Will a new law change the internet forever?

0:00This BBC podcast is supported by ads outside the UK.

0:30they want to do. To see how ServiceNow puts AI to work for people, visit servicenow.com.

1:00The next one.

1:30built an ID facial recognition surveillance state, it's really hard to roll that back. Hello and welcome to The Interface, the show that decodes how tech is rewiring your week and your world. I'm Thomas Jermaine. I'm Karen Howe. And I'm Nikki Wolf. Today on The Interface, the proposed new law that could end the internet as we know it. The hidden human labor that's watching you. And are wired headphones actually better than Bluetooth.

2:04So we wanted to first start today with a bit of a somber update on the news that we were talking about last week, where we mentioned that the Wall Street Journal had reported that Anthropics AI tool Claude had been used in the military operation to bomb Iran. And there's since been more reporting that has come out, including from the Washington Post, that has mentioned that Claude was specifically used to analyze a lot of intelligence data and then identify roughly a thousand bomb targets. One of the places that ended up being bombed was a school. And in fact, it was bombed twice. So when it was first bombed, a bunch of first responders and parents then came to the school to try and rescue people.

2:53And then it was bombed again. So there There were significant civilian casualties, many of them under 12 years old. And there is some legitimate speculation now from observers about whether or not this school was identified as a bomb target by Claude. Because one of the things, as you might remember, if you watched last week's episode, is that large language models are highly inaccurate. And they're very ill-equipped to be integrated into highly sensitive contexts, certainly life and death situations. And Futurism publication actually tried to pose this question to the Pentagon, whether or not the school was in fact identified as a bomb target by the AI model.

3:40And they declined to share any further information. It's all just a grim story all around. It is very grim. And with that, I think we have a really great show for you today. We're going to talk about some hopefully more fun topics. And horrible stuff that's a little less horrible. It's not that bad, but it's bad. So, Nikki, why don't you take it away? All right. So the way into this story is a little complicated because of the way Congress works. But a set of laws has progressed through Congress to the next stage of voting. The main one is called the KIDS Act, K-I-D-S. Congress likes to use these cutesy little acronyms for all of this stuff.

4:24Basically, what these pieces of legislation do is mandate age gatekeeping for much of the Internet. So on anything judged to have adult content, we'll need age verification. Now, that's not you may be used to clicking. I'm over 18 when you go to a site like Reddit or porn sites or anything like that. What this will mean is that you will now have to upload a picture of your government issued identification card. And this is really split public opinion straight down the middle, because on the one hand, yes, unarguably, we want to protect children from seeing things that might be inappropriate for them.

5:08There's all kinds of things online that you could make an argument for online gatekeeping. I think online gambling is one that I think we need some kind of protections from. On the other hand, one, that gives the government a massive database of people's identification. And we live in an age where there's a lot of reasons, especially in the US, that you might not want the government to have that kind of information about you or private companies to have that information about you. There's huge privacy concerns and huge surveillance concerns. And also what this does effectively is change the internet in a really meaningful way from a place where fundamentally you visit places anonymously to a system of gated communities where your identification follows you around everywhere you go.

5:58There's some important nuance here, right? So there's this question of age verification. If you live in half of U.S. states or in the U.K., for example, there are these laws that have been passed that on porn sites in particular, you have to verify your age beyond a shadow of a doubt, like with either some kind of facial recognition scan that determines how old you are or like flashing your government ID, like you're saying. in the United States, it's more complicated than that. The Supreme Court ruled on this with laws about porn sites that that's OK. But for the rest of the Internet, it's like not clear that this is constitutional.

6:35So these laws like the Kids Act are not outright saying that you have to do age verification. They're just saying if you're not very, very sure that there aren't kids, you're going to get in enormous trouble. And then like it's the sideways in that then you might have to do it. There's a balance here. We're trying to protect kids, but then at what cost to like your comfort using the internet? I feel like we should talk a little bit about like play out what would it look like for people's day-to-day lives to access information with this law in place? Because this was something that I didn't fully understand that both of you were trying to explain to me.

7:13So the worst case scenario is a government can designate anything as unsafe. And then that would mandate identifying anyone who goes onto it. So for example, WikiLeaks, or for example, the Epstein files, or the New York Times or the BBC, right? Access to information about abortion provisions in southern states where abortion is under assault by the state governments could easily fall into this. It could be almost everything. It provides the infrastructure for a very restrictive and very surveillance state way of operating the internet right now you know you're constantly being monitored everyone knows that but what if there was this even more direct tie to like your real world identity how might that affect not just the things that you choose to access but the things that you choose to say on the internet the real concern here, right?

8:12This is all about, at this point, social media. Do we want kids using social media under what circumstances? So like, and I think politicians are thinking like social media is an app, but it's also just a vector for speech and expression and communication and information. And by putting up this barrier where it's like, you need to tie your real world identity to what you're doing, that might change what you're comfortable doing online. There's a lot of knock-on effect here that certainly this kind of legislation does not prepare for. And that's even before you get to the problem that there is no good technological way of doing this, right?

8:51Everywhere it's been tried has been absolute chaos. There was a massive data breach at Discord last September. Which is like a gaming communicate, like it's like a chat app for gamers. 70 ,000 people's identification was leaked. And this was like their government ID. Yeah. Oh, wow. I think looking to porn is a really interesting way to think about what's going to happen here, because this is a truism of technology that whatever the future of tech is, it happens in porn first, because that's where the biggest economic engine is in the world. That's how the internet is made. Right. Porn is always the canary in the coal mine.

9:34Exactly. Anything in terms of privacy and surveillance and technology, it always happens in porn first. Everyone can agree that we don't want children accessing pornography, right? Like if you talk to Pornhub and the company that owns Pornhub, which is called Ailo, I had interviewed them last year for a story on this subject. They said, we're on board. We don't want kids on our website. The problem here, like you're saying, Nikki, is these laws make it more difficult to access these mainstream porn companies that are doing more to moderate the content on their platform. So this attempt to make children more safe just pushes them to more dangerous places.

10:16It seems like we're rushing ahead with this like not particularly thoughtful solution to a problem that does have consequences that are very, very clear. This new legislation updates another piece of legislation which has been going through called COSA, the Kids Online Safety Act. This new legislation actually removes a clause from that called the duty of care clause, which would affect the big social media companies and was designed to regulate their addictive structural practices that we've talked about in a previous episode. This new legislation actually removes a safety clause for that. Do we know who, which of the tech companies has been doing the lobbying?

10:56Yeah, this is a really interesting question that has kind of split the tech industry into different parts. So Meta, the company that runs Facebook and Instagram and WhatsApp, they're all in on age verification. But what they want is they want the operating systems to handle it. Wait, what? They should have to check your ID. because that way it's like the responsibility to protect kids is like no longer in Meta's hands. What would it look like for the operating systems of your devices to be doing this? Like, is that supposed to be better? That's a great question. So in the UK right now, there's a law that says essentially if you're like a sufficiently large online platform, then you need to be doing age verification to keep kids off of like content that they shouldn't be accessing and this has along the way hit companies like spotify wikipedia like you know was freaking out about this because wikipedia was going to have to start checking people's id so it's like where does the verification happen does it happen when you get to the website that you're on which is how the proposals were rolled out initially.

12:08A lot of people think that's a huge problem because now the chilling effect on speech happens immediately. You get to the site, you're there, and like, oh my God, this particular site is going to get information about what I'm doing here. The alternative that has been proposed is called device-based verification, where what that would look like, there's a couple of different ways you could do this, where it's like one time your iPhone or your laptop or your Android phone or whatever it is, it verifies your age one way or another. And then as you careen around the internet, you download an app, whatever it is, the app or the website can ask your device like, hey, is this person over 18?

12:49And people like that a little bit better because like there's less potential exposure, like in terms of who's getting your data in terms of like, you know, the number of places where there could be a security breach. But everyone I've talked to about this says that that's not a good plan either. And again, there's still this concern about chilling effects on free speech when there is a really easy alternative to this, something we already have, which is parental controls. Parents can go set this up and they can step in and decide what they're comfortable with their kids doing online instead of the government doing it.

13:32It's kind of interesting that you mentioned parental controls because a few years ago I did a story about parents who, like helicopter parents in the digital age, who have become very obsessed with using parental controls to monitor their children. And I hear what you're saying, that it does feel like at least a better solution than ubiquitous, everyone gets the same rules. But there is something to be said about how parental controls is also not the perfect solution. Like I was talking with some LGBTQ teens who are saying that, you know, having parents who do not support the fact that they are gay, and then using parental controls to surveil and control their access to health information was in and of itself like a traumatic experience growing up.

14:20So like, it's a more fundamental question than actually just technology itself in a way of like, this has been a perennial problem in society of like, who do we trust to take care of our children? That is exactly it. In Kansas, they passed an age verification law about porn sites. And the question is, what counts as pornography? And if you read the text of this law, it like lays out every possible sex act you could imagine. And then in the middle, they also say that you have to do age verification for acts of homosexuality. Right. So we cover every kind of possible sex. And then specifically we're like and gay stuff, too.

15:04And Marsha Blackburn, the senator who is one of the main proponents of this law, COSA, the Kids Online Safety Act. She has openly said that one of the goals of the law is they will limit dangerous information about the transgender movement. And at exactly that same time, Kansas has passed a law saying that trans people must change their ideas to the sex assigned at birth. It is one of the most restrictive. Right, and then all the driver's licenses were invalidated, like overnight, with no warning, right? Yeah. What we're talking about here, if these laws go through, you know, unchanged, is a very different internet than the one that we all grew up on.

15:41Would you rather that these kinds of laws, these kinds of protections had been in place when you were growing up? Do you think you'd be a healthier, more adjusted person? Christ, no. I mean, look, we grew up in, certainly I grew up in the age of a basically completely free internet. It was an absolute free for all. The internet had a sense of freedom and fun and hope. Now, maybe that was naive, but it seems to me that the damage that's been done to the internet has been done by the corporatization of it. It's been done by companies like Meta making their platforms so neurologically addictive. Occasionally, when I was a teenager, catching a gif of like naked people doesn't come close to the kind of harm that we're talking about.

16:30I guess the question for you guys is we know that age verification has all these potential consequences. We know that parental controls have their own flaws. Would it be better to rush ahead with this stuff before we'd answer these questions because we're so concerned about protecting children? Or would it be better to do nothing? Because I'm really not sure. It's basically like, are we OK with the current status quo, I guess? Yeah. Maybe the question then is like, what are the harms that we're seeing currently that are not being addressed? And they're very nebulous, right? This is not a law proposed because of anything suddenly that has come up, any sudden new movement.

17:08But I think the problem with putting in a bad law and a bad structure of laws is that that infrastructure is then there. Once you've built an ID, facial recognition surveillance state, it's really hard to roll that back. Well, if this story didn't get you really concerned about your privacy, then my story definitely will.

17:38professionals use LinkedIn ads. LinkedIn has grown to a network of over 1 billion professionals, including 130 million decision makers, and that's where it stands apart from other ad buys. You can target your buyers by job title, industry, company, role, seniority, skills, company revenue, so you can stop wasting budget on the wrong audience. It's why LinkedIn ads generates the highest B2B return on ad spend of major ad networks. Spend$250 on your first campaign on LinkedIn ads and get$250 credit for the next one. Just go to linkedin.com slash broadcast. That's linkedin.com slash broadcast. Terms and conditions apply.

18:14It's 2009 and we're in the German mountains. A man straps himself into a car on the world's most dangerous racetrack. He whispers to himself, it's time to put my balls on the dashboard. As he starts the engine. In 15 minutes, he's in an ambulance unconscious. In 15 years, he's a billionaire. This is Total Wolf, Formula One's most powerful team boss and the breakout star of Drive to Survive. This week on Good, Bad, Billionaire, how Toto Wolff made his billions. Listen wherever you get your BBC podcasts. There was a recent investigation that was done by two Swedish newspapers, SVD and GP, as well as a freelancer in Kenya, Nepanoi Lepapa.

18:57And this is the craziest investigation because they basically discover that if you are a meta Ray-Bans user, people are watching the recordings that you are making and there is no way for you to opt out. These are smart glasses, right? We're talking about glasses with a little camera, a little computer. You can record stuff you see. One of the immediate problems with them came up with people recording people in public without their consent. But also people were recording themselves either intentionally or unintentionally doing very intimate acts at home, like sitting on the toilet, having sex.

19:39In one case, the investigation found that, you know, a man put his Meta Ray-Bans on the bedside table and then left the room, but it was still recording, unclear whether he realized that or not. Then what appeared to be his wife walked into the room, definitely unaware that the Ray-Bans are recording and started getting undressed. And so all of this footage is gathered up by Meta. And it turns out that some of it ends up getting sent to a third party contractor in Kenya called Sama. And the investigation interviewed over 30 Sama contractors, as well as former Meta employees in the US, who then revealed that, you know, there is this huge supply chain of data, intimate videos being sent to these people for review and your sex video can end up in an office in Kenya.

20:38People have been outraged by this and this investigation triggered a probe by a UK watchdog and now a class action lawsuit in the US. So this is kind of like a crazy situation because I mean, well, I mean, obviously, it's like, really, this is like, this is partly a crazy situation because of the way that Meta advertises its glasses. So it has this page dedicated to privacy of their Meta Ray-Bans products. And at the top of this page, it says in big bold letters, designed for privacy, controlled by you. Oh, good. Oh, well, what are we even talking about then? That's fine. Neither of those things.

21:27Every word in that sentence is a lie. Yeah. So this is what Meta's defense has been in response to this investigation is, oh, but in our AI terms of service, which is a totally different page, there's this like tiny little print that says, we may sometimes conduct review of the things that you put into meta AI. Sometimes that will be automated review. And sometimes that will be done by humans. And then in other little tiny print, it's like, maybe don't share sensitive data that you don't want shared to us. Well, I think, you know, the good news here is that almost everyone reads all the privacy policies.

22:10So you'd probably know this. Right. Well, Karen, this is one of my favorite parts of your book. This is like something that you've investigated specifically. And there's a whole other interesting thing here. Can you remind us, like, why are there people who don't work at these companies in other countries reviewing all of this content in the first place? Yes, I've been so obsessed with this topic. I've I've been reporting on this for almost eight years. And basically the entire internet is built on top of hidden human labor. The reason why people in the global north get a clean social media experience is because, once again, there are people in the back end that are reviewing all of this grotesque content to take it offline.

22:57And the workers themselves consider this to be a career. It's not like, you know, like oftentimes when people talk about this work, they really minimize it as it's menial labor. It's, you know, it's drudgery. But it's like it's really hard. And these workers, you know, they take pride in the fact that they're keeping the Internet safe for people. There's also people that review and do data annotation. So the reviewing is sort of similar to what we're seeing with this Meta Ray-Bans. Data annotation is this function where in order to even train AI systems, there are people that are literally teaching the AI model every possible thing that it needs to know.

23:37So the fact that chat, GBT can chat is because there's literally tens or hundreds of thousands of people around the world that are typing into OpenAI's large language model and showing it, this is the correct answer. This is a not so great answer. And then there's my all-time favorite category of human labor is artificial artificial intelligence. Right. Which we'll definitely get to at some point in a later episode. But suffice to say, you think that, you know, self-driving cars are always just driving autonomously. But oh no, there's actually a team in the Philippines that sometimes takes over and is remotely driving.

24:15Literally everything that you possibly use probably has hidden human labor behind it. Like there was a story a few years back about how Amazon Alexa was also sending audio clips of people's intimate conversations to third party contractors as well. There was another story about how Roomba, after they attached cameras to their vacuum cleaners, they would also film people on the toilet as well and film people doing other types of intimate acts. And then they were sending it to a third party contractor called Scale AI. And then the workers were literally taking the Roomba photos that people did not even know were being recorded of them and posting them on Facebook in these Facebook groups to talk with each other about like, oh, like, how am I supposed to address like labeling this thing or how am I supposed to review this thing?

25:07So then people's potty photos were literally going on to Facebook and then like traveling to then public spaces because then they would leak from the Facebook group. I think it's fair to say that if you have a device in your house or you're using an app or you've got a thing and it's like got a system that continually gets better over time, right? It's like, oh, these smart glasses, they're getting smarter. That means that whatever data it's collecting, there is a guy who is watching. So, you know, you're strapping this camera on your face. Other people don't even know it's recording and you think it's your footage, but is it?

25:47So we reached out to Sama for comment and they got back to us saying that Sama is compliant with international regulations, including GDPR and CCPA, these are privacy laws in Europe and in California, and that they operate under rigorously audited policies and procedures designed to protect all customer information. Well, I feel better. I don't know about everyone else. Yeah, I mean, thing is, all these companies comply with GDPR, right? And yet, you know, things still seem to happen. But, you know, I'm glad they're not breaking the law. Sama was also the same third-party contractor for OpenAI for a hot second.

26:28OpenAI, they were trying to develop a content moderation filter that would basically protect future users of their GPT models from being exposed to some of the toxic content that the models are actually trained on. And so they end up hiring Sama as their third-party contractor to help them develop this content moderation filter. There was this brilliant investigation that was done in Time Magazine by Billy Perigo that exposed the fact that these workers were day in and day out be exposed to the worst possible content on the internet, as well as AI generated content where OpenAI was literally prompting its own AI models to imagine even more grotesque scenarios to cover the basis of like all of the bad content that could exist in the world.

27:21world. Yikes. And these workers were just getting like totally traumatized. And a lot of the times the, you know, the companies that are paying people to do this, like they're not providing them with any like, you know, preparation. There isn't like a psychologist you could talk to if you see something that gives you PTSD, which is a real thing that happens all the time. Like it's like just the human gears in this giant machine that we never see, you know, from our perch thousands of miles away. Yeah, I was just talking with this worker actually last week, not in Kenya, but somewhere else, who mentioned to me, he strongly believes there's no such thing as consenting to this work.

28:00This is the defense of contractors like Sama, where they're like, well, we asked the workers whether or not they want to do content moderation. And this worker was like, you literally cannot possibly conceive how bad it is going to be before you have done it. So you consent, but it's not informed consent. And so one of the amazing things about what's happening right now is one of the reasons why we're seeing so many more stories like this is because the workers are not okay anymore with being hidden. Like they are actively organizing now to get their story out, to make international headlines, to say like, hey, we are literally here doing some of the most essential work to make the internet function.