Ben Whitelaw 0:00

So Mike, what have you been seeing on the internet this week? What's been making you smile? Tell me all about it

Mike Masnick 0:04

So, so this one, this may be very specific to a certain age group, and I'm wondering if you are not in that age group, 'cause I know you are younger than I am. So I'm going to tell you about something, and it is entirely possible that you will have zero idea of what I'm talking about. But I I have a question for you. when did you first start using computers?

Ben Whitelaw 0:25

Oh, God.

Mike Masnick 0:26

what year about?

Ben Whitelaw 0:28

Um, later than I would like to admit. I, it's probably kind of 2003, '04.

Mike Masnick 0:35

Okay. You,

Ben Whitelaw 0:36

actually that's a lie. I was at school in 2000 and I was using computers, but I didn't really have a good home setup until kind of

Mike Masnick 0:42

Okay. So then you've, you have almost certainly never experienced this. do you know what defragging a computer means?

Ben Whitelaw 0:48

Don't know, but it sounds very painful.

Mike Masnick 0:50

No. Yeah. Well, it was. So, in the sort of '90s, if you had a Windows computer, what happened is, like, as you would use it, stuff would get saved to the hard drive. But it was a hard drive. It was, like, the physical spinning hard drive before we had all this, fancy solid state nonsense that we all use today. And it would, like, just put all the data into different sectors, and that could be anywhere, literally, like, physically on the hard drive. And that would eat up space and make it awkward because there would be empty spaces, but it wouldn't be able to dump new data there because you would have empty spaces. So there was a thing that was recommended that you do, like, every year or so, which is you defrag the hard drive. And it's literally a process where you set it up. You're not gonna be able to do anything with your computer for hours, depending on how large the hard drive is and how much defragging it needed. And you would set this thing to go, and you would click a button, and you could... You would have this screen that would show you, a map with a whole bunch of pixels sh- demonstrating, like, where data was on your hard drive, and it would move it all, sort of push it all in so that, the beginning would be where all the data is and the end would be empty so you could keep filling, and it would give you back a bunch of space on your hard drive. But it was something you would have to set aside. It would make your hard drive make noise 'cause it was constantly spinning and it was very noisy, and it was a whole process, and I had completely forgotten about doing it, though I used to do it basically once a year when I was a kid to clean up my hard drive and get more room. But there is a website that lets you relive that experience called defrag98.com, because it was a Windows 98 thing. And if you go to that, you can see exactly the experience that some of us went through, including you can, have it make noise as if it was a hard drive defragging. You can pick which drive you're defrag- This is, like, the exact experience. Like, literally, this is the Windows 98 defragging

Ben Whitelaw 2:47

It, it looks like Minesweeper.

Mike Masnick 2:49

Yes. It has that sort of mi- there's all these boxes, but that is data or empty sectors, and you're moving data. So it would, like, copy data from one part of the hard drive and place it on another, and it was a process that for a period of a few years you kind of had to go through if you had a Windows computer in the '90s, to keep it working in, in good working condition. And this just brought back that kind of, uh, memory and nostalgia.

Ben Whitelaw 3:15

It, looks, quite mindful, I'll be honest. It looks like

Mike Masnick 3:18

Yeah, it, it could be... It's, it is a, a sort of meditative experience to defrag your computer. There was always the slight fear that defragging it would mess something up and destroy your hard drive and everything, but yeah.

Ben Whitelaw 3:30

Well, I mean, this m- my-- similarly, my kind of joy from the internet this week is derived from a website which, which may not make a lot of sense to you either. I mean, it, it barely makes sense to me. But there's a, a, tool called MVNT.studio that's been created by a bunch of Korean v-video creators and enthusiasts, I think. And what you can do is you can drop any YouTube link, SoundCloud, or file into this tool, and it will create you a dance move. and you can pair that with a bunch of avatars and characters, some of which, you know, I'm sure it doesn't have the ability to actually use. but you can get kinda Bart Simpson doing a dance to your track, whatever it may be. you know, it's pure Korean internet gold, I would say. Um, just a few weeks after, you know, you mentioned about those, websites that allowed you to deliver food that doesn't exist. You know, it's, it's something only could really work in Korea. Um, and

Mike Masnick 4:34

think, both of our, sites this week are, are generationally specific.

Ben Whitelaw 4:38

Yeah. Our listeners will like one or the other, probably not both. Hello and welcome to Ctrl Alt Speech, the podcast where we make sense of the major debates shaping online speech, platform power, content moderation, and the future of the internet. It's September the 3rd, 2026, and this week we're talking about what a giant ID leak means for online age assurance, the tension at the heart of AI companionship, and for our extended Patreon episode, how AI is the social media era all over again. Ctrl Alt Speech is the not always hope-filled podcast about our technological future, co-hosted by Mike Basnick, founder and editor of Techdirt, and me, Ben Whitelaw, from Everything in Moderation. Mike, I don't want to do any dance moves. I, I'm, I'm too old for it. you know, my hips are sore. Um, you know, not even having a tool like MB&T kind of suggest something for me is gonna, is gonna make me do that. But that's, we've made a venture into short-form video ourselves on Ctrl Alt Speech.

Mike Masnick 5:47

Yes, we have.

Ben Whitelaw 5:48

not dance related, but, uh, sharing some of the

Mike Masnick 5:51

Not yet. Not yet. Give us time.

Ben Whitelaw 5:54

exactly. We, we-- I think a lot of people have been asking us, like, "Why don't you share video of the podcast? Why aren't there clips that we can see of you handsome fellows talking about the internet?" And I guess we've, we've held off for quite a while, haven't we?

Mike Masnick 6:06

Yeah. I mean, you know, video is, is always a, it's a different medium, and there's, there's work that needs to be done in order to do video. But, we're dipping our toes into the video space, uh, which is that we are not releasing full videos of the podcast, because n-nobody wants to watch us. I mean, you know, the live, the live recordings, sure. but I don't think that people wanna see the recording of us every week.

Ben Whitelaw 6:34

if you do, listeners, let us know

Mike Masnick 6:37

but clips. we are experimenting with, various length clips of the podcast, of different sections of the podcast with the two of us talking, and we're putting them all around social media. It's a scavenger hunt. Uh, you can go find various clips in different places. I think we're experimenting certainly with Blue Sky. We'll have some clips on Blue Sky. we have some clips on YouTube. We have clips on Patreon, if you are supporting us on the Patreon. and Instagram where we have now set up Ctrl Alt Speech accounts as well. So you can find a Ctrl Alt Speech on Blue Sky and on Instagram

Ben Whitelaw 7:19

And, and all of these accounts will be in the show notes. I guess, you know, our ask for our, our listeners is, go and s- follow the accounts. we know that, you know, not everyone is on, on those social media platforms, but if you, if you use them regularly and you want a bit more control of speech in your feed, go and follow them and share the clips with people that you think might find them interesting or useful. you know, it really hel- does help us grow the podcast. It's one of the many ways that you can support us without becoming a Patreon supporter, which is, one of the big ways you can do

Mike Masnick 7:46

And these are, these are not mutually exclusive. You can do both.

Ben Whitelaw 7:50

True, true. Do, do the free things and then, stump up for the extended episodes, of which we, we should talk about. We got a few Patreon supporters this week, so, uh, big sh- big shout-out to those people who, who signed up. we are no longer doing our Trust Con special code, so we're back up to full price, but it is very much worth it, and we've got our, two tiers. We've got our founder and insider tier, who pay $20 and get extended episodes, early access, and the chance to suggest stories and episodes that we should cover and produce. We've got a few really good ideas from our, Patreon supporters this week. Plus our supporter tier, who pay $10 and get the episodes and the early release time. So lots of reasons to support via Patreon. you can go to for more. also to note, Myke, you know, you've got a new camera set up because we're now doing more video. I got a haircut because, you know, last week was, last week was, not acceptable. So, you know, this is, this is becoming a professional operation.

Mike Masnick 8:53

I wouldn't go that far yet, Ben.

Ben Whitelaw 8:56

cool. So we had a bit of tr- a bit of difficulty this week, I think, working out what stories we wanted to cover. it's been quite a fragmented news week, I would say, compared to last week when we had the Meta settlement, and it was very clear that that was the big topic of the week. but we, think there's a lot to say about these stories and, and we expect listeners won't necessarily be fully up to speed on them. So you're gonna start us off, Myke, with one that's, like, fairly, fairly niche or perhaps maybe hasn't broken out into the mainstream news as, certainly the Meta settlement did last week.

Mike Masnick 9:26

Yeah. This one, but I think is, is actually a really big and important story, and it comes from Brian Krebs, who has been a long-term cybersecurity reporter. I think he worked for The Washington Post many years ago before striking out on his own, has been very successful, has, uh, his own website, Krebs on Security, um, which does incredible work. He's breaking really interesting stories all the time. But this week he broke a massive story, about, a leak, apparently a real-time leak of scanned IDs from a company called IDScan.net, which is apparently one of the bigger ones that is mainly focused on, like, real-world ID scanning. So within the leak, there were 153 million driver's licenses. There were a number of millions of other kinds of IDs as well, and he found his own, driver's license in there, and then, figured out the date that it had been added and realized that he had rented a car from Hertz Rent-A-Car, that day. and then found some other people in the database who he knew and asked them to check what they were doing on the day that it was added to the database, and a lot of them were renting cars. So it appears that a lot of it was from, scanning your ID at a rental car place. but there were other things where, like, there were some pot dispensaries and I think, maybe Target, uh, used them as well. So there are a number of different, offline places that when they would, verify your identity, they would scan your ID, and that would go into this database. And apparently, hackers had access to it for over a year, and basically every day were adding, you know, maybe a few hundred thousand different records to this, which is pretty incredible and pretty scary. The point that ties it back, there are a few different ways that this ties back to the, the internet world, one of which is the ongoing debates and discussions that we've been a part of as well about age verification online. And as I've said and other people have said is that age verification is really identity verification. And in fact, if you look at this company, IDscan, they're very much also in the age verification world. They are promoting their tools and services for age verification. They have actually a fairly, comprehensive and in some cases useful website that tracks all of the different age verification laws for social media. they clearly see that as a, growth field for them as well. they've written about some of these laws. They had an article all about KOSA, the, proposed Kids Online Safety Act in the US, which would, even though they claim it wouldn't mandate age or identity verification, realistically would, heavily encourage it at least. and so you have this vendor that, uh, presents itself as one of the top vendors in the space. They have a whole page, a whole... It's literally called their Trust Center, that talks about all the ways in which they keep the data private and secure, and it's certified by all these things, and they have all these banners about how it's GDPR compliant, it's CCPA compliant, it's compliant with all of these different standards. And yet, for over a year, apparently hackers had full access to the database and were extracting every record that was uploaded basically in real time.

Ben Whitelaw 12:47

Pretty shocking stuff

Mike Masnick 12:48

Yeah, it's, it is incredibly shocking, and you could basically buy anyone's ID for 100 bucks, if you wanted it, and,

Ben Whitelaw 12:55

including Pete Hegseth.

Mike Masnick 12:57

Right. So one of the ones that, that Brian found was Pete Hegseth's, and you see this very clear image of Pete Hegseth's driver's license, uh, and you can pick it up for 100 bucks.

Ben Whitelaw 13:09

wonder what car he rented. That's what

Mike Masnick 13:11

Yeah, who knows? Who knows? But, uh, you know, I mean, within hours after Krebs reported on it, the site where you could see all these things, which was on the dark web using Tor, disappeared because of course they realized they were about to get scrutiny and they, took it down. But they still have access to all of that, data. one hopes that IDscan has figured out where they were getting it from and shut down that path, but we don't know quite yet. As far as I know at the time of this recording, IDscan hasn't said anything in particular, about it, which is not great. apparently the FBI is investigating this now, so it's serious enough that the FBI is involved. I'm not sure right now how competent the FBI is, but, you know, they're involved and they're investigating it, which I think is important. but, I think this is really important for part of the age verification debate. Now, obviously, you can say, like I've been very critical of a lot of the age verification mandates and the a- all of these demands, whether they're, child social media ban, teenage social media bans, or requiring age verification or age estimation or whatever. But I think this is important because one of the points that lots of people have made is that when you're doing identity verification in any form, especially online, you're creating a privacy hazard. There is a risk. Someone has to be collecting all of this data, which is valuable, which makes it a target. And, some people have even said, "Well, you know, offline, identity verification isn't so bad because you just hand somebody your driver's license, they look at it, and then that's it. And, you know, then they hand it back to you and it's forgotten." We're already seeing that's not true. this sort of proves in a lot of cases that's not true. It's being scanned into a system which is then being connected to the internet, which then allowed hackers to exfiltrate the, the entire database. but in the online context, obviously it is connected to the internet, obviously it is being stored somewhere, and as much as the different vendors say they're keeping the data private, we have to contend with the fact that that is not always true, and you've created a massive target, and hackers are very good at figuring out ways into these different things. And we should, and the people who are pushing these laws should be willing to contend with the risk that they are creating and the liability that they are creating and the, you know, potential negative impact on the people that they're, you know, whose identity is verified, where the claim is always that it's being done for their safety, and yet you're increasing their risk in a different way. And I think this story is so massive and so complete that, it has to be reckoned with.

Ben Whitelaw 15:50

Yeah, I wa- I wanna come back to that 'cause I have some, thoughts on that. But I think it's important to point out, since we are a podcast that, cares about nuance and is important to, to be clear about, we say there's, there's age identification, which comes as a result of an ID or some sort of government documentation, and then as the industry define it, there is age, assurance or age estimation, which again, is certainly age estimation is a different thing that often uses kind of biometric data, a face scan or some sort of, uh, voice, often of a kind of voice signal to estimate how old somebody is, and that, facilitates access to a site or to an app. Now, we know there are problems with both of those. We're gonna talk about kind of primarily the former here since it's, this, uh, system had government IDs within it. We know that there are problems with the latter, the accuracy of it, the inability to s- estimate the age of particular demographic groups. There's a whole bunch of, of issues there. but you know, I feel like listeners might be thinking, "Well, you know, those two things are different." and we should kind of be clear about what we're actually referring

Mike Masnick 16:50

Yeah. So, the one point I'll make on that is, you know, Eric Goldman wrote a paper last year, about a year and a half ago, which makes the argument that they're not really different. They're all backstopped to the same thing because as you note, like, the other tools, the estimation tools, you know, and I have said this recently, like, they can tell the difference between, like, a 60-year-old and a 15-year-old, but, for the ranges and the s- specificity that you need to, like, figure out how do we have a different experience for a 16-year-old versus a 19-year-old, those don't work all that well. And so even in cases where you can do some of that, they always backstop it with an ID verification at some point, at some point in the process. It's not saying that everyone has to share their, government-issued ID or anything like that, but at some point, all of these systems have a point where you can upload a government-issued ID. and I feel like oftentimes this, Oh, we have all these other systems," whether it's, like, the face scan or the voice or whatever, which we all know has, real problems, all of them backstop it with some sort of scan your government ID at some point. And they're not necessarily as transparent as they need to be about how often it gets to that point, but it does. And so that is a part of this process no matter which of these tools you're

Ben Whitelaw 18:10

Yeah, that's fair. I mean, more and more platforms, app services are gonna have to go down an ID route because of, of the regulation that we're talking about. I'm just thinking that the, uh, the industry itself thinks of itself as, two things, whether that's fair, fair or not.

Mike Masnick 18:26

Sure. And, and I mean, like, again, Eric Goldman's argument is that that, that's all marketing to some extent, right? they're doing that on purpose because they know that the ID verification part of it feels really scary, but if they can switch it and, and say, "Oh, this is just age assurance or age estimation or whatever," it doesn't feel as scary. And his argument is that, you know, that is marketing, that is a sleight of hand to sort of distract from the, reality.

Ben Whitelaw 18:53

it. So, so let's talk about the kind of reaction to this then, because my, issue here, as somebody who is a skeptic of age assurance for the same reasons that you mentioned and, and the same as many of our listeners, you know, is it's not accurate. It's a terrible user experience, which we often forget. but it's this idea that it doesn't feel right to kind of make the connection necessarily between this kind of big, mostly offline ID collection service and the age assurance, age verification piece that we, can obviously see coming online. it's a bit of a leap really to say that one thing is going to necessarily lead to the other, in the way that lots of people have characterized it on social media over the last week or so. And again, I, I get the need to be cautious around age assurance and verification, but a lot of people using it, as often people use stories like this as a kind of reason to jump straight to a solution, which is to not have anything kind of at all. and so it felt a bit like kind of point scoring. It felt like this is the kind of perfect story, and, and some of the quotes in the, the Brian Krebs piece actually made that connection very quickly, in a way that felt a bit weird. You know, we weren't, we're not talking about necessarily online age assurance. We're talking about this offline piece, this, this company that has legally collected and stored for much longer than it seems to have told the people who use it, data. we need to, we need to discuss what, what that industry looks like and then draw the parallels, I think, with the age assurance question that we're facing, and it, and that, it felt a bit, point-scoring.

Mike Masnick 20:29

Yeah, maybe. But I mean, again, like the argument that I would make is that for years, one of the things that we've heard is that the offline in-person identity verification piece of this was clearly much safer and much more secure than the online one, right? I mean, even people who were worried and concerned about online age verification were like, well, you know, kept trying to distinguish it from like, oh, you have to show an ID to go to a bar or whatever. And yet why I think it's not necessarily just point scoring but a legitimate complaint is you're saying like even in that situation, even in that situation that we've been told for years, even among critics of online age verification, that okay, well that kind of identity verification is okay. Now we're finding out that even that is not. And again, this is not even the first time that this has happened. There was a case a few years ago in Australia where they had a law that required, I, I forget the exact details, but it was like bars and clubs I think it may have only been in one part of Australia, that they all had to be linked up to a, an identity verification system so that as you went in it wasn't just the bouncer, you know, looking at the ID and deciding if you're old enough to enter the establishment. Rather it was scanned by a computer and that system was also breached and all of those IDs were leaked on the dark web as well. so we've had this before. But the larger point stands that any system you have that is somehow scanning IDs and collecting that information in order to verify you, verify your age, verify information about you, is a risk, right? It is collecting private personal data and that alone becomes a target. And so therefore like yes, I get it, like people who are already upset with this including myself are using this as a story that, kind of highlights how dangerous this is. But I don't think unfairly. Like yes, it is a different situation, but this was the situation that was supposed to be less bad and now we're finding out it was really, really bad which should raise more questions about the online age estimation or age verification systems that are out there because we don't know that they're any safer because we have no reason to believe that they're any safer because this was supposed to be the safest version of

Ben Whitelaw 22:49

Yeah. I suppose, you know, it makes sense for a car rental company to take your ID 'cause,

Mike Masnick 22:54

you

Ben Whitelaw 22:55

that. Yeah. but there's a bunch of companies that this ID scan company works with that you question why they're working with them necessarily. you know, Target is in there, all right? And, and like yes, probably buying alcohol is involved. but again, why is that data being kept on file? Like that, that's the fundamental question is like why are we not furious at companies like Target who you buy alcohol from in one second or you buy weed from, if you're a pot store, in that second, and then it's-- you give them your ID and that data is kept for seemingly months at huge risk to yourself and at the risk of, of a leak like this. Like, to move so quickly into-- And I guess maybe it's, because w- we, have control over what happens with online regulation in a way that we don't with these systems that are already set up in the real world, so maybe it makes sense to concentrate on that. But I would-- I'm just like, if I was a North American, you know, if I was a, a US citizen or a Canadian citizen that had data in this database, I'd be like, "Why, why is that happening, first of all?"

Mike Masnick 23:57

Yeah. I mean, again, it did feel like these hackers had real-time access to it, so it wasn't even how long they kept the data because it appeared that they were able to get it out of that database, every day. so even if, you know, who knows, maybe the company did delete that data after a certain period of time, didn't seem to matter because the hackers had already gotten access to it. but yeah, I mean, I think the underlying point is that, any collection of private data becomes a risk, and we're concerned about it when, you know, and there are lawsuits and all these other things when Meta or Google or whoever else, whichever tech company is collecting private data, we're demanding that they keep it secure in all sorts of ways. And there are all these other companies out there as well that are collecting, in some cases, even more secure and more private data, and there's nothing sort of dealing with that. Like now we have the FBI investigating, but again, it's been over a year, and this is a way that, I think most people didn't even realize. Like, when I've rented a car, and of course I hand over my license at some point, but I'm not sure I even realized that they were scanning it. And, you know, Krebs says that in the article where he talks about it and he's asking people and like, you know, yeah, you hand over your driver's license, but nobody remembers that it is actually being scanned. Or maybe, you know, like I'm trying to think the last few times I've-- Yeah, like I've remembered it being photocopied, right? but did I realize it was being sent to a company that had like half of North America's adult driver's licenses in their database? Like, no, not at, not at all. and that's kind of terrifying, and this is why anytime we're doing anything that mandates the collection or verification of private information, we should be really deeply thinking about the implications of it and how, secure is it really because there's a risk. So yeah, like maybe it's a little bit of a cheap shot to, tie this back to the online age verification debate, but to the extent that it should raise the alarms about how are we doing this? Do we really trust these companies? How do we know that they're keeping the data secure? How do we know that the data isn't leaking? I think all of that is valid, and this is just a really clear example of why we're raising these concerns because, you know, lots of people say, "Oh, you know, this is all just, Chicken Little claiming the sky is falling. You just don't like this technology. You don't like the way it's, being implemented." But no, there are real reasons why people are concerned about it, and this is a, I think, a really good example of why people should be concerned about it.

Ben Whitelaw 26:33

Yeah, definitely, definitely scary and definitely worth noting. I mean, I think cheap shot is, is a good way that, of describing it, but it does merit a lot more, a lot more study and a lot more thought. And hopefully we'll get more details as this company, does an investigation and lets us know what's, what's gone on. it links quite neatly, Mike, into a next story. Uh, a couple of eye-catching stories that I think we, believe go well together. and the first one includes We've talked a lot about OpenAI's approach to teen safety and some of the work it's done over the last year to create defaults for teens that, teens and their parents can opt into in order to have a safer experience. In the last week, it has made those voluntary protections as part of a new product that it's calling ChatGPT for Teens. And the way it verifies whether somebody is a teen, somebody between 13 and 17, is by looking at behavioral characteristics, things like what you search for, the time of day that you're online, and also, you know, your kind of account history more generally. So there's a kind of question there as to whether that's gonna be accurate, which we can, we can discuss. the good news, I guess, from, for people who are concerned about the way that children use these chatbots is that ChatGPT is thinking, a lot about kind of how to encourage critical thinking amongst teens. There's been a lot of research about whether, you know, we're kind of removing people's ability to think critically, particularly children. And so this new product has a kind of study mode that doesn't give the answer straight away as, as a lot of the chatbots do, and encourage teens to think, you know, over the course of a few responses, what the answer is. Now, don't know what you thought about, about this as a product launch. You know, we... Do you feel a bit cynical towards this? Is this a kind of... This is a, as a, kind of lipstick on a pig situation?

Mike Masnick 28:25

I, yeah, I mean, it's interesting. I, I would love to see the research into this. it's funny because right before this came out, I had actually been reading about a separate thing that, OpenAI had done in Estonia, where the Estonian government, I think this is like a year ago or so, had worked with OpenAI to generate a, academic version of ChatGPT, which was specific. You know, it made sure that the data wasn't being sucked back up in for training, was kept in Estonia, I think. But one of the things that it had was it sort of required a Socratic method. Uh, you know, if you went to ChatGPT and said, "Summarize 'Catcher in the Rye' for me," you know, as a high school student might do, it would come back with you, "Well, what do you think?" It's, uh, you know, and sometimes that's appropriate and sometimes it's not. But, you know, when I, read about that and I was like, "Oh, that's actually interesting," because so I have this- theory that I've talked about in the past that I think, the absolute best way to learn a subject is to teach it. and this was, like, personal experience, right? You know, when I was in college, I studied statistics, and I took multiple levels of statistics until I was, pretty far in and, you know, could do fairly advanced statistics work. And then I taught incoming freshmen statistics. When I was, a senior, I taught a class, for incoming freshmen, and I-- that's when I learned statistics. I thought I knew it going in, but because the students would ask these questions, I'd be like: "Oh, wait," you know, I really h- you really had to deeply understand it. So that experience has always stuck with me as an example of, of really learning something. And so one of the things I've always hoped for in the, the AI world is if we could replicate that experience, if we could figure out a way to have the AI be a student more than a teacher in a way that then forced you to actually learn the subject. So that part of this is interesting to me, right? You know, if ChatGPT actually acts that way, in a way that sort of, pressures you into really learning the subject, that could be really interesting. Whether or not it works, we don't know. Whether or not the implementation actually sticks, we don't know. Whether or not kids will use it in a way that, that encourages this behavior, we don't know. And that's things that should be studied. But, like, as an approach to dealing with teen users who are just trying to use it to, get answers and to cheat makes it interesting because we know that that is happening, and we know that that's bad, and we have research now showing that students who are doing that don't really learn the subject and perform much worse on tests, and that is a real concern. But if the technology can actually sort of force you in a way to really try and learn the subject, that's really interesting. How do you balance all those things? I don't know. You know, there are cases where you don't want it to do that. if somebody actually needs help, right, you don't want it to go into the Socratic method at that point. there's a whole bunch of questions about how that actually plays out, and I think... So it's interesting to see it happen, but I, you know, I don't know how well it'll work.

Ben Whitelaw 31:39

I mean, I'd go one step further, Mike, and say that I would like and hope that OpenAI has that data now

Mike Masnick 31:46

Yeah

Ben Whitelaw 31:46

and should release it as part of these, product launches in the way that, we're starting to see other companies do when they release products or models. they must have, based upon that Estonian experience, have some data that shows that teens are, their usage is more effective. They're getting more out of it whilst they're using the chatbot, that when they encounter an issue, that the issue is less severe or that they're able to kind of find their way, out of it. and also the, this age assurance piece, you know, they must have data that shows that the system can identify teens and, you know, have a proportion of people who have tried to evade it by putting on a, a mustache or using a photo of a, of their sibling. So I think all of that data must exist, and yes, there should be more research on it, but I would like to think that that data, is available and should be published when these announcements are made rather than, you know, it being like a kind of, I think one piece I, I read described it as like akin to putting a ribbon cutting ceremony on a building that's yet to pass a safety inspection. Like, that's kind of what what it always feels like

Mike Masnick 32:51

Yeah. And I, I think, you know, I can understand why they might be afraid to release some of that data for competitive purposes, but this feels like the kind of thing that the industry itself as a whole is better off if everyone sort of recognizes what works in this scenario and what doesn't. Because, without that, then, you know, if, if Anthropic and Claude do something bad for kids, that's gonna reflect poorly on the entire industry. And this is a space where if they were more transparent about what actually works in encouraging kids to actually learn or even, like, encouraging kids to put down the chatbot at certain points, I think that would be useful if we had that, that information. But, yeah. I, I find this a, a curious and interesting experiment, but I'm not sure if it's... It's, it's tough to tell how relevant it is until we have data.

Ben Whitelaw 33:40

Yeah, for sure. an interesting Washington Post piece kind of dovetails neatly with this, doesn't it? one of the big questions is how teens use AI for emotional support and this Washington Post piece doesn't speak specifically about teens, but talks about the, the increasing reality of people using these tools for that

Mike Masnick 33:57

Yeah, and it's based on this academic study that looked at, how people, adults increasingly were using AI for emotional support and found that a lot of people are, a large percentage, I think it was like 37%. I'm, I'm suddenly blanking on the percentage 'cause there are a couple different things. They looked at how they were using the different AI tools. and the interesting thing to me was the reasons that people gave for using the chatbot functionality for emotional support, and one of the clear ones, and we've heard this before, but it was interesting to see how many people said, was that they knew that they could speak to it without facing judgment. Because it wasn't human, they actually found it more valuable. And in most cases, it feels like people, they know that... Like, there were, there was a lot of fear early on that like, oh, people won't realize that you're not actually talking to a human, and that is a problem. And here, people recognize that they're not talking to a human, and that is seen as a benefit, that they feel they can open up to it more. It's different than... You know, and so there's always people who say, "Well, you know, you have issues and, you know, you need emotional support, you should, go to a therapist or go to a professional or, talk to a good friend or something like that." But what, the responses to the survey said is like, there are certain things that people just don't feel comfortable talking to other people about, and yet they feel that they get benefits from this. and I know that that still horrifies some people, but I think it's actually important to recognize that that is the reality for so many people, that they gain comfort from this, and we have to deal with that reality rather than pretending it's not a reality or that it's somehow a bad thing that some people find that the responses that they get here help them.

Ben Whitelaw 35:41

it is really interesting and I, and I, you know, I've, I've actually used it for, not regularly, but for kind of one-off similar instances. I don't always think it gives the best, response, but I don't think I'm probably of, of the generation, even if I'm slightly younger than you, Mike, as, as previously discussed, that, that, you know, seeks the validation that the chatbots provide. but I do wonder what life without judgment looks like. You know, there's this idea that judgment is bad, but I often think and-- about what, what I would do if I wasn't being judged. You know, if you weren't judging me for the stories that I was bringing to the podcast, if my wife wasn't judging me for what I was wearing when I came down in the morning, you know, what people would say if they saw me when I'm left alone i- in an evening to myself and what I watch on TV. You know, judgment is, is perceived to be bad, but-- and I think there's probably a generational feeling about how, about the benefits of judgment and what it does

Mike Masnick 36:40

Yeah. I don't think the summary here is that we're moving to a judgment-free world. I think that there are some things that people bottle it up inside and are afraid to talk about it because they're afraid of judgment. And in those situations, this can be helpful. Because realistically, what happens right now is people, get very deep inside their own heads on these things and, they do have this conversation, it's just with their own internal voice, for people who have their own internal monologue. And I recognize that some people actually don't have an internal monologue. I do, and like, if it's something that is judgy, like that can actually be psychologically damaging if you're sort of always criticizing yourself as well. So there is some benefit in being able to talk through things. And that's not to say that like there should be no situations in which judgment is important, but I'm, I'm not convinced that like having a space where you can ask a judgment-free zone question means that there aren't other scenarios in which judgment is still there and is still important. And, it's not one or the other. These aren't mutually exclusive situations. so I don't think... Like, it is worth considering, right? And there may be some people who sort of retreat into this world where they're so afraid of judgment in any other context that they don't interact with any other humans. I think that is a larger problem, but I think that is probably narrowly scoped to a small group of people who, need real help and we need to be able to deal with that. And in those cases, like it would be good if the systems themselves recognize like, oh, someone is now over-relying on this chat tool, then that's a problem and that should be a signal and that should be diagnosed and, dealt with and, try to move people to, to get real help. But I think the situation, which is what, this article is really talking about, what the study is really talking about, is the certain one-off cases where there's a particular topic that you don't have anyone that you feel comfortable talking about with, and so therefore you turn to a chatbot. And we've seen different examples of this, and I had mentioned this a few weeks ago where it was like former guest host on the podcast, Bridget Todd, talking about her experience with, her, her family and, her parents being ill and, and passing away, where she just needed something to talk to, and it turned out that ChatGPT was really helpful. And I had recently seen another, and I'm blanking on the name, but it was a really interesting, very long video on sort of how to critique AI and was very critical of different AI systems, but at the end she was saying like, I'm here explaining all the problems with AI, but there are times where it is really useful and we have to admit that." And she talked about how she was being evicted from her apartment and she needed the AI to sort of help her through it because it was like such a traumatic experience that she was unable to process it on her own and she couldn't go to anyone else and the, and ChatGPT or I don't know which model she was using actually helped her deal with the situation. And I think we have to recognize that there are those situations where the tools can be helpful. The problem is when people use that to deny reality and hide from reality, that's not good. But o- there are a lot of sort of edge cases and corner cases where it is a useful tool that helps get you through something and we should acknowledge that rather than determining that it's, it's all good or all

Ben Whitelaw 40:00

Yeah. I suppose in the same way that the internet has always been. there have been, you know, people's experiences on the internet shared in forums and in listservs and in groups, and those are incredibly helpful in times of, crisis and, a lot of people are, are in that increasingly, I think, in, in the world. So yeah, I think there's, an interesting tension there which we will probably chart over time. And, and, you know, this week it's worth mentioning that OpenAI was designated a very large online search engine by the European Commission, means a VLOS, um, the, the sibling of the VLOP. And, you know, we won't go into the details about whether it is a search engine or not and what, that means, but it does mean it has kind of stricter obligations around transparency, around risk assessments, of which, teen safety will be very high on the list, I think.

Mike Masnick 40:52

Yep. Yeah, it'll be interesting to sort of see what, comes out of that,

Ben Whitelaw 40:55

Yeah. talking of judgment-free zones, Mike, we're gonna go to our Patreon segment now. See what I did there? f-free listeners of Ctrl-Alt-Speech, thanks very much for joining us this week. we've got a great segment coming up which you should definitely tune into. if you're not a Patreon supporter already, go to patreon.com/ctrlaltspeech. We're gonna talk about how we're all the same mistakes of the social media era using a couple of great stories from Rest of World. Definitely worth tuning into.

Announcer 41:24

That's the end of this week's free edition of Ctrl Alt Speech. To hear the extended episode with additional coverage, become a Patreon supporter at patreon.com/ctrlaltspeech. That's C-T-R-L Alt Speech. If your organization is interested in sponsoring the podcast, contact us via our website at ctrlaltspeech.com. Thanks for listening