Ben Whitelaw 0:00

Come on then, Mike, tell me what fun you've had or seen on the internet this week

Mike Masnick 0:05

Well, here's a fun one. the NBA championships, I don't know if you're a basketball fan at all. Are you? Do you, do

Ben Whitelaw 0:11

I wish I was a, I wish I was a basketball fan. I'm not.

Mike Masnick 0:15

Okay.

Ben Whitelaw 0:16

Te- teach me. School me on b- what I need to

Mike Masnick 0:19

basket, the, finals, the NBA championship finals began last night. and the New York Knicks, which are my team from growing up, but who literally the last time they won a championship was the year before I was born, are in the finals, which is a little bit exciting because, when I was a kid they were okay, and they had a few good years, but, uh, it's been 20 something years since they were even remotely good. I mean, I guess last year they were all right. But they're in the finals, so I'm actually a little excited about it. But...

Ben Whitelaw 0:52

what's happened? What's the special, the special sauce? What have they done?

Mike Masnick 0:54

Uh, they're just getting good players. I mean, that's, that's the trick.

Ben Whitelaw 1:00

That old thing

Mike Masnick 1:02

but the thing that I found this week online, which is perfect for the NBA championships, is a site called 82 and 0, 82-0.com. There are 82 games in a season in the NBA, and the idea is that you sort of spin a wheel, and it presents you a team and an era. So some decades, you know, going back to I think the '60s. I don't know how far back it goes. and you have to pick one player from that team and that era and build out a team, and they have some system that will determine if it can go 82 and 0. So the goal is to try and build a team that would win every game.

Ben Whitelaw 1:38

Nice

Mike Masnick 1:39

it is difficult, and I have not gotten a team. gotten teams that were, you know, good, but I've not gotten a team that is close to 82 and 0, I do not have that much s- free time. But I am spending a lot of time trying to build a good all-time basketball team.

Ben Whitelaw 1:57

That sounds like an incredibly valuable use of anyone's time. Um, do you need to know about basketball? Can, can I have a go? Do you think I could

Mike Masnick 2:05

could go, you could go. You can figure out, I mean, you can figure out the basics pretty quickly. I mean, you want a team that scores a lot of points, but you have to be able to match positions. So, like, I was playing with it earlier and, you know, certain positions tend to score more points. So if you get someone who's kind of good, you put them in. But if you fill that spot, you can't fill it again later. So Shaquille O'Neal being a very famous basketball player, so I... But he's a center, and I just got him, the, you know, the period when he was great, except I'd already filled the center position with a less good player and could not use Shaquille O'Neal. It was very disappointing.

Ben Whitelaw 2:40

uh, I'm into, I'm into fantasy football games, so that will be, that'll be up my street.

Mike Masnick 2:44

There you go. Check it out.

Ben Whitelaw 2:46

I will. I mean, am I, I can, can't confess to have been doing anything as quite as kind of analytical as that, but I did find a video that brought me immense joy this week. obviously now being a father, you get served all kinds of stuff on all kinds of platforms, and the, video that I got kind of sent on Instagram, it's actually, on my kind of for you page, my, my recommended page, was a toddler in a silent disco. Have you ever seen a toddler in a silent disco? it's the purest amount of joy you could ever see. and when I say silent disco, I mean, you know, this is a two-year-old wearing his dad's kind of running headphones and being played some sort of like Disney classics. the kind of whirling of the arms, the pumping of the body, it's, it's honestly something kind of quite mesmerizing and

Mike Masnick 3:37

you sent this to me a couple minutes ago, and I will say, this kid is getting down. I mean, this kid is, is really feeling the music.

Ben Whitelaw 3:45

Yeah. It's flowing through every single sinew of his body, and he's completely kind of immersed in it, and obviously no one can hear, you know, what he's listening to, and he's completely kind of incommunicado with, with his parents. So yeah, I think we should take that energy into today's podcast,

Mike Masnick 4:01

Absolutely.

Ben Whitelaw 4:01

basically. and, uh, yeah, let's see how it goes. Hello, and welcome to Control 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 June the 4th, 2026, and this week we're talking about Meta marking its own homework on teen safety, what the Bricks and Minifigs story says about online speech, and for our Patreon subscribers, internet despotification and what to do about it. have a drink before you say that one. Uh, Control Alt Speech is the weekly therapy session of Mike Masnick, editor and founder of Techdirt, and me, Ben Whitelaw, founder and editor of Everything in Moderation. Y- you seem to like that one, Mike, weekly therapy. Yeah.

Mike Masnick 4:54

a good one

Ben Whitelaw 4:55

I think that hits the nail on the head

Mike Masnick 4:57

A- absolutely.

Ben Whitelaw 4:59

Yeah, I've, I think there's more, more to say on that. yeah, before we jump in, big love for our, uh, Patreon subscribers who have been both kind of keeping the lights on and, also you know, supporting us in the early stages of this new experiment we've been running. we've, been running our Patreon for five or six weeks now, and we have supporters who pay $10 a month to get extended episodes of our dulcet tones and occasionally some analysis as well. And our founders who, for this weekend only actually, will get the honor of being one of our earliest supporters, plus they get the chance to have a direct line to us and send us stories that we can cover on the podcast. After this weekend, that founder tier will go slightly up in price and will become something a bit different. So if you are interested in supporting the podcast, if you've been listening a long time, now's your chance to, to

Mike Masnick 5:50

chance.

Ben Whitelaw 5:51

ac- to claim that founder supporter title. I've heard it gets you into a lot of very exclusive clubs and venues, Mike, if you say that you're a founding Patreon supporter of Control Alt Speech. I haven't

Mike Masnick 6:02

need,

Ben Whitelaw 6:03

that myself

Mike Masnick 6:03

we need to have a special handshake, which we'll, test out at Trust Con this year.

Ben Whitelaw 6:08

Yeah. Yeah, exactly. I'll, I'll let you do that one first. Uh, and it goes without saying that if you're an organization or an institution who wants to reach our, very attractive and clever audience, um, by sponsoring an episode or three perhaps, you can go to ctrlaltspeech.com, find out more about our sponsorship offering, which, is one of the best out there, I'd say. or by reaching out to Mike and I via our, uh, podcast@ctrlaltspeech.com email address. How's things this week, Mike? What, what's, what's new in your world from kind of being quite nervous about basketball, it sounds like?

Mike Masnick 6:45

Well, the Knicks won last night, so I'm happy about that.

Ben Whitelaw 6:48

Yeah, what was it like before they won though? Probably

Mike Masnick 6:50

oh, it was, it was a little nerve-wracking 'cause they were, they were down by a lot and they, they came back and, and won. Uh, but this is not a basketball podcast, so we're not, we're not gonna, we're not gonna go there. But, uh, no, it's been a busy week. we'll talk about it later, but I, published a big piece this week, which has been making the rounds and getting plenty of attention, and I'm, I'm glad that that's out 'cause I put a lot of time into that. and so that was big, and then there's a whole bunch of other stories going on, some of which we'll be talking about. And so it has been-- I mean, it's always a busy week, but this one feels particularly crazy. So how about you?

Ben Whitelaw 7:24

Yeah, yeah, it's been good. I, been kind of running around, went to a couple of events here in London, which really interesting to speak to folks who are, you know, deeply in the weeds on this stuff. and I think, yeah, you're, you're gonna talk us through some pretty high-profile dramas, storylines that have been, taking place this week that have a speech element, which is always fun. I like it when those, our quite kind of quirky, nebulous world of online speech and internet regulation breaks out a little bit and reaches new audiences as, as we'll discuss. we should get straight into it 'cause we've got lots to discuss and, we're gonna start, Mike, by talking a bit about product defaults, and that is not a reason for our listeners to switch off.

Mike Masnick 8:05

audience gets it. I don't think you have to give that sort of, you know, caveat.

Ben Whitelaw 8:10

No. Maybe not. Maybe not. But I mean, I, I'm very fascinated by product defaults. I think they are, you know, everyone thinks about, uh, internet speech and safety as about content moderation, about, as about policy, and product defaults are, you know, obviously one of the kind of blanket ways, one of the, kind of most impactful ways that you can change the way that a platform works and the way people be- behave on, platforms. And I'm gonna talk a bit about Meta. Do you remember last year when, Meta announced some changes to how they were doing teen defaults? we talked a bit about it on

Mike Masnick 8:40

Yes, we did

Ben Whitelaw 8:42

and this was a fairly big story at the time. they essentially rolled out, plans to, have a new teen experience that meant that content was more suitable for, teen users, users between kind of 13 and 18. And they were, piloting this in the UK and, the US, Australia, and Canada. this was a time when there was loads going on about teen safety. you had the Australia social media ban had been announced but not kind of enacted. That was gonna happen in, December, a few months later. The Online Safety Act was being enforced, and there was a big focus on, harming of children. You also had, our good friend Jonathan Haidt doing many, many interviews in the media about his book, "Anxious Generation." And so there was this kind of cacophony of, noise and pressure around child safety and particularly kind of Instagram's role in that, which has been long documented. well, this week Meta announced a plan to roll out that teen default to a larger audience, a more global audience, but also, onto its other platforms, onto Messenger and to the main Facebook experience. So it's not something that's got loads of attention, but it's gonna impact a significant number of users' experience on those three platforms, which we know are, are very widely used. I wanted to ask, Mike, did you come across this teen default? Did your kids kind of come to you and say, "Hey, I've, my, my experience on these platforms has changed"? I've

Mike Masnick 10:02

No, I, I had not. I had, had not actually realized until a little bit afterwards that, this rollout had happened. I will note my, kids don't really use social media that much. They're not as, you know, beyond YouTube. YouTube is the experience that they, they enjoy. But, they're not big Instagram... They're certainly not Facebook users. Nobody is.

Ben Whitelaw 10:22

yeah.

Mike Masnick 10:24

But,

Ben Whitelaw 10:24

that ship has sailed.

Mike Masnick 10:25

yeah, yeah. But they're not real Instagram users, so we didn't hear anything about it, in terms of that, that aspect

Ben Whitelaw 10:31

Fine. I was really interested in the kind of this announcement and the justification for rolling out this new experience to so many users globally. And I'm gonna talk a bit about the stress test that basically Meta ran. says it did a study, but as we'll talk about, it's not a particularly, academic one or very robust. And then I wanna talk about what I think it, kinda says about the nature of Meta and how it thinks about teen safety at the moment. So just to give you the background, Meta ran, it calls a study, with a company called Alice. Alice is a, a well-known trust and safety vendor technology company. It used to be called Active Fence. My sense, I don't know for sure, but I think they have a long-standing relationship with Meta. They have been a, vendor of Meta's, trust and safety team for a while, is my guess. they've worked with a lot of big platforms. And anyway, Meta worked with them on this study, to validate the teen default experience and whether it actually did the job of reducing harmful content in those teen accounts' feeds, but also in search and in reels and all those other different interfaces that you have on, on Instagram, 'cause it's not just the feed. The way they did that was by creating 36 different accounts as 14-year-olds. Okay? So they created 12, with the kinda new teen default experience. they created 12 with Meta's limited content experience, which is something that it rolled out as well, which allowed parents to basically take a kinda stricter view of content that the teens saw on Instagram. And then it also created 12 accounts on what it calls a competing platform, which I'm assuming is TikTok. It's not, it's not mentioned, but my guess is that

Mike Masnick 12:08

I, yeah, I, I was scouring the details to try and find any hint of who it would be, but the list is very short of who you could consider it to be, and, TikTok is the obvious one. So it's almost certainly TikTok, yeah. and I would, I would actually argue if it's not TikTok, then it's misleading.

Ben Whitelaw 12:28

Yeah

Mike Masnick 12:28

because in terms of similar experiences, there's really only TikTok, and anyone else, the context would be very different

Ben Whitelaw 12:35

Agreed. Agreed. and so, so they created these accounts, and I think what they're trying to show is that this new teen default is both safer and more robust for teens than this unnamed competitor. So the fact that they don't name it is certainly kind of undercuts the reason for doing the research, I think. But nonetheless, they, they prime these accounts, by which they mean that they, try to get them to behave in different ways. So they, they follow certain accounts and they, get the accounts to, be either a kind of normal 14-year-old who likes sports or likes books or likes food, or they get the accounts to, they prime them to kind of seek out content that is less suitable for teenagers. And they, specify those different categories, suicide and self-harming, substance abuse, risky behaviors, sexually suggestive content. And so you have these kind of cohorts of different accounts that are essentially designed to act in different ways. And the study is then on the basis of that priming to see what content comes through on the feeds and in the search results. It's essentially kind of trying to audit the, algorithmic recommendations, of these new accounts, this new account status. And they view almost 17,000 pieces of content, Mike, and they categorize that content. They label it as either potentially mature or benign. And unsurprisingly, I don't think anybody will be that shocked to hear that the kind of key finding is that the prevalence of mature content of this new teen default kind of account is much less than the competitor. basically, the change has worked in terms of reducing the likelihood that a teen will come across, more adult content, and that's even more so if the account has been put into a limited content mode, i.e., a parent has kind of said you're gonna have a more restricted experience than the default. and the numbers, you know, are impressive, 68%, 96%. And that's the basis on which they basically rolled this new teen default out globally. So that, you know, you could say, Mike, th- this is, piece of work done fairly well, a company getting an external party to audit an experience before rolling it out. there's a, school of thought that says that this is Meta doing something right. Before I kind of give you my maybe more cynical take, what, what did, what is your sense on the numbers and the kind of methodology side of things?

Mike Masnick 15:00

I mean, I, I do think it's interesting, right? I mean, I think, you know, one of the things that, uh, I imagine most listeners to this podcast recognize is that is not that companies don't take child safety seriously at all, that There are certainly people within these companies, there may be questions about priorities and resource allocation and all that kind of stuff, but there are people within all these companies who do look seriously at how do we, limit the potential for any harm related to children or, or for any users. That is the entire purpose of a trust and safety role. And so difficulty is in, know, some of those other questions around resource allocation and, pri-prioritization and how do these things impact other aspects, and also in the fact that it is a very complex topic area where decisions are not clear-cut and, doing one thing helps in some ways and harms another, and as we always say, includes trade-offs, right? So each one of these decisions includes trade-offs. But it's, it is interesting to me to see how the different companies think about these things and how they approach them. And so I found this interesting to just as a thought process and as a, a mental experiment to, like, how would you actually measure these things. It's interesting to see the methodology that Alice came up with and how they implemented it and tried it out. I am not convinced that it is the most effective or that it is, particularly telling in terms of what impact it will actually have on safety versus what safety story they can tell to regulators and to the media, which I think is sort of, kind of the overriding part of it. but there are interesting things in here, and I'm not s- trying to dismiss the entire study entirely. it's interesting to see how they went about it and the different choices that they made, but they are choices, and you would get different results if you made different choices within that process.

Ben Whitelaw 16:51

Yeah, agreed. And, you know, there's, I guess, a world in which the report isn't published for us to read.

Mike Masnick 16:57

Yep.

Ben Whitelaw 16:57

maybe, the partner, in this case Alice, that they worked with isn't named, and it becomes even more of a, black box decision, w- which would not be a great thing. but yeah, I agree with you that there's those headline numbers of the, the kind of reduction in prevalence of, content in Instagram that might affect teen users is what, was kind of announced. It's what they're gonna go to the press and, and try to kind of, share as a story. It's what they're going to talk to regulators about. we did this piece of work in October last year. It's had this material impact, this third party says so. there's a couple of parts that I struggle with a little bit. One really interesting aspect, which I, I'm sure you will have thoughts on, is the movie ratings element of this study. So as well as kind of comparing, the prevalence of mature content on Instagram to this unnamed platform, it also does so to films. It, it, it kind of draws that distinction between content on social media and content in films, and it does that by scraping a lot of data and comments from IMDb. I won't go into the exact details, but it's an attempt essentially to s- say that social media content isn't really as bad or, or it's actually n- no worse than content that people are letting their children see of 13-plus films. this is something that goes back to the announcement back in October, where there was this feeling that talking about social media content in the way that we talk about films gives people a framing that they understand. Parents know that when they see a film that's 13-plus or 15 or 18 or where- whatever it is in the US, I forget, they're kind of clear on, the decisions that they're making as to whether they allow their child to watch this content. Maybe social media could do something similar and alleviate some of the pressures the platforms are, are facing. the Alice report goes back to that idea in a fairly unconvincing way, by scraping lots of data from IMDb and trying to, to kinda make that distinction again. What did you make of that? Did, how did you, how did that feel to you?

Mike Masnick 18:54

I mean, there are a couple things. So one, if you remember when, when they first announced this plan, they actually used the framing of the movie rating system and called it PG-13, which is what the US rating is for, recommended that you be at least 13 to see the content. And the, Motion Picture Association, the MPA, got very upset about this, and there was a lot of back and forth, and eventually there was a settlement of some kind that, Meta would not use that framing of PG-13. And in fact, in their release on this announcement, even though they, they just talk about 13 plus, not PG-13, there is this note at the bottom that, uh, there's a lot of differences between social media and movies, and we didn't work with the MPA when updating our content settings. And basically like, you know, we've made it clear, like we are not associated with it, which is kind of funny. But I, you know, I think as we pointed out, at the time when that first came out and they used this sort of PG-13 rating, it's like that is a very different context. and I don't think it applies, and I think, you know, sort of going back to it here, other than to try and associate to make that connection in the minds of parents or, or media or regulators that, okay, we let teenagers see some content that is a little bit more mature because we have these movie rating systems. You know, other than that, it just feels really irrelevant. Like it's, it's such a different context and it's such different types of content that, the only purpose in sort of going back to the movie example is just to like prime people to believe like, well, we recognize that there's different kinds of content that teenagers can be able to see, even if younger people we don't think is appropriate for them. But other than that, like it just feels different, and it just doesn't feel like a useful comparison beyond the, the simple priming aspect of it.

Ben Whitelaw 20:41

Yeah. Whether the Motion Picture Association keen for it or not, I don't think it works as you say, but the fact that they, they squashed it pretty quickly, I think made that, decision for them. So, so there is that element which feels odd, and I would, recommend listeners going and if they have time, digging into the methodology there. There's a lot of interesting kind of nuance. The other part was, what happens after, Alice came up with the data and, and basically figured out that there were some opportunities for Meta to improve. they highlighted essentially two areas that, Meta should address, and one of those was that there was a, a bunch of 18 plus accounts that these teen default accounts could access still. no one really understands how. Apparently it was due to a kind of model error, it was described as. And so, you know, if you're 14, you can see accounts that are 18 plus, that's obviously not good. And so Meta very quickly closed that loophole. There's also a kind of, they identified a policy gap around what's called car surfing. I never knew you could surf a car, Mike, but apparently it's, it's a big thing

Mike Masnick 21:45

thing. It is a thing. I, I was aware of it.

Ben Whitelaw 21:49

Are you a, you a regular car surfer?

Mike Masnick 21:51

I, I am not. and I would recommend to our listeners that you do not engage in couchsurfing. It is a very stupid and very dangerous practice.

Ben Whitelaw 21:58

Don't say you heard it, about it here. but, but, you know, h- having seen obviously a bunch of content being recommended to these 14-year-old accounts, Alice was able to say, "You should, you should create a policy that doesn't allow these users to see this content," and Meta quickly closed that loophole. Now, that's, again, a positive reaction from a platform in response to new information and data, and you could see it as such. However, if I was, for example, the oversight board, Mike, the, you know, Supreme Court-style, content moderation group of experts who frequently engage with Meta and who, provide recommendations on the platform, I would be pretty annoyed about the fact that Meta have made some changes in relation to their own data, but not to stuff that the oversight board have, brought to them. Similarly with other, third-party organizations like the out-of-court dispute settlement bodies, which are, part of the regulation in Europe. They're frequently saying, "Meta is not engaging with us as much as we'd like on, some of these key topics." and here we have, Meta making some very swift changes to data that they've commissioned for themselves. So I think there's a slight double standard here, Mike, when it comes to changes that Meta is making, in relation to data that they have commissioned or that, you know, these very close party vendors have brought to them, and less so with these bodies like the oversight board who are trying to make changes on the platform but struggling to, get the engagement they'd like.

Mike Masnick 23:26

Yeah, I mean, I, I think th-this is the interesting thing, right? So you have the oversight board in particular, which is staffed by a bunch of human rights experts and people who have thought deeply about these things and have sort of deep philosophical, careful takes on, you know, how do we balance different interests of different people. That is a very different situation than Meta going out, hiring a company and saying, basically, Do this, particular safety audit to effectively say that what we're doing is, is working." and this sort of gets at one of the things in looking at this overall report that gets, to a, deeper issue where you can say, "Okay, we reduced, access to this content by X percentage or whatever." There's no indication of did that stop any actual harm? th-because there's a question of seeing this content, accessing this content, and the actual impact on harm, and also, like, if you're blocking this content, are you also blocking other content, right? I mean, the trade-off situation comes up all the time where we're talking about, the example often is with suicide-related content. if somebody's looking for resources, are you giving them help, or are you just cutting them off entirely, which could make situations worse? We have none of that sort of thoughtful analysis. That's the kind of thing where, outside experts could look at these things and, explore the trade-offs. This report doesn't do that. It just comes up with a, "We reduced access to this amount of content," without then any further analysis in terms of what is the actual impact of that. And so in that sense, when you look at it that way, you say, this report Even though there may be some impact on safety, we don't know that from this report. We know it reduced access to this content. Does that actually increase safety? We don't know. and that would be a more interesting look, and that's something where an outside body that has other types of expertise, I think would be more interesting. And that's why this feels like even though there, there are some interesting things here, and it could lead to a safer platform, we don't know that from this data and this report, and that's why it feels more like a PR exercise than a real safety exercise.

Ben Whitelaw 25:28

Yeah. Agreed. And it's not even comparison to a pre-October 2025 benchmark, there's no, "We did this back in October before we launched teen defaults, and this is what it looks like now." It's comparing kind of apples with oranges in a way. It's, It's looking at the kind of prevalence of a completely different platform

Mike Masnick 25:49

With completely different standards and again, because I don't even name the platform, again, we're sort of assuming that it's TikTok. But, if you wanted to goose the numbers, like pick X as the competitor platform, right? Like you could pick a platform to make the Meta numbers look better if you wanted to. And I'm not saying that they necessarily did that, but there's a limited number of platforms, and you can make choices that would impact these numbers greatly

Ben Whitelaw 26:11

Yeah, agree. so yeah, so that, that was my big interesting story this week. I think, you know, product defaults not always interesting, not always sexy, but I think in this case, the data we have about it made it a little bit more interesting, I, think, than

Mike Masnick 26:24

I, do think it will be worth watching, right? If they're rolling this out more broadly, let's find out what the actual impact is, and there should be more data going forward, and hopefully it means that also outside researchers, academic researchers at universities and elsewhere can now do some comparative, research themselves on the data that is available to people outside. It's a little bit more difficult obviously if you're outside. But we do have a sort of before and after, and maybe we'll see some other evidence show up somewhere else as well.

Ben Whitelaw 26:52

Yeah, for sure. so we're, gonna, as with our new format, Mike, talk about, our second big story in the Patreon episode, which was, it's coming up. Before we do so, we wanted to run through a couple of eye-catching stories that we, both spotted that we found interesting. And, uh, talking of Meta, uh, you wanted you wanted to bring about, uh, a story that doesn't necessarily reflect that well on its use

Mike Masnick 27:14

Yeah. Uh, so this, this story made the rounds this week. you know, it was sort of noticed over the weekend that a couple of high-profile Instagram accounts were hacked, including the Obama White House account, and the chief master sergeant of the US Space Force, 'cause now we have a Space Force, Ben. and

Ben Whitelaw 27:33

About ti- about time, I'd say.

Mike Masnick 27:34

about time, yeah. and y- y- what came out was that effectively, Meta has been relying more and more on AI tools for a variety of things, and one of those is for customer support, and that's not unique. Lots of companies are experimenting with these things. But, you know, apparently they've given that, AI-powered customer service, tremendous power, and some people figured out, and it was apparently making the rounds in various hacker circles on Telegram, that you could trick the AI customer support at Meta in ways that would effectively allow you to take over an account. The full methods, reporters are being pretty good about hiding it. I've read a few different articles on it. The one I really liked is the KrebsOnSecurity, because I think he's always very thorough and, and thoughtful in, in his writeup on things. But effectively, what you do is you use a VPN to show up at an IP address that is in a similar location as that account would normally be, because apparently there is some geographical check. and you convince the AI through a f- a prompt injection that you need to change the email address that is associated with the account, and then you change it to one that you control, and then you can reset the password because you have the email address, and you can, take over the account that way. That is really bad. There's no two ways about it. and there's lots of talk about how with AI systems the big risk is prompt injection, which is, it's sort of the modern version of what used to be called social engineering, except instead of tricking a human being into doing what you want, you're tricking an AI, and there are tricks that, work. these are systems that are not determinative, and you can figure out ways to, effectively jailbreak their controls and get around them, and in this case, it feels like this one should be a f- fairly obvious one to make sure it's locked down, and Meta failed to do so. But it, you know, this certainly raises on the trust and safety angle, as people are rolling out more and more of these AI tools to do different things, figuring out how to really thoroughly lock them down and have guardrails against this stuff, you know, especially against prompt injection, within a customer service arena, becomes a bigger and bigger deal.

Ben Whitelaw 29:50

And it's not as if those accounts are easy to get back. You know, if in, non-AI, hacking situations if you just lose your account or, you know, there's a, an old-fashioned hacker who's, who's managed to, use social engineering to get access to your account, it's really tough actually to, get your account back. It's happened to me once, and we talked about it before, Meta doesn't have a very good kind of customer support service So if it's gonna happen at a scale that we've never seen before,

Mike Masnick 30:16

And in theory, right, I mean, that's, why the, the use of AI as customer service itself is interesting because, you know, the technology industry has never been particularly good at customer service. I used to joke about how Google's customer service was just like a giant white monolith. Like, there was no way to reach anybody there ever, if you had a problem. They've-- I think they've gotten much better about that over the years, but same thing, customer service has never been a focus of the, big internet companies certainly. And so there is a world in which you look at this and you say, "Oh, well, this is interesting," because they can effectively staff up their customer service so that customer service can actually be more responsive because the AI can take on the work that they didn't have human staffing to do before. But it has to work, and it has to be secure, and if you can do something that feels as obviously bad as this, it suggests that they haven't really thought through the guardrails of how the AI within the customer service realm should be working.

Ben Whitelaw 31:12

Yeah, I agree. a little bit worrying, but yeah. part of the course right now. Everyone can have their accounts taken over at any given time. keep an eye out, everyone. I was gonna talk a bit about an FT story, Mike, but w- I wanna, I wanna jump to, probably the, big, narrative on the internet this week, I'd say. I'll include the FT story in, in the show notes. It's a kind of interesting story that builds on, on what you've been talking about, about AI. But we wanna spend a few minutes about this Lego dispute. I'll be honest, when you brought this to me, earlier today and said, "This is what I wanna talk about," I was like, "What the hell is bricks and minifigs?" I'd completely not paid attention to this whatsoever, and your-- luckily, your very long article on Techdirt helped give me some background. But break it down for us and talk a bit about the kind of online speech aspect.

Mike Masnick 31:59

again, let's be clear. this is one of our shorter stories. To accurately explain all of the details here would take more than I think we've ever recorded. So much of the discussion is also happening in really long YouTube videos, and competing videos and different viewpoints. There is a huge mess here, most of which we are not going to talk about. So I'm going to skip over and, get past a whole bunch of things. There is an underlying dispute which went completely mega viral all across the internet. My kids are reach-- they're talking to me about it, which is like, to give you a sense of, like, how far this has spread, of a guy in Oregon who had-- his father actually had collected over a period of decades, a very large and fairly valuable collection of Star Wars Lego kits. and they decided to sell it off to help for the grandkids' college, and they went to a store called Bricks Minifigs, which is a franchise operation. There's a whole bunch of them where you can buy and sell different Lego pieces. I had not been aware of its existence prior to a couple weeks ago, but they have about 300 stores, around the US mainly.

Ben Whitelaw 33:07

I'm glad it's not just me

Mike Masnick 33:09

I'm not as big into the Lego world, I guess. But anyways, it's, it's a, a You know, there are series of stores and you can buy and sell your Legos, but this was a, a deal, it was a consignment deal, which is, you know, technically he still owns it, they're selling it on his behalf, and they had advertised it, and there was a whole big thing. There was a, big issue where the franchise owners who had signed this consignment deal wanted to get out of their store and wanted to either sell it or shut it down or whatever, and skipping over a whole bunch of, partially important details, the Bricks and Minifigs corporate showed up, took over the store with almost no notice. Literally showed up, kicked the people out of the store, demanded the keys, and said, "We're taking over." and caught on video, the owner says, "Well, I have this big consignment with all these Legos, and I owe this guy money. It's not ours. It's not our inventory." And they said, you know, like, "It's okay. We're-- We'll accept it, and we'll take on that liability," and then basically said, "No, these bricks are ours." There's a whole bunch more that has happened, and there were threats of lawsuits and this and that but basically, the, the company was trying to get away with taking the, the remaining Legos. Then enter another Ben, not you, but a a, a Ben named Ben Schneider, who has a YouTube account called Reckless Ben, who does these sort of crazy stunt things. in my mind, they're a little bit cringey, but he goes to extreme lengths to expose things, and I think his heart is in the right place. Uh, his brain may not be. And so he found out, yeah, he found out about this story, and he was just like, "I'm gonna get back these Legos," and then went to extraordinary lengths. I mean, th-this is where, you know, he I-I'm not gonna get into any of the details of this at all, but at one point, he registers a fake religion, sets up a lottery system, sets up multiple other companies, sets up a franchising system for his other companies, prints fake signs. he's based in LA, I think, or Southern California somewhere. He goes up to Oregon multiple times. He goes to Utah multiple times. He confronts people. He, sets up fake contracts. there are layers upon layers upon layers, all of which is too much to get into. So it went viral about a week ago because of his videos and it also turns out he gets arrested multiple times, he gets backed off by police, all these things. There are now state lawsuits in Utah mainly that have been filed. There've been a whole bunch of small claims lawsuits that have been filed. But let's get to the, the online speech component of this

Ben Whitelaw 35:44

Okay. I'm, I'm ready for it. I'm, I'm excited

Mike Masnick 35:47

which is... Well, first of all, the aspect that, something bad definitely happened here, and the guy who was the victim, the, the original owner of the Lego sets, had really no recourse, and that comes out very clearly. He sort of went to lawyers, and they basically said, "It would be more expensive for us to help you than y- we would get back from the company." and so the only remedy was this online speech aspect, that someone as crazy as Reckless Ben would go do reckless things, and draw attention to it. And that is fascinating because it's, it's a different thing that, you know, the way the world works today. The ability of a random, wacky, sort of cringey YouTuber to draw attention to something is fascinating, and to actually make change in the world, I think is, is really fascinating.

Ben Whitelaw 36:33

Yeah, we're not past that point, are we? Like, I thought that was something that was kind of prevalent 15 years ago maybe on the internet, but we-- And I thought maybe it wasn't, didn't exist anymore, but, but no. But Breakfast Ben has shown us it's

Mike Masnick 36:43

it is totally possible and, and as I was mentioning to you before, like now this has just become a, competing series of YouTube videos and my YouTube algorithm is now completely overtaken with videos about this, this account. Then the final point is that because the different lawsuits have been filed, there's a big one in Utah which was filed by the Bricks and Minifigs Corporate against Ben and co-conspirators. It's a, a RICO conspiracy case among other things. and they were able to get a temporary restraining order which is way too broad and clearly violates the First Amendment and it orders a bunch of things that are potentially legitimate. Ben has to stay away from some of the people involved. He can't trespass on their properties, so all of these kinds of things. But it also requires him to take down all of his videos and to shut down his Patreon account. And so the lawyers, which is a big time law firm, Dentons, for the Bricks and Minifigs Corporate, they sent these temporary restraining orders to, I believe, YouTube and to Patreon. We don't know... The videos are still up, so clearly YouTube didn't take them down. But Patreon, Patreon CEO Jack Conte put out a video. It's a 50-second video, and it's so good that I'm gonna just, we're gonna play it right now

CLIP: Jack Conte 37:57

Dear creator community, my name is Jack Conte, and I'm the CEO and co-founder of Patreon. On May 29th, Patreon received an official takedown request filed by Bricks and Minifigs for media and accounts related to Reckless Ben and an ongoing dispute regarding a local Lego consignment issue. This is the verified complaint with exhibits. This is the request for immediate content removal pursuant to temporary restraining order, and this is the motion for the temporary restraining order and preliminary injunction. So after an extensive review and investigation by Patreon's trust and safety team, we have in fact unfortunately determined that Bricks and Minifigs can stuff it. We're keeping Ben's page up, and if Bricks and Minifigs doesn't like that, they can sue us. Patreon out.

Mike Masnick 38:44

so if you heard that video, that is Jack Conte, CEO of Patreon, basically saying, "Yes, we received this," and as you heard, telling Pricks and Minifigs to stuff it

Ben Whitelaw 38:54

yeah. Go, go suck it.

Mike Masnick 38:56

Yeah. I mean, it is fascinating to, one, sort of join in in the, sort of viral video aspect of this, but, like, there is a real legal thing to this, which is he's right. if they want Patreon to be required to take down a temporary restraining order, he's not a party to the lawsuit. So the only way he gets to be a party to the lawsuit is if they sue, and he tells them, like, "Yeah, you, if you wanna sue us, go ahead and sue us." and so this is why things like Section 230 exist, to say, you know, the, you're protected. You can't be forced to remove content unless, you're a party to the lawsuit and then other things have to happen. The First Amendment protects the speech, and you can't do this sort of prior restraint of saying, like, "You are required to take this down." And so despite all the other crazy things involved, at the heart of this, this is an internet speech case.

Ben Whitelaw 39:42

Yeah. And it's good to see actually a CEO come out, and, I suppose essentially kind of back what is essentially a trust and safety decision, right? So to kind of publicly and vocally explain, maybe not explain the dec- the decision-making behind it, but at least, pin the kind of colors to the mast as to say, "We're not taking this down. This is, too broad, this is unnecessary, and we back ourselves to, that we're on the right side here." That doesn't often happen. You know, CEOs often kind of hide in the shadows when making such decisions. So, you know, fair play to the CEO of Patreon we, Mike, have come to the kind of end of the regular episode of "Controllable Speech." This episode is, will always be available to listeners on all the major podcast platforms. we are now going to segue neatly into the extended version of "Controllable Speech," the, the plus version, the, the extended part of it. And Mike and I are gonna go deep into his latest essay, which builds on a lot of his work that he's been doing over tech, over many years, and is a kind of, I would say like a kind of descendant essay of your "Platforms Not Protocols," paper that you wrote, which we can talk a bit more

Mike Masnick 40:45

and specifically about sort of the impact on centralized systems and democracy. So broadening it out beyond just like the impact on social media, but rather society at large, and why the architecture of platforms matter for society and democratic values. And so, I think we'll, we'll have a good deeper discussion on that

Ben Whitelaw 41:05

Yeah, and before we just, we go over to that episode, Mike, just explain how it came about, uh, think people might be interested in a bit of that

Mike Masnick 41:11

Yeah, th-this is published on liberalism.org, which is a relatively new, you know, online website that is focused on sort of liberal values is the, is the concept. And they had reached out to me a few months ago and said, "Would you write something on decentralization and democracy?" Was basically the framing that they gave me. And I said, "Sure, I'll have something for you next week." And that was three months ago. it turns out that writing this piece has-- it w-was really... Uh, I went deep to actually write it. And the more that I wrote, the more I realized, like, Oh, I have a lot to say here, and it's, it's a difficult one to get out in the right way." And so I wrote it, and I rewrote it, and I wrote it again, and I... At one point, it was up around 5,000 words. The final piece is about 2,500, so I cut out a lot. it was a, a, struggle to get this piece into the shape that it's in, but I'm very, very proud of the end result, which I think I've been hearing from a lot of people. I think it's a, it's a useful way of thinking about the internet and the infrastructure that we build and what that means for our society.

Ben Whitelaw 42:13

Okay, great. Let's get into the details.

Announcer 42:17

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