Your Bluesky Posts Are Probably Training AI

0:00This episode is brought to you by the Electronic Frontier Foundation. The election is over. The United States and the world might be headed into uncharted territory. That means defending digital privacy is more important than ever. Consider that Google search data can tell the police if you looked up the address of an immigration attorney, or the fact that online chat logs can reveal that you talked about an abortion. What was often benign data before could quickly become potentially criminal evidence now. For more than 30 years, the activists, technologists, and attorneys at the Electronic Frontier Foundation have worked to protect your privacy and your ability to control your personal data.

0:41Data that can reveal where you go, who you talk to, who you love, and what information you seek online. EFF defends user privacy, free expression, and innovation, regardless of the obstacles. In these uncertain times, their work is more crucial than ever. Learn more at eff.org slash 404media. Support the Electronic Frontier Foundation. Because in a world where your data can be used against you, defending digital freedom is defending yourself. That's eff.org slash 404media.

1:18Hello, and welcome to the 404 Media Podcast, where we bring you unparalleled access to hidden worlds, both online and IRL. 404 Media is a journalist founded company and needs your support to subscribe go to 404media.co as well as bonus content every single week subscribers also get access to additional episodes where we respond to their best comments gain access to that content at 404media.co i'm your host joseph and with me are 404media co-founders sam cole hey amazing timing on that a manual member right hello hello and jason clever will also be with us but he's going to call in remotely for the second segment that was like a dub hall that was great it was right in front of my apartment literally as you introduced me and now it's gone so that was nice all right let's keep going um these first couple of stories that sam wrote all about blue sky and scraping and data and privacy.

2:19The one you just published, the headline is your blue sky posts are probably in a bunch of data sets now. Let's go back in time though. Where does this start? It starts with a database of 1 million blue sky posts, right? Yeah. Yeah. So last week, a machine learning librarian at Hugging Face, which is this platform that's for open source, primarily machine learning data sets and research data and things like that, where people post these things, posted on Blue Sky that he was releasing a data set made up of a million Blue Sky posts. And they included everything that was basically attached to the posts, So like the person who posted's ID, like their handle or whatever, whatever they call it, a blue sky, I don't even know.

3:19Their timestamp, any kind of like images that were included would say whether they were imagery and all of that. And of course, the content of the post and people freaked out. People really lost their shit. Like I'm not exaggerating. People were like, this man should be thrown in jail. And worse. People were comparing it to like rape. Yeah, it was really crazy. Yeah. I think just to add one other bit of context, you say they're from Hugging Face. What did they say that this database of one million blue sky posts they made, what did they say that was for exactly? They said it was for machine learning research, which could mean a lot of things, right?

4:05Like, datasets from social media are compiled all the time. It's not usually something that people compile in a non-anonymized way. So that definitely made this different. And yeah, it was just kind of put on a huggy face just to be like, hey, here's this dataset that I made to play around with different, you know, like you could make an LLM with it in theory. You could do like anthropological research on the users of Blue Sky if you wanted to. Like you can make a bot. I mean, there's just like lots of ways that you could use that kind of data that people do, you know, use social media data for all the time.

4:45But I think partially because it was not anonymous and you could track people's ideas back to their posts. and because people were leaving Twitter in part because they were protesting the scraping of their data, their user data to build LLMs. Right, well, Elon Musk or rather X had come out and said, we reserve the right to scrape and to use whatever you post onto X and we're going to use that for our own machine learning. They have their little chatbot called Grok, right? which is supposed to be like a woke chat GPT or something stupid. But people don't want to contribute to that. So they go to Blue Sky and then somebody else is scraping all of their posts as well.

5:32I mean, I guess we should say this up top just so it's obvious. Like we know public data is going to be scraped. Lots of Blue Sky users know that as well. why do you think people reacted to this so aggressively because we see the data set you cover it that brings more attention to it and then people um voice their opinions against this person who uh who made the data set why do you think the reaction was like that i mean i think in general i think people when people find out and we've seen this happen with other platforms too. When people find out that their data is being used in ways that they didn't realize or didn't consent to or didn't look into or didn't occur to them that it could be, they get pissed off.

6:20That's the case across pretty much any platform. So it's not some kind of special phenomenon that's happening with Blue Sky that people are mad that their stuff is being used without their consent. I think it is a little bit special to Blue Sky in that people, like I said, people are very anti this sort of scraping in general. They're by and large, at least in the worlds that I'm following and I'm still somewhat new to Blue Sky, people are very anti-generative AI, anti-AI generally, anti-machine learning. I mean, it's just, it kind of reaches, and I wrote about this a little bit in my Behind the Blog last week, but like it kind of reaches this point where people are just sort of dogpiling it's like they see something that they can jump into and say hey fuck this and they just you know type hey fuck this fuck you um and then hit reply and then move on and that accumulated into like i don't know it was like 800 different replies people saying um that they did not like this and then the guy eventually i mean not event not even eventually it was like within like a couple hours took it down um and said, you know, hey, I hadn't considered the ramifications of this, whether those ramifications were getting into trouble or actually, you know, use of people's data without their consent.

7:48He didn't really elaborate on, but he took it down, which I think people were pleased to see. And he was like, okay, I get it. I'm standing down, you know. It's gone from Huggy Face, it's gone from his blue sky page all that so yeah well he took his one down but in the same way there was a response with the uh people who were angry or upset on blue sky there was a sort of other set of reactions which is now other people have made even more larger data sets i mean what are some of the ones that have been made in response and what's sort of the the trolling going on here because it's not just like oh hey here's research it's like some of these people are making data sets specifically to piss off blue sky users yeah yeah so this reaction to this one guy's post and data set went so viral on blue sky that it broke out of those usual like containment zones where people are like anti-generative AI and bled into Twitter basically is kind of how I saw it happening because a lot of people are still on both.

9:03It's not like an either or for most people. If you're on Twitter, you're probably also on Blue Sky, vice versa, or you're likely to be. So yeah, people saw this going viral on Blue Sky and on Twitter, they started posting about it there too, which turned into more people making more data sets and even bigger data sets. There were the 2 million one. The 2 million one was super popular, by the way, and had a lot of downloads. There's an 8 million one. There's almost 300 million one, I think is how much it was, right? Yeah, 298 or something. It's like a crazy amount of posts. but like this is still probably like a fraction of like the post total on on blue sky and we know this because the fire hose that uh blue sky is set up to kind of like make public contains all this data and it is public um so it's very easy to just kind of put it all in a data set as a lot of people did and yeah like you said like they're trolling they're like i think one of the um the the 300 million one was like, um, had this like weird, like rant in the description that was like addressing people saying that they didn't want to be scraped and kind of saying, you know, if you, if you don't want to be scraped, basically don't post to social media period, log off or start a blog, which I thought was hilarious because we have started a blog and we still deal with scraping every day.

10:41And it's like, it's, that's not really, it's not a solution. And it's also, um, wrong and dumb but like it's well it's almost like the extreme yeah it's almost like the complete opposite reaction on one side you may have people who and i don't think anybody really believes this but it's sort of a straw man in that on one side you may have people who are just like what you can scrape on the public internet that's outrageous that sort of thing and on the other side there's these people saying they're almost being defeatist about it or even more than that they're actually enabling it and performing that scraping themselves and you seem to have both sides of that spectrum here when it come and you touched on this because blue sky is open there is a fire hose of data that people can access you can't do that really on twitter anymore you have to pay for api access right and an extortionate fee to the point where researchers don't even do it anymore i used to scrape twitter for various purposes and it was pretty easy to spin up a python script to be able to do that because the API was available.

11:45So that openness about Blue Sky, you touch on it in the piece and you call it like a double-edged sword. What do you mean by that? Yeah, I mean, the way, what makes Blue Sky different from Twitter or Threads or Facebook or any of these others is that it's decentralized and it's built on this protocol that's open and you can port your content around. You own your following, you own what happens on there in a way that you might not on Twitter or threads or some of these others. So that's what's appealing about Blue Sky to people in a lot of ways. If you care about that sort of thing, that's the reason why a lot of people are on it.

12:31And that's also the reason why it's vulnerable to this kind of dataset collection. And, yeah, I mean, it's definitely, there definitely are, like, two very passionate sides to this where, you know, like you said, it's like some people are just like, never, ever, fuck no. And then other people are like, well, someone's going to do it eventually and it might as well be me, I think is, like, the attitude that people have always had, especially with, like, not just tech, but, like, it's really prevalent in, like, machine learning and AI in those communities where, like, people are like, well, eventually, like, we sell this with deepfakes, we see this with all kinds of harms that have to do with general AI.

13:14It's like, someone's going to do it, someone's going to have this idea eventually, so I might as well do it. Do you see that mentality across different things? You mentioned deepfakes then, and, like, you know this, and Emmanuel knows this as well better than me, but, like, was that part of the thinking? Not necessarily the person who first coined the term deepfakes and started doing that, but was the idea that, well, somebody's going to do this it might as well be me and i almost don't quite follow the logic and but was that part of the sentiment it's not super logical it's mostly kind of an excuse in my opinion to do the thing um yeah i mean that's that's almost verbatim what a lot of people who have developed deep deep fake type tools um have said to us in the past and keep saying um it's just this kind of...

14:06I think people are in our free-for-all grab as much as you can while the legal gray area is still gray. Whether or not these data sets are even legal under GDPR and copyright and things like that, it's all still being decided because these are problems that we hadn't had before. Not at this scale. These are problems that have always existed. Not with AI. Not with AI. Not when we're talking about machine learning. So, yeah, it's something that's still being like litigated literally right now. So I think it's just kind of a mess. Like it's people just don't know what ground we're standing on ever when it comes to whether this stuff is like not just putting aside like moral, ethical, safe.

14:55Like it's really not very safe to scrape like non-anonymous data into a data set like this. There's a lot of like personal information out there that people are scraping in these datasets. Um, but even like legal, I think is like what everyone's kind of waiting on. And we see this in all kinds of industries. It's just something I touched on in this story, but yeah, it's just right now it's like, get it while you can kind of attitude in the industry. Yeah. And again, these are independent individuals for lack of a better way of putting it. And it's not blue sky itself taking user data and training AI or doing machine learning or anything like that just to pull it into this conversation what is blue sky's stance on scraping and training ai with user data itself what's what's their stance so blue sky has said in the past that they are not gonna you know train a train ai on user data um which again and sets them apart from like Twitter threads or any of these others that are actively using user data to train their own LLMs and things like that.

16:03But they said they won't, but there's really nothing. And they've said this also publicly, like not verbatim, but like in a nutshell, it's like there's nothing stopping anyone else from doing that because of the nature of the thing. And I think they're like, they've, I think they're going to end up being forced to. And also like they've said publicly that they are thinking about how to address this because people are so put off by it. They're thinking about how to work better consent tools, basically, into the platform so that this sort of thing doesn't happen if you don't want it to. There's no...

16:46You can't have a private account on Blue Sky, which is a big problem for a lot of people. and that also opens you up to scraping if you don't have a private account. Yeah, they're working on it is kind of their response. But I don't really know what there is to do. Yeah, I don't know what you would do because as you say, it's open and the only way you could stop somebody scraping is by blocking the specific account which is viewing your profile. but I mean you don't know what account that's going to be presumably it's not going to be an account that says hey I'm a big scraper and I'm getting all of your posts and also uh presumably I think you could probably actually scrape it and without an account potentially as well you know you can view summer blue sky while you're not logged in um so there isn't really anything users can do is it more people just have to wait to see what the company does if anything i guess um i'm not like i'm again it's like i'm not really sure if it's on the platform or users to do this it's like it's again it's like it's this problem of like the people doing the bad need to stop doing the bad like it's like they cut it out they just like there's it's hard to like build you can only build so many walls against this kind of thing.

18:12It's just something that like we agree generally. I think that taking people's content without their consent is usually bad. Not nice at least. But it's not illegal. So like, is it ethical? Is it moral to do that? You know, it's like, that's kind of where the argument is. So like this guy should have known better who did it originally. The people who are doing it again probably should like are doing it for the trolling and to get people more pissed off, which is very classic Internet behavior. So I don't know how much you can tell them to cut it out. It's just like we need to like socially agree that this is shitty if that's the way this is going to go, you know.

19:06Or we agree that it's fine and whatever and take the data and go nuts. I don't know. I'm sure there's some kind of middle ground that we can figure out. And this is something that researchers have been doing for a long time, too, is using data sets and anonymizing them and giving people the chance to opt out, which is part of the law in the EU and UK. people should get a chance to say I don't want this I don't want to be part of this data set and I don't like I think it should be anonymized at the very least like at the bare minimum people's like username should be attached to it because again sensitive information and also deleted posts are in these data sets you know it's just you don't know what you're getting in there it's like the 300 million one is like there's like a thing on the description that's like is it not safe for work yeah for sure there's probably like foreign in there there's definitely a hole in there like you know whatever you've ever been on blue sky before like that's literally half the website yeah and like i don't think it should be on the users to stop posting hole that's kind of the ethos that we've always followed here we've always followed uh yeah we don't need to get into that about our potential earlier names beyond 404 media don't say it maybe we'll save that for subscribers.

20:28Maybe we've already mentioned it. All right. We will leave that there. When we come back, we'll hear from Jason about Redbox and I guess the Redbox removal team. We'll be right back after this.

20:54There are so many things to worry about when you're starting a business. One thing I worried about, how to get 404 Media set up as a legal entity. I shouldn't have worried. I found LegalZoom, which walked me through every step of the process so I could cross another thing off my list. LegalZoom helps business owners like you take the first step and every step after from reliable business formation to experience guidance in legal and tax. Setting up your business properly and remaining compliant are things you want to get right from the get-go. And LegalZoom saves you from wasting hours making sense of the legal stuff.

21:31Launch, run, and protect your business to make it official today at LegalZoom.com and use promo code 404media to get 10 % off any LegalZoom business formation product, excluding subscriptions and renewals. Expires 12-31-24. Get everything you need from setup to success at LegalZoom.com and use promo code 404media. LegalZoom.com and use promo code 404media. LegalZoom provides access to independent attorneys and self-service tools. LegalZoom is not a law firm and does not provide legal advice, except where authorized through its subsidiary law firm, LZ Legal Services, LLC. The holiday season is here, which means that if you have a business, you're probably pretty busy.

22:17I've handled an influx of new merch orders using Shopify. Simplify and empower your business and online store using Shopify's all-in-one solution to host your shop, make it look the way you want, manage inventory, and fulfill orders. Shopify's simple but powerful backend gives you everything you need to start and manage your store so you can spend less time researching and more time selling. Upgrade your business and get the same checkout we use with Shopify. Sign up for your$1 per month trial period at shopify.com slash media, all lowercase. Go to shopify.com slash media to upgrade your selling today.

22:58Shopify.com slash media. Lumen is the world's first handheld metabolic coach. It's a device that measures your metabolism through your breath. And on the app, it lets you know if you're burning fat or carbs and gives you tailored guidance to improve your nutrition, workouts, sleep, and even stress management. All you have to do is breathe into your Lumen first thing in the morning and you'll know what's going on with your metabolism and whether you're burning mostly fat or mostly carbs. Then Lumen gives you a personalized nutrition plan for that day based on your measurements. I've been working out more lately and I've been trying to get better sleep.

23:37I've realized that my metabolism is my body's engine, which helps me perform well when I'm exercising and helps me feel alert and focused throughout the rest of the day. Lumen's recommendations have helped me learn more about my body and how it turns food into fuel. It helps me know when to eat and helps me know what to eat so I can make sure I'm ready for a workout and ready to face the day. So if you want to stay on track with your health this holiday season, go to lumen.me slash 404media to get 15 % off your Lumen. That's L-U-M-E-N dot me slash 404media for 15 % off your purchase. Lumen makes a great gift too.

24:18Thank you, Lumen, for sponsoring this episode. Let's face it. After a night with drinks, I don't bounce back the next day like I used to. I have to make a choice. I can either have a great night or a great next day. That is, until I found pre-alcohol. Z-Biotics pre-alcohol probiotic drink is the world's first genetically engineered probiotic. It was invented by PhD scientists to tackle rough mornings after drinking. Here's how it works. When you drink, alcohol gets converted into a toxic byproduct in the gut. It's this byproduct, not dehydration, that's to blame for your rough next day. Pre-alcohol produces an enzyme to break this byproduct down.

24:59Just remember to make pre-alcohol your first drink of the night, drink responsibly, and you'll feel your best tomorrow. I've been trying pre-alcohol before drinks, and I definitely noticed a difference the next day. Last weekend, I took pre-alcohol, hung out with my friends, then woke up, surfed, walked my dog, and reorganized my living room. I had an actually productive day. With the holiday season upon us, I know I'm going to be consuming a bit more alcohol than usual. With pre-alcohol, I know I can stay on track and not let the season throw me off course. Go to zbiotics.com slash 404media to learn more and get 15 % off your first order when you use 404media at checkout.

25:38Zbiotics is backed with 100 % money back guarantee. So if you're unsatisfied for any reason, they'll refund your money. No questions asked. Remember to head to zbiotics.com slash 404media and use the code 404media at checkout for 15 % off. Hackers and cybercriminals have always held this kind of special fascination. Obviously, I can't tell you too much about what I do. It's a game. Who's the best hacker? And I was like, well, this is child's play. I'm Dina Temple-Raston, and on the Click Here podcast, you'll meet them and the people trying to stop them. We're not afraid of the attack. We're afraid of the creativity and the intelligence of the human being behind it.

26:20Click here. Stories about the people making and breaking our digital world. AI machines. Satellite. Engine ignition. Click here. And lift up. Click here. Every Tuesday and Friday, wherever you get your podcasts.

26:44All right. And we are back with Jason calling in remotely. um jason can you hear me i've been here the whole time just listening i appreciate it like this voice from from nowhere um you've done a couple of stories on red well a few at this point on red box this latest one is called the red box removal team where you went on a trip with some people liberating these boxes step back a little bit and you know because i wasn't familiar with this Can you just explain, for those who don't know, what is Redbox and what happened to the company? Redbox are these DVD rental kiosks that popped up all over the outsides of grocery stores, Walgreens, CVS, Dollar General.

27:36And it's sort of, I feel like it replaced Blockbuster kind of as a physical location that you could go and rent a DVD from. And I think for a while, they also did video games. The big thing about Redbox was that there was essentially no overhead. So it's like Netflix came and essentially killed Blockbuster, but Netflix was a mail service for a while. And so it would take a day or two to get your Netflix rental in the mail. whereas with Redbox, it was at a convenience store around the corner from your house. So if you were deciding last minute to rent a movie, you could just go there and rent it.

28:22I believe for a while, it was just a dollar a day, which was a big selling point. And I remember this being quite popular. My family used them all the time. and in a kind of amazing turn of events about maybe like two or three years ago, I don't know the exact timeline, Redbox was sold to a company called Chicken for the Soup Entertainment, which was the entertainment conglomerate arm of the Chicken Soup for the Soul self-help books. Uh-huh. Are you aware of Chicken Soup for the Soul, Joseph? I was a big fan. Or is it an American thing? I have no idea what you're talking about. But I'm glad Sam does.

29:15I mean, Sam, what do you mean you're a big fan? We're a big fan. Dude, Chicken Soup for the Soul? I was a very wholesome teenager, so I had Chicken Soup for the Teenage Soul. There's like a Chicken Soup for every soul. Chicken Soup for the Cat Lover Soul. It's like, yeah, I was going to go to the dog lover soul. There's surely other types of souls that there's, there's like hundreds of these books, Joseph. And they were like all over borders and Barnes and Noble and perhaps sometimes grocery stores themselves. And it was one of them. Rather than go to therapy, what we did in the United States was you bought one of these books for your like troubled teen.

29:54I never had one. Emmanuel, are you aware of this? I have definitely heard the name, but I don't really know what it is other than like a series of books. I was going to say it's like, what is like the, not Miss America, like, it's like a series of American Girl something. Yeah, I was going to compare it to that. But the thing that I would compare it to is For Dummies, the For Dummies series. It's like mental health for dummies, tailored directly to you. Because they all have the same cover. They all... And there's like 500 different versions of them. So anyways, this company, Chicken Soup for the Soul Entertainment, purchased Redbox because they briefly got into trying to have a streaming service and trying to do movies.

30:50And shocker, this did not work. And so Chicken Soup for the Soul Entertainment went bankrupt earlier this year. And they didn't go Chapter 11 bankrupt. They went Chapter 7 bankrupt, which is where... Is that the bad one? It's the bad one. It's where you go fully, you have no money. And you basically go to court and you say like, Hey, we're fucked. We're out. Goodbye. and what happened was there's something like 20 ,000 of these kiosks all over the country and you know they were servicing these for a while stocking them with dvds fixing them when they were broken and they were paying rent to these convenience stores that had contracts with them and they filed for bankruptcy and they were like we're out goodbye like we're not paying you any money, we're just done.

31:41And so these like six or 700 pound devices, like these big red steel, they seem like they're made of steel to me. Tiosks have just been totally abandoned. But the really interesting thing is that they still work. A lot of them still work. And so what people have been doing is they have been going and quote unquote, renting DVDs and just keeping them. And then there was this whole community of people who decided, wait, why rent all the DVDs when I can just bring my pickup truck and an angle grinder? And I can like, because they're all like bolted to the ground. And so they get permission from the store, and they take them home.

32:29And they've been reverse engineering them. There's like two or 3 ,000 people in a discord right now who are reverse engineering these things, figuring out how they work, figuring out the different problems with them and taking them home and i wrote one article about them but then there's this other part of it where there is a company called junk luggers and there's a few other ones that are essentially taking these to these giant kiosks and taking them to recycling centers and like trashing them more or less right um just before we get to them and sort of your trip out in the field getting one of these boxes, you said that this community who reverse engineering them, they go to the store, the Walgreens or whatever, they get permission, and then they take the box away.

33:20But wasn't there some sort of tension around, was it Walgreens corporate learned of this and then stopped giving permission? Like what happened there? Yeah, so I think for a while, I have not gotten internal Walgreens communications with managers. But it seemed like in the first few weeks after Redbox went bankrupt, Walgreens was like, oh, we need to get rid of these. And so individual store managers were just telling whoever wanted them that they could take them home. And I believe that Walgreens corporate eventually learned about this and said, please don't do that. And I have to assume that the reason is because they're like 700 pound devices that are bolted to the ground and they're connected to not just like a regular power plug, but they're like connected directly to outside power in a scary like you can get electrocuted to death kind of way.