to see them. Feedback at daily techchnewshow.com. DTNS is made possible by you the listener. Thanks to Paul Fe, Ali Sanjabi, AB Puppy, and Johannes. >> There is still more to talk about today. Let's get to the briefs. >> We have got lots of gaming news today related to how you get your games. Montto noted the story that Sony is fighting a class action lawsuit in a court in San Francisco where it must prove that reasonable consumers would not misunderstand licensing terms for digital games. Meanwhile, Eurogamer reports that Microsoft's forthcoming Project Helix console would support physical media as opposed to previous rumors that said it would be all digital. In an interview with BBC News at Gamescom in Cologne, Xbox CEO Asha Chararma declined to share anything new about Project Helix, but she did say, >> "Look, we've been hard at work on a great uh a great next generation and a great family of devices for Helix, and we'll share more soon." >> That made a lot of people speculate Xbox might continue to make versions that support disc. And that also led Eurogamer to report that Sony is considering pushing off its December 31st, 2028 deadline for making physical media versions of its games. >> Yeah. So, uh, obviously, uh, Sony fighting that class action lawsuit is only tangentially related to this. Uh, it's it's trying to fight the fight that a lot of companies have fought to say, well, we have it in the terms of service, uh, but the California law means that it has to be easy to understand. It's not going to change the fact that digital content is a license, not ownership. Uh, but it is bringing more attention to digital ownership, quote unquote, versus physical ownership, which by the way, you still also you only own the disc. You actually still are only licensing the game that's on that disc. >> Um, you could sell the disc, though. So, that that's a a tangible difference between that. And I think what we're seeing play out here is Asha Sharma very astutely noticing that people are mad at Sony for announcing it's going to stop making games 2 years from now and saying, "Well, shoot, maybe I should get out there and imply that we're not going to stop that." Uh, I'll leave my options open by making it pretty vague. Uh, but that's that's a great way to win the war of public opinion. And if you ever doubt that the war of public opinion ha has an effect, uh it it's not a guarantee to change things, but it can. And and we're kind of in slow motion watching that happen here. If you've got Sony going, well, maybe we should push off that date because we don't want to give Microsoft an opening here. >> I'm not a betting man, but I am a guessing man. So if I had [laughter] to guess like at some point >> at some point somewhere a discussion has been had inside of Microsoft like you know we're going to get rid of disc at some point but we haven't told Earth yet that we're going to do this. So let's just not say that yet. It would be it would behoove us to just not say the things that we're thinking about doing years down the road because maybe they weren't two years away, maybe they were three years away. But you know what? There's no point in telling Earth and getting everybody upset about something that's like maybe almost at the end of this decade if not the next one. So yes, this absolutely benefits Microsoft to just even if you were thinking about it, even if some leaks come out to where you were thinking about it, well, we never said it officially. Sony came out and said it officially. We're not trying to upset you. We're just thinking about it. We thought better of it and we're we're not necessarily reversing course. We're just not we're just not ready to tell you that that's what we're going to do eventually. >> Yeah. And Sony is going to have to stop making physical games. And the that that is the sad fact is people aren't buying them in greater numbers. every year they're buying them in smaller numbers. And what Sony did was, I think logically projected out like, yeah, by the end of 2028, uh, we'll have made enough that we can just have them in stock and that'll allow us to to ride out the demand until it's gone. Microsoft knows that's true as well, but like you said, they're like, well, Sony went ahead and talked about it first, so we don't got stun. [laughter] >> Yeah. Yeah. I I remember this little company called Netflix that they were You might have heard of them, Tom, but they used to just send you DVDs in the mail. Man, it was like it was like the coolest thing ever. And no point ever were they ever thinking about getting rid of DVDs until, of course, they decided to, you know what, I'm going to spin this DVD business into another business called Flickster. Run it for a year or two and then we'll kill it. Um, but it but there was a time when you would have never thought that that's what they were going to do, even though they were having those conversations for five or six years before they actually moved on. I would imagine that Microsoft's doing the same thing. >> Yeah. And and it's it's a matter of of you know whether Sony needed to make it public. I think I think Sony was just trying to be honest honestly like yeah we're going to stop. Uh and there is there is a it is not a popular public opinion to not like physical games. Uh and Asha Chararm knows that. So very very wise on her part. NASA launched the Nancy Grace Roman Space Telescope on Sunday on a mission to enter orbit four times farther from Earth than the moon, so it can deliver a view 100 times wider than the Hubble Space Telescope. A SpaceX Falcon Heavy rocket lifted off at 7:25 a.m. Eastern time and sent Roman on a trajectory to reach the L2 Lrange point in three months. The telescope is named after Nancy Grace Roman, NASA's first chief astronomer. Uh sometimes you may hear Nancy Grace Roman referred to as the mother of Hubble for her essential work on what led to the Hubble Space Telescope. Uh it should deliver its first images by the end of the year. The telescope, the Roman telescope is a 7.9 ft diameter 2.4 meter device donated by the National Reconnaissance Office. So it's a it was meant to be a spy telescope. Uh so they just took it, built an observatory around it. It wasn't like plugandplay or anything, but it sped things up. Uh and one of the missions for this telescope will be to look for evidence of dark energy, which would help improve or possibly disprove and adjust the standard model of physics that could be incredibly uh helpful for physics and for our understanding of how everything in the world works. Uh it will also be able to image exoplanets, planets around other stars. These images are high resolution. They say you would need half a million 4K TVs to display an image. Uh it will send down 1.4 terabytes of data each day for a total of 20 pabytes over the course of its 5-year mission. NASA created a platform called Roman Research Nexus to host the data on Amazon Web Services. Roman telescope was supposed to launch in May next year. So, it beat its deadline by nine months as well as staying within budget. Uh, just lots to love about the Roman Space Telescope, >> Tom. You know, we we tend to be able to look up at the moon almost almost every day or every night and see it. And we don't think of it as being as far away from us as it is, but it's pretty darn far away. It's not close. It's not like it's it's not like just you walk a half a mile and then you're to the moon. This telescope is four times further out uh, you know, from Earth than the moon. So to even think of that as orbit just feels kind of kind of odd because it is so far out. Technically it is it is you know it is it is orbiting but it's a long way out there. So it'll be really interesting to see the type of images that can come from this because they've gotten it far enough away that you can actually look at some stuff that you just we physically weren't able to see because of the proximity to you know you know to Earth before. >> Yeah. And I I don't think we understand just how small of a slice of the sky telescopes have been able to look at. I think Hubble could look at like.1% maybe 1% uh of the sky at once. Uh and this telescope, the Roman telescope, is going to be able to look at 12% of the sky at once. Now granted, that's still not all of the sky. That is a big big chunk of sky, which is why there's so much data associated with this. Uh and it's it's going to take a lot of processing to to look through that, sift through that data. But we're going to have so many astronomical and possibly physics related discoveries because of it. >> Yeah, this the sky is a big place. How big is it that there are an order of magnitude or two more stars in the sky than there are grains of sand on the earth. When you when you think about that, like you take all the sand on all the beaches and in the bottom of the ocean and say there's more stars in the sky that we can observably see. The sky is even bigger than that because there's so there's some things that are so far away that we just can't see the light from them yet. So we don't really know how big the the ultimate size of the universe is. Only what we can see the observable universe. But it is a big place. So the fact that we can even see 001% of it is is pretty impressive how big it is. That's still a lot. >> Yeah. And and this being able to see 12% is really going to be able to image galaxies and filaments. Uh if you if you know what those are. Uh it's it's going to be incredible. And if you are still sitting there thinking like wait if the images are that big how am I going to look at one they they will downsize some images they will make some a lot of lot of the images that you see from telescopes uh are certainly not all the data the telescope uh collected anyway >> guessing they might use some artificial intelligence Tom actually make you know to to some of that stuff down to make it >> they will be absolutely >> Instagram announced it will reduce the reach of any account that has an LLM generator persona that does not label itself as an AI generated profile. That tag is meant for profiles that feature an AI generated person. There's been a rise in video posts featuring influencers that are generated by machines. >> Yeah, this is one of those big stories out there uh that I think is catching people's eyes who use Instagram and have been frustrated when they saw a doctor telling them something and then found out later that's not a doctor, that was created by VO uh or something like that. And this is Instagram saying, "Ah, okay. uh people don't like that. We need to we need to put a label on that so that folks know the difference between someone who is human pretending to be a doctor and someone who is absolutely not even a human. >> There is an interesting dance that companies are having to participate in right now because there are some AI related things that people have just shown we just don't care. It's like a give me the AI more is cool. But there are some things that we care about immensely and it's it's like you didn't care at all about this. We moved literally one meter to the left and now you you you care to the point to where we're the most horrible company on the planet. So, so it's it's it's very interesting because the these things that you know where you are creating personas um even if you know it's kind of clear that they're personas when you're watching it. They're saying no, we're going to make this absolutely clear because we know that eventually it'll get to the point to where you won't be able to tell that this is not a human. So, you need to label it and if you don't label it, we're going to derank you. We're going we're going to downplay your video if you do not let people know. And that's the interesting part is if you label it, you're not your reach is not restricted. So so these these virtual influencers can still get the reach. They just have to admit that they're not people that they are an AI generated person. >> Yeah. And and a bit of the problem for them is that's if they are actually trying to play on people not knowing that they are AI that is that's just dishonest and and that's what that's what Instagram is saying. No, we we can't allow the dishonesty here. >> And potentially that reduces the number of those that that do that. uh when there are are perfectly, you know, fun entertainment reasons to have a virtual persona uh that some people won't mind as long as they know that it's virtual. >> Just let me know it's Johnny 5. >> Yeah, exactly. Johnny 5 is Well, Johnny 5 said he was alive. I don't know that these these things are actually [gasps] uh we are alive though. We we do live streams uh every week. Uh we do mailbag and we do DTNS Live. You can catch all of those by becoming a subscriber at youtube.com/dailytechne show. [music] Let's get to the quick headlines that are good to know. Going to make it look smart. Archive has extracted 12 to 13 terabytes of data from an endpoint on Valve's legacy infrastructure. The data is from between 2003 and 2013 and includes internal projects and third-party releases, including an early weapon asset from Halflight 2 episode 3 and Portal 2 concepts. >> Yeah, I I got to admit I got a kick out of seeing uh the early version of GLaDOS waking up uh that they showed. It's just it's fun. It's like seeing a behind the scenes or a making of where you see those early mockups and stuff. Uh, I know Valve didn't intend this uh to go out, but it is fascinating to look at. It's also Tim Cook's last day as CEO of Apple, and he sent a goodbye memo expressing enormous comfort that John Turnis is taking over the role and reminding folks that uh he he's not quitting. He's going to continue to work at Apple as executive chairman. He'll be responsible for Apple's relationships with governments around the world. Uh, basically, he's becoming chief lobbyist. Turnis' first big event as Apple CEO will take place on Wednesday, September 9th. Apple's expected to announce a foldable iPhone, amongst other things. Also, right before the show, we learned that senior vice president Phil Schiller is going to step down from his role leading the App Store and product events, but he's going to stay with Apple. Uh he'll just be working on unspecified initiatives. Microsoft is instructing users to ignore notifications that say Microsoft Defender anti virus is turned off. That may appear even if settings confirm Defender is working after the latest updates are installed. Microsoft says it's working on a fix. Also, the August 2026 nonsecurity preview update reverts your mouse settings and then won't you let you change them back? Is investigating that, too. >> At least that one's in preview. Uh but yeah, you know, it's our it's our regular uh Microsoft is working on a fix update. Schools and other organizations that use the paper cut print management software uh need to install a second emergency security update to patch two actively exploited vulnerabilities. There was a patch that came out, but then researchers found workarounds to those patches. So now there's a second patch. The OpenClaw Foundation released version 2.0 of its open source agentic platform, including shared cloud sessions, a smoother onboarding process, and a rebuilt browser app. The information has sources saying China's CXMT is producing its first batch of HBM 3 chips. That's a form of high bandwidth memory. And of course, high bandwidth memory in big demand for data centers. So that's an very profitable thing for CXMT to make. Now those chips are still a generation behind HBM4 which is made by Samsung, SKH and Micron. Uh and CXMT still probably three to five years behind those companies but catching up. This is a significant advance for Chinese memory makers who are being forced to get ahead because of restrictions placed on them by the US government. LG Energy Solutions has agreed to buy battery lithium carbonate from Smackover Lithium, a joint venture of a Norwegian and Canadian company that extracts lithium from the US state of Arkansas. >> And that's the opposite. China has restrictions on lithium. Uh, and so this is causing other companies to go, well, let's go find other places to to get the lithium. And they're getting >> I was just going to say this is one of those ones where it's almost like uh, you know, fair fair play is turnabout or turnabout. Fair play turn. Literally, we're doing this because of what's happening. So, yep. >> Huawei may be making a great comeback, gaining 20% of the market in China, but it is not immune to the memory shortage as it posted a steep profit decline for the last quarter. Revenue rose 10%, but net income fell 36%. >> Ramageddon is hitting everywhere. No one can escape it. >> Y OpenAI said his advertising business had reached a $1 billion annual rate. For comparison, Google's annual ad revenue is $294.69 billion. So only $270 times more than the company that just met some, you know, some some standards that is now going to be looked at like a very large company in in, you know, European Union. So >> I mean the the the advertising business is working for Open AI. It's just, you know, it's it's it's not going to be threatened. >> It ain't working like it's working for Google, though. >> Yeah. Yeah. They got a ways to go for that. Uh Nvidia is investing $3.5 billion into Taiwan's chipmaker MediaTek. Uh and in exchange, MediaTek will use Nvidia's Envy Link Fusion and NVHBM techch to help components communicate. And 18 months after passing 6 billion downloads, free and open source media player VC has passed 7 billion downloads. >> Ah VC, an oldie but a goodie. Glad to see it's still kicking. >> Yeah. Well, folks, we end every episode of Daily Tech News Show with some shared perspectives. Today, Bod has a trick for telling if a piece of sleep tech is reliable or not. >> Yeah. Uh, Bod writes, "Hi, DTNS friends. I love the sleep talk across the different hosts. Today, I was listening to Jason and Jen talk about devices designed to help with circadian rhythms. For the average person, it's hard to know if a device actually does what it says it does. Earlier this year at CES, the National Sleep Foundation, a respected organization in the sleep world, rolled out something called Sleepmark. The idea is pretty simple. NSF, in this case, National Sleep Foundation, not Science Foundation, independently evaluates sleeprelated products using scientific testing to see whether they actually do what they claim to do. Companies apply, have their submissions reviewed by staff who evaluate the claims against NSF objective criteria, consensus standards, and the foundation's published guidelines. The companies must provide documentation and formally attest that they have evaluated and can support their claims. The designation essentially means the product aligns with the National Sleep Foundation's principles and standards, but it's not exactly a technical warranty or a clinical guarantee of the manufacturer's specific health claims. Products that meet the criteria, though, can slap the NSF Sleep Mark logo on the package. I'm not saying someone shouldn't buy a device just because it doesn't have the Sleep Mark. The certification is still pretty new. There are plenty of good products that haven't gone through the process. But if a device does have it, you can feel a little more confident. Just thought it might be a useful resource the next time sleep tech comes up on the show. You can find more about the National Sleep Foundation at thensf.org. Thank you, Bodie. >> Yeah, you know, we we generally know Bod for, you know, being a firefighter or uh or more so talking about EVs and stuff like that, but that guy is really into sleep um sleep. So, he he knows the stuff here. He knows the stuff here. >> Yeah. Yeah. Uh, and and a good thing to know like and I like how he's like, "Look, this doesn't tell you it's foolproof, but it tells you that somebody looked at it and vetted the the claims and said, "All right, you know, the these might be worthwhile." So, so something to look for. I I appreciate that. >> Well, folks, we want to know what you're thinking about. So, if you've got some insights into a story, please, please, please share it with us at feedback ataily technew.com. [music] >> Yeah. Big thank you to Bodie for contributing to today's show. Thank you for being along for Daily Tech News Show. You're the folks who keep us in business. Become a patron, which you can do very easily at patreon.com/dtns. Helps us out. Help us out. Thanks. Bye. [music] The DTNS family of podcasts helping each other understand. >> Diamond Club hopes you have enjoyed this broker. [laughter]