There Are No Girls on the Internet - The Racist AI TikTok Scam That's Flooding Your Feed

Episode Date: April 14, 2026

Fake AI-generated Black women are flooding your social media feeds, performing self-hatred, pushing porn sites and drop shipping scams, and raking in cash for anonymous creators who'll never be held accountable. Jeremy Carrasco, the internet's go-to AI debunker, breaks down who's behind it, why platforms keep looking the other way, and why one Black student researcher's testimony made him want to do something about it. AI Blackface: Profiting on racist depictions at scale: https://www.riddance.ai/p/ai-blackface-profiting-on-racist AI videos of sexualised black women removed from TikTok after BBC investigation: https://www.bbc.com/news/articles/c070e283k8vo Bridget explains the evolution of digital Blackface from its antebellum roots in this Youtube video: https://youtu.be/yh-2il78AEA Digital rapper FN Meka and the new era of digital Blackface: https://omny.fm/shows/there-are-no-girls-on-the-internet/digital-rapper-fn-meka-and-the-new-era-of-digital Let us know what you think by emailing [e-mail] or leaving a comment on Spotify. Pre-order Bridget's forthcoming audiobook about AI and intimate relationships at LoveAtFirstPrompt.com ! Follow Bridget and TANGOTI on social media!  ||  instagram.com/bridgetmarieindc/ || tiktok.com/@bridgetmarieindc ||  youtube.com/@ThereAreNoGirlsOnTheInternet || bsky.app/profile/tangoti.bsky.social See omnystudio.com/listener for privacy information.

This is an I-Heart podcast. Guaranteed Human. Another podcast from some SNL, late-night comedy guy, not quite. Unhumor me with Robert Smygel and friends. Me and hilarious guests from Bob Odenkirk to David Letterman help make you funnier. This week, my guest, SNL's Mikey Day and head writer, Streeter Seidel, help an a cappella band with their between songs banter.

Where does your group perform? We do some retirement homes. Those people are starving for banter. Listen to humor me with Robert Smigel and friends on the IHeart radio app, Apple Podcasts, or wherever you get your podcasts. Run a business and not thinking about podcasting. Think again. More Americans listen to podcasts than ads supported streaming music from Spotify and Pandora. And as the number one podcaster, IHearts twice as large as the next two combined. Learn how podcasting can help your business. Call

844-844-I-Hart. Hey, it's Edwin Castro, also known as Castro 1021. And I'm Konki, his best friend and business manager. And we've got a new show called The 1021 podcast. I'm taking you behind the scenes on how I became one of Twitch's most popular streamers. We also love sports. And with the World Cup right around the corner, we'll be breaking down the biggest storylines ahead of the big tournament here in the USA. Listen to the 1021 podcast on the IHart Radio app, Apple Podcasts, or wherever you get your podcasts.

Hey, I'm Deanna Maria Riva. And on my new podcast, How Hard Can I. Can it be? I call on my Gen X squad from Ohio to Hollywood as we navigate Midlife's most fantastic BS. Unfiltered conversations from night sweats to fupas to scheduling sex. Wait, what sex? Is it just me or does every woman my age want to look at Pinterest instead of having sex sometimes? They say we can't polish a turd, but we're sure going to try. So let's get blunt with laughs, tears, or tears of laughter. Listen to How Hard Can It Be with Diana Maria Riva on the IHeart Radio app, Apple Podcasts, wherever you get your podcasts.

There are No Girls on the Internet as a production of IHeart Radio and Unbossed Creative. I'm Bridget Todd, and this is There Are No Girls on the Internet. Someone, probably a guy in a basement somewhere, has figured out that if you generate a black woman and make her say degrading, racist things about herself, you can make money on the Internet. Probably not a lot of money, maybe just a few hundred dollars a month, but enough to keep on doing it. TikTok, Instagram, and every other social media platform, they're serving it up on your social media feed right now. It's AI-generated digital blackface,

and it is absolutely everywhere on social media. There are accounts built entirely around it, fake black women with impossibly dark skin, performing self-hatred, and funneling viewers to things like porn sites and drop-shipping scams. So this is not one of those future AI problems we might have to grapple with down the line, this is happening right now.

And Jeremy Carrasco is one of the people actually trying to do something about it. He's kind of become the AI debunking guy on social media. He started out as a video production guy working in live streaming and technical producing, and that background has given him a sharp eye for spotting AI fakes. When Jeremy partnered with the BBC to expose over 100 of these AI-generated accounts,

TikTok actually listened. pulling 20 of them down within three days. It's a start. But as Jeremy will tell you, it is a very small dent. My name's Jeremy Carrasco, creator at Jeremy Finds AI and director of riddance. So I love that you kind of come at this from a technical place as somebody who got, you know, was working in live streaming, making content. I feel that so much of the conversation around AI has really been set by tech companies and their sycophants, frankly. and so it's just nice to have a different kind of voice.

Do you feel like that producer, live streamer, video, I has helped you develop a framework around AI deepfakes? It's the voice that I wanted to have and the background that I felt like was needed in this space. I think exactly what you said. When I created my pages in June of last year, the main reason was that a lot of the ways people were talking about AI video, they were pretending that these were tools for my industry.

like the one that I knew people in, the one that I didn't see many good uses for in general. So I just wanted to speak up and be like, hey, no one really wants this. Like, we're not asking for this. Stop using the industry as a reason to push out AI slot making tools. Just be honest that you're making AI slop or you're trying to maybe replace Hollywood down the line. But I felt like a lot of it was just either inauthentic or like very, very rose-colored glasses that people were going to start making independent films with AI videos.

And in fact, a lot of the opposite has happened. A lot of people who are true creatives have rejected this. But it has definitely encroached on transactional spaces. And I spent a lot of my time in places of video commerce, where a lot of that is kind of under threat by AI. So that's one of the reasons I started was I just felt like a lot of that was disingenuous or too rosy. But the technical background I have,

I had no idea that this is what it would be used for. When I was a live streaming technical producer, I was also what they call an engineer in charge, which means sometimes I would have to visually check lip sync, like very, very specific thing, down to like a frame of accuracy. I didn't know that that would come in handy for looking at AI lip sync later

or looking at errors in encoding and video would come into play when I was looking at AI videos. So, yeah, it's a very weird, niche skill set, but that plus my obsession with internet culture and politics kind of brought it to a more media literacy place rather than just spotting AI video because I think that AI video spotting is a part of media literacy toolkit that some people don't have. So like in that case, they should have other parts of media literacy toolkit because it's going to get harder and harder. I'm not under any illusion that people are going to be able to do this for

that much longer. This is a little bit of a non-securter, but to your point about how when video tools, when AI video tools like SORA were being talked about, I don't know if you see this on your social media feeds, but I cannot escape it and it drives me fucking crazy where AI sycophant, I guess, like, DeVotay's will say, did XYZ AI platform just end the need for Hollywood or CGI and then they will post the wackest looking most out of context, ridiculous video and I'm like, are you all

are you seeing what you posted? You're saying that this is going to like like, like this is what people are going to pay to watch? Are you, are you really saying this? So there are two things here. First, like the whole Hollywood is cooked thing is so funny because Hollywood's not exactly like a burgeoning

profitable industry right now. Just like stick to tech. Like why are you trying to to enter? this like industry that is being threatened in so many different ways. So just the obsession with Hollywood as like a cultural phenomenon is just so funny to me, especially like the anti-Hollywood people. I'm not a big Hollywood person, by the way. I'm a social media person.

I've never worked in Hollywood. I just think the fascination is funny because I specifically wasn't working in Hollywood because I didn't think it was the place to be. Anyway, and that's nothing against the people who love movies. It's just not how I grew up. The other thing that is like missing here is we, have to ask ourselves, like, why we watch things and why we would watch that in the context that they're presenting it. So, like, why would you go to a movie theater to watch something if AI could make anything? Then, theoretically, nothing really matters on, like, aesthetic grounds

and it's all about the meaning that it would have. And if you can't get that meaning out of the artistic work that was put in, or even, like, the celebrity that you like going to see, like, then what is the point? And I don't think that they've thought that far ahead. I think there are a lot of useful tech demos, but they haven't gotten a good way around that point, which is if you can make that at home on your desktop, then why would anyone go watch it in the first place?

It doesn't mean anything. So it's a little bit of like a problem. It's like a zero-sum game. We have too much content out there. How do we decide what we want to watch? It's not just because it looks like Hollywood. it's because it means something or we have some sort of connection to it. Yeah, it doesn't mean anything is a great way to put that,

where I'll watch these clips that people are telling me are so phenomenal, and I feel nothing. They just feel so flat. And my listeners all know, I am a movie person. And it's like, the thing that I feel when I go to the movies, it's like a very specific thing. It's not just, I guess the attitude that boils that down into AI, made this very quickly and cheaply,

that's not the thing that gives anything meaning. Yeah, and just to be clear, I like to say that I'm not a movie person because I've seen an embarrassing few number of movies. Okay. I still love seeing movies sometimes. I just like whenever, you know, there's like, I'm embarrassed by it. The, whenever I go though is because I really want to see something.

I really want to watch it. And, you know, that doesn't just come because of like some impressive technical feat that happened. You know, I just, I don't think that we, do a good job of separating why we watch certain things because I think the natural tendency is like well people just doomscroll they watch Slop so like yeah but like we're not watching we haven't been

watching social media videos and movie theaters this whole time like there the bar to to create things has never been lower it was never lower before AI media like we were still figuring out reasons to watch something on our phone versus something in the theater there weren't a lot of movies coming out that

like specifically used iPhone cameras even though they were way cheaper because you have to rejigger your entire workflow around an iPhone camera. You have to remake your entire workflow around AI video production if you're going to do that because you can't cut between a 4K cinema camera and a 720P AI video. That's what all of them still are by the way. You can pay extra for 1080p or 4K, but they still look like AI videos, especially if you're looking at them on a big screen. So there are technical limitations to this happening. There are also like meaning relationships that just aren't being made with AI videos.

But I think a lot of people who are super into it are just like, they're into the AI hype, they're into the technology. So they just don't see that because for them, the purpose here is the technology. That's like, that's the entire point. I just don't think normal people do that. I don't know if this is true or not, Jeremy, but I read that you actually started from a place of being quite optimistic about AI

and that initially you wanted to do a YouTube channel about kind of how people can use AI video tools. Is that right? I read it in Verge, I think. Yeah, it just, that was before Google Vio3 even came out. I mean, it was basically, like, I chose the, my pages used to be called show tools AI. Because I was, I was going to make shows with tools and some of those were going to be AI, very witty. Except for the fact that, like, Google came out with Vio, the week that I was going to launch my page,

I had to, like, stop everything I was doing and being like, what is happening here? And yeah, I wasn't really ever thinking that I was going to use, like, AI video to make movies. I was more interested in how it could be used for storyboards, maybe, or how it could be used for green screen backgrounds or stuff. And you can actually see me experiment with some of that and, like, try it out in my, like, my early days. And I was very, very upfront with it. I disclosed all of it. I was like, I was trying to be, like, the demonstrator of, like, how to label and do a responsibly. And then I just realized it just didn't really mean anything to me at the end of the day. I didn't find it was that

helpful. And I just hated the process. I just hated prompting AI videos. And I still do. I just think it's boring. I think it is not an efficient way to make media. I think that there's a lot of reasons why you can't convey an entire scene over text and you do have to have some like visual medium to tell that's like just more efficient for me. I don't know. I can point to can. camera at something in some ways quicker than I can write it given, you know, the right context. So, you know, the reason that Hollywood works the way it does and the reason they haven't moved over to a phone is because this is harder to use in the context that they are making movies. They need more control. Their producers expect more control. That's the same way I felt about AI videos. It did

not give me the control that I needed in order to make the thing I wanted. So then what was the point? And the whole thing that I think it's kind of ontologically evil and it stole a bunch of data. There were a bunch of other reasons that wasn't about it, you know. Is that really what changed for you from going from like a kind of an AI optimist to being like, actually, I don't know about this. This maybe isn't the thing I thought it might be. Yeah.

I mean, I don't think I was ever an AI optimist. I think I was always an AI pragmatist and like, how are we going to use this? Like it's here. How is it possible? And then I just saw a surprising amount of reaction against AI. I don't think it was a given at the time. when I started my pages and I was starting to do like these AI video spotting videos, people are like, they don't care, they just want to watch whatever, they'll just do and

scroll, they'll watch Slop. No one cares, but it turns out people do not want to be tricked. Even if they want to watch Slop, like they want to watch Slop and know it's Slop. They don't want to watch something real and find out that it was AI later. That's just like a really bad feeling. We want to have real connections with people. We want to assume that we're interacting with real people. And oftentimes the, even the Slop interactions, are like you're meta-aware that you're watching Slop and you're engaging with the comments and like you're all meming each other.

It's like it still can be a social thing. But I think that it was just pretty early when I started and I knew it was early. And the learning process was really quick because all of a sudden a lot of people cared about AI video and I just didn't think that would ever happen because I was also under the impression that people didn't care.

And I'm glad that I was wrong. I spend a lot of time talking about people lying on the internet specifically for nefarious means, for social or political control or to influence people. But a lot of people are just good old-fashioned financial grifters trying to make some quick cash, especially when it comes to AI-generated content on social media. And I think that understanding that dynamic, the power of algorithms and platform monetization to push everyone toward extremist, click-baity content, is central to understanding.

the internet in general. Jeremy says that accounts chasing that monetization or other scams are very different than someone who was using AI to just make slop entertainment content, like those fruit Love Island AI microdramas that recently went viral. On our podcast, we talk a lot about things like misand disinformation, AI has been a huge part of that conversation. And a lot of times when we're talking about it, we frame it as just inherent. kind of nefarious, right? That it is that AI and other forms of mis and disinformation are manipulating people for some sort of political or social ends. But from watching a lot of your content, I've come to realize that, yes, that is true, but also a lot of times, it's a lot simpler than that.

People are just trying to make money. This is a little bit of a financial grip. Do you see it that way? And if so, is one of those motivating factors like worse than the other? There are too many reasons people would use AI video. The first is just to make slop. And this is something interesting. So people understand that I'm like the AI debunking, AI spotting guy. And so like, you know, the AI Fruit Love Island videos, for example,

I was not tagged once. No one brought it to my attention at all. I had to find out like everyone else did just through culture, publicate, you know, like, whatever. Like, it was just the normal way. Learned about it on the street. Yeah, yeah, out on the streets. Exactly.

Most of the times I find out about videos because people tag me in them, people send me them. And, you know, if an AI video gets over 100 to 500,000 views and I'm not tagged in it, sometimes I'm a little bit surprised or I'm worried because no one's caught it. But not one person tagged me in that because people knew. People were aware. Now, a lot of the things I get tagged in are scams, exactly what you said, which is just like a drop shipping scam that.

pretends that their goods are handmade. That is the thing that I'm seeing everywhere right now. I'm tagged in at least five or ten of those every day of like this new Chinese good that pretends to be handmade. And a lot of the times those are also reusing media from people who are actually making those goods and rescinding them with AI. And it's a very frustrating thing for those real people to go through. And then I also see coordinated campaigns or just miss or disinformation. but, I mean, I think a lot of the people who, you know, there is a reasonable, there is a reasonable viewpoint that an AI person, so like a person who shows up and tells you something, like you should not trust them because they don't have any accountability.

You don't know who they are. They don't have a history. They don't have a past. It's completely reasonable to see this and go, I don't trust you. And therefore, I'm just going to disagree with you because you have to check what they are saying anyway because they don't have any stakes. They don't have any accountability. So at that point, I don't think that they're a very reliable piece of information. You're not going to get good information from them. This is actually different than using a large language

model. You can ask a large language model questions. You can ask it for sources. You can, you can, debate with a large language model. So it's not even just that AI itself can't tell the truth about things. But if I see an AI avatar saying something on social media, I'm going to not trust it because I'm going to ask where this is coming from. I think that that's a good. instinct that people have. And we're seeing it used a bunch of different ways, but like, you can largely throw a lot of that information away. And I'm glad that that is the direction we're headed. I'm glad that that's the instinct people have rather than the opposite right now. I want to get into this. You mentioned the sort of AI generated fake, fake handmade good stuff.

I see so much of that. And I will say, like, a lot of what I see is also I guess I would almost call it like racialized guilt marketing where it'll use some sort of, it'll be an AI generated black person selling something using a kind of racialized sob sorry to sell it. So it'll either be, I got, I am the victim of a racist hate crime, so buy my handmade item. Or I am a black person who makes a handmade item and the only people that I trust to support my business are white women. So white women buy my handmade item. What is going on here? So there are unfortunately examples that I've seen where you, you know, basically have a A and B experimentation on like one of them is a white woman, one of them is a black woman and the black woman

usually performs better. So there is something there. And it's usually because they're both using exploitative language like you just said and that performs better. for the black character. I've also seen pages that have inconsistent characters so they might go from like a white guy selling dog supplements to a wise Asian man selling dog supplements

and then it's just like boom, it just takes off. And part of the reason, if I want to be very charitable about it, is there something just like shocking about this? It's like real people don't manipulate people like that or if they do, they're shamed and it's embarrassing. And so there's just something about AI, like that's an arbitrage opportunity for AI

because no one has to be held accountable and they can literally just like make a racist comment in front of a black character. And like that's okay for them because it's just a white guy making them usually. It's like some dude in Kosovo. Like I see that a lot. And so I, you know, it's an exploit that AI is uniquely able

to manage as far as like what it says about us. I mean, I will I will leave that to people smarter than me. I can see the patterns. I can I hate it. I've seen it like manipulated in very big and harmful ways. And I've actually, you know, the articles I've done about it, I oftentimes like will like ask a black woman how it makes them feel because I can't feel that the same way. I like I feel uncomfortable about it. But like it's harmful. It's harmful. to have these depictions out there, just like in general. Like, it, it is inherently harmful. It makes people feel bad when they see it. And, like, that harm is being done. And, like, that's not harm being done to me. But it is something that I can spot and make people aware of. So that's the role I've taken.

Let's take a quick break. Another podcast from some SNL late night comedy guide, not quite. Unhumor me with Robert Smygle and friends. Me and hilarious guests from Jim Gaffigan to Bob Odenkirk, to David Letterman, help make you. funnier. This week, my guest,

SNL's Mikey Day and headwriter, Streeter Seidel, help an Acapella band with their between songs banter. There's that worst singer in the group? The worst? Yeah. Me. Is there anything to the idea that because you're from Harvard, you only got in because your parents

made a huge donation. The group. The yard birds, right? That's the name. The Harvard Yard. But they're open. Do you have a name suggestion? We're open. Since you guys are middle-aged. Uh, one erection. Listen to humor me with Robert Smigel and Friends on the IHeart Radio app, Apple Podcasts, or wherever you get your podcast.

Humor me. I need some jokes to make me seem funny. Run a business and not thinking about podcasting, think again. More Americans listen to podcasts than ads supported streaming music from Spotify and Pandora. And as the number one podcaster, IHearts twice as large as the next two combined. So whatever your customers listen to, they'll hear your message. Plus, only IHeart can extend your message to audiences across broadcast radio. Think podcasting can help your business.

Think IHeart. Streaming, radio, and podcasting. Let us show you at iHeartadvertising.com. That's iHeartadvertising.com. What's up, fam? It's Isaiah Thomas. And I'm C.J. Toledano, and our podcast's point game is about defying the odds. Like LeBron heading into the playoffs without Luca and Austin Reed.

And finding ways to win no matter what. He's the smartest player to ever play the game. His IQ is at a level that we've never seen before. And he knows without Luca and Austin Reeves, I got to manipulate the game. We get a player's perspective on the challenges of the playoffs. I think Joker's going to be exhausted this series because when they don't have Rudy in the lineup,

he has to really guard guys like Nas Reid. He has to guard Julius Randall. And then he has to give us everything he gives us on the night-to-night basis on offense. And when IT's friends stop by like Quentin Richardson, we dive into some playoff history too. Steve Nash will get that thing. That man, hell get the flying.

He running up the court, licking his fingers why he got the ball. Like, you go through a training camp with that, Isaiah, you figure it out real quick. Oh, yeah. Get your ass up and down the court, and you're going to get the ball. So listen to Point Game on the IHeart Radio app, Apple Podcasts, or wherever you get your podcasts. Hey, I'm Deanna Maria Riva, actress, mother, lover, and a Gen X woman walking through life, one hot flash and hormonal crying jag at a time. You ladies know what I mean.

I'll bet you a perimenopausal chin here you do. So let's talk about it. Join me on my new podcast. How hard can it be with the Adamani Arriva, where I call on my Gen X squads from Ohio to Hollywood as we navigate midlife's most fantastic BS. All of a sudden, I'd had hanginess happening on my own. I was like, what the hell is that?

I was married when I had her, so I didn't even consider how empty that nest was going to be. Swings, night sweats, fupas, sex drive. Wait, what sex? Dating at 45. How high can it be getting naked at 50 with the new guy? That one's kind of hard, no. Well, that's lighting.

They say we can't polish a turd, but we're sure going to try. So let's get blunt with laughs, tears, or tears of laughter, and dive into it, unfiltered and unbothered and ask, how hard can it be? I cannot believe I'm about to say this out loud in public. Listen to How Hard Can It Be with Diana Maria Riva as part of my Cultura Podcast Network available on the IHeart Radio app, Apple Podcasts, or wherever you get your podcasts. I're back. I've been talking about digital blackface, both AI-generated and otherwise for a while.

Like when Google's V-O-3 first came out and racist skits depicting AI-generated black women as guerrillas, or depicting AI-generated enslaved people on plantations, blooded TikTok. I really thought that tech platforms should have had to answer for it. but I was pretty much screaming into the void, and stuff like that just became more and more ubiquitous. Earlier this year, a black student researcher named Angel Nulani reached out to Jeremy. Angel had been seeing those very same overtly racist AI-generated videos depicting black women. She had found 40 social media accounts with AI-generated black women that perpetuated racist and anti-black behavior. An angel told Jeremy how it feels to use.

social media platforms as a black woman, when those same platforms are awash in this kind of racist, misogynistic garbage. She said, what upsets me is not that these characters are self-hating, but that there is no self. For the majority of the people who are behind these accounts, the only black people they know are the women they generated. They were not born black. They chose to be black, and yet they spend so much time distancing themselves from it. All Lives Matter shirt can be paired with booty shorts. Crime statistics can be overlaid on thirst traps. Videos crying over their features are spliced between suggestive yoga poses.

It's not enough to pander to people that say eminy more than African, because the appeal is not simply a black woman desiring white men. The appeal is a black woman desiring a complete absorption into whiteness. The men they attract are not suitors, their savior. I've tried so hard to disassociate. As a black woman that's been online since I was young, I assumed I saw it all. Nothing really hurt me. At first, all I wanted was to be objective and rational and done with it. But this cuts me down to my core. So that's what Angel told Jeremy, and I really understand how she feels. In a collaborative investigation with the BBC, Jeremy worked to uncover

100 social media accounts that depict the likenesses of black women through AI-generated characters. TikTok took action and removed some of these accounts from their platform, and I'm really glad they did. But Jeremy says that doing work like this sometimes just feels weird because the content is so deeply racist and exploitative and gross. But he's also very aware that it isn't really his pain to bear, like it is for black women like me and angel. Yeah, I wanted to ask you about this. wasn't even sure how to ask this. We have been talking about digital blackface for a very long time, both AI generated and other kinds of what I would consider to be digital blackface. We did an episode about it way back when in 2022 where this like AI generated rapper named FN Mecca was

released. We had a whole conversation about it when SORA and VO3 were being used to create these skits that I think you and I would probably agree are like essentially digital minstrel shows. we did an episode about it. I don't know. In my opinion, that, especially when V-O-3 and SORA were becoming things, that was like an urgent moment to me. I was like, this is, we need to like have this conversation. And I was kind of surprised that that moment did not get the traction

as like an urgent change that I would have thought it would. And when I came across your videos, I was like quite thrilled to see them getting such traction, especially the videos that are about, you know, race and the way that black women are depicted in these things. And I guess my question, and I'm not even sure if this is something you could speak to, but do you think there is an element of, of like, who is the messenger here? Because I wonder if like me as a black woman saying, these videos make me feel X, Y, Z. These videos are trafficking in X, Y, Z, misogynistic, racist trope. I think that there's something where a lot of people can just be like, oh, it's just black women complaining about something again.

And then when you speak about it, not just from the perspective of a white man, but also somebody who, you know, has this technical background. Like in your substack piece, you really break down, like, the color gradients of why the kinds of skin that is depicted in some of these videos is actually impossible and what that means. I am really grateful and happy that you are talking about this and amplifying it. But I wonder, if like, if there's like a thing with the messenger there where it's more, people are more like, oh, well, this kind of technical guy is saying it, so we better pay attention. I agree. I think that that's pretty likely. I also am very uncomfortable with that because like, I know that's why I'd,

so I'll say, I actually have not talked about this. This is a great time to do it. So the way that I found out about the, well, not found out. So I had seen other, uh, um, um, um, um, um, I had seen other, uh, the AI-generated sexualized black women before. And I feel uncomfortable talking about it because I think that the depiction is quite harmful. And what actually made me write that article was I had a young woman who she's black and she had found a lot of these accounts

and she sent them and she's credited in the piece. Her name's Angel Nalani. And I was really thankful that she was there because she was able to be like the emotional core and like I wasn't I didn't feel like I was like taking advantage or something so I felt really bad like I don't want to get views on something exploitative. I don't like I do feel like a discomfort here and I think that it's a discomfort on so many levels it's all new. It's like this is this is all like quite new like the technology layer obviously you know racist depictions and digital blackface aren't anything new but like this is a new kind of intersection of like technology and and race and and so. I think it's like a really difficult thing to navigate. So the reason I'm answering this way is like I agree that there's something there. I struggle to navigate it because I really don't want to be taking advantage of these stories anymore than I should be.

I would love other people to talk about it. And I, you know, my opportunity to platform Angel is someone who like, I want you to be a mispiece because like you came to me. Like this is personally affecting you. You told me that this is a difficult thing for you to go through. I want to like pay you to do this so that like you can be a part of this. And like I don't think that's like, I'm not doing that to like pat myself in the back or something. I'm just like I'm aware of the discomfort there. And it doesn't like, I'm very, very meta aware of this.

I think that just kind of the nature, like I'm very careful about like, you know, the language. And it's actually, it keeps me from making a lot of videos sometimes. It like definitely puts a pause because I don't want to, um, the problem worse by showing more depictions. When we did the story that you're referencing, we went through a ton of different variations on if we were even going to show the depictions, if we were even going, like how we were going to show them. When we worked with the BBC, the BBC decided not to name any of any of these accounts. We decided to name the accounts. That's a difference in how we decided harm could be done, right? The BBC decided that by directing

people to these, it could be harmful, whereas I wanted to call them out. And so I think that there just isn't one answer, and I'm seeing those things all play out in real time and navigating them and just kind of doing my best. But I agree, and I do hope that, you know, I can provide any sort of platform I can to people who aren't white men to help, because it's not, it really is like, AI is very good at making a lot of identities really quickly. It's going to impact a lot of different types of people. Yeah, it was really clear from the piece, the intention and care and thought that you would put into kind of framing it. And it is like just a tough growth thing that have to talk about.

The quotes from Angel, I thought, really were so resident. She says, like, I'm terrified to think there's a little black girl somewhere feeling as terrible as I do right now because some guy in Malta wanted to make three euro a month. Or, you know, they make us seem so ugly, but we're so beautiful. like all of these, I just was really moved by a lot of what she had to say. And so I'm grateful that her voice was included in the, like, throughout the piece. And it's just very clear that you all are taking very intentional steps to talk about something that is like sensitive and tough and thorny to talk about. Because I don't feel like everybody would be, a lot of people would be like, what a scoop.

Can't wait to get a million views on this TikTok about horrible. racist, ugly depictions of black women. I also, there are similar things happening with the way that children are depicted online that have been very troubling for me. A lot of stuff related to, like, frankly, there was one, like example I could give. And this is not, this is not to say

that someone is doing this the wrong way or right way. I'm just, like, stating the difference that I would cover and how I would cover it. Like, there were a lot of AI depictions of girls with Jeffrey Epstein that got a lot of views by people on Instagram who were like pointing out of me like, this is wrong. And like, I don't feel comfortable doing that. I just, I don't. I don't feel like there's a responsible way for me to cover that. And so like when I was doing, when I, when I did videos about the piece we're talking about, like I decided to blur them all

out because I don't want to be trafficking in the same like shock value that a lot of these have. Like, there is a lot of shock value here. And I think that's unfair. But I, Again, I don't think that there's one right or wrong way to do it. I think that it depends who your audience is and what your angle is. Thanks to Jeremy's investigation alongside the BBC. Within three days, TikTok banned 20 of the 37 accounts that his team found pumping out this racist AI-generated content. TikTok also applied AI-generated labels to the videos that Jeremy and his colleagues pointed out.

So we're talking about this investigation you did with the BBC. from that investigation, TikTok banned 20 accounts for using these AI-generated black women influencers to drive to sites. Some of them were driving to, like, porn sites where you could then pay for porn. The accounts are, they're upsetting, I guess I'll say. They're overtly racist. Yeah, they're overtly racist. Yeah. And a lot of the comments, a lot of the posts will kind of use these like race play terms,

Like the post will be like, I love white men and things like that. I, it's clear to me what's going on. But I think to your point, it's one of those things that like, it's difficult to talk about like, oh, someone has figured out that using super racist depictions of black women might generate clicks back to this pay porn site. And to the point that you made earlier, there's certainly, you know, adult content that is super racist. racist and horrible and violent. But the idea that you can create that kind of content without a human, right? So like all of the harm that you that would be resulting of that, you could do that without any person signing up and consenting and being like, okay, I'll do this.

Like you can just make it quickly and cheaply. You can, there does still have to be a human in a loop right now, just to be clear. So like, I think we still have to remember that. But there are very few. And so it just takes one person rather than a team. of people and someone to be like, this probably isn't the right thing to do. And then there are people who are just intentionally making racist depictions and know it and they don't care.

And this is a very cynical way to just to view how to use these tools. And so I think that there are different types of accounts that's something else I wanted to say. There are accounts that are run internationally that are not as overtly racist and they really are just using like a black woman's likeness, which is problematic, but they They aren't doing anything different than a similar white character that they might be running. So they're just like running different identities. But, you know, so there were a gamut. You know, we found like 136 different accounts.

And they weren't all the same, but there were a lot of trends for sure. And I would say that probably over a third were overtly racist and over a half were suggestive or at least like dabbled in raceplay terminology in some sort of direct way. And then there were some that were just like way too far, even for Instagram. And Instagram didn't take his money down, but Instagram took a few down that were extremely problematic too. So, yeah, there were a lot of them, but they had a lot of the same patterns. Do you have a sense of where videos like this originate from in the world? Everywhere. It is an international problem. We did track across, I think it was five different continents.

I think Australia wasn't included in an article. are the only two not included. It was pretty wide-ranging, I would say. I think a lot of people will make assumptions, but this is something that anyone can do from really anywhere in the world. And there were operations run, so there were accounts that were run by content farms,

and then there were solo accounts. It's a pretty wide range. So I try not to pin it down anywhere, but you can get location data from Instagram, especially if they let you see it. And those can be spoofed too. So it's not always the most accurate, but it at least tells you that there's a range.

Let me ask you this. So I know that TikTok did take down a lot of these accounts. Instagram was like, they're actually, we're looking into it. What explains the difference there? So TikTok will, so TikTok actually cited exactly the podcast, that they broke and they were very direct about it.

And part of the difference is that Instagram doesn't have as direct of language about this in their policies and their terms of service. Their inauthentic media is more protecting real people from being deepfaked or copied, whereas TikTok takes a broader view of inauthentic behavior. So that's the first difference. The second difference is that Instagram is just generally where a lot of content is and that's because it is more permissible of it. So like there were, I think, 96 accounts that we tracked on Instagram and it might have been 36 on TikTok or something along those lines.

And like the fact that we reported more of them and few of them got taken down on Instagram is also why there are more on Instagram. So it's a little bit of like a self-replicating cycle there that Instagram has built. That also means, ironically, that videos I do about it do better on Instagram too. So it's like, it's affecting everyone. If I make a video that includes so much as like an AI woman wearing a bikini, I have to blur that on TikTok. Otherwise it'll get shadow, you know, deprioritized, right? Like these are just things we're aware of as creators. So it's, it is consistent with how they each run their platform. I think that's what I'm trying to say. Yeah, that makes sense to me, just as someone who has covered a lot of meta and their policies and how they

abide by or don't abide by those policies, what you're saying really checks out to me. I guess checks out. Yeah, not sometimes. Yeah, I didn't have to think that hard about it. It was consistent. Consistent. That's a good way to put it. One of the videos that you talked about in the piece took a real model from Malaysia and then gives her that impossibly dark skin using AI. And just as a black woman to me. Like it almost feels like it is using technology to create this impossible standard of beauty where you, where it's a woman who is not black who has so like, of course, has non-black features, but then has not just dark skin, but like impossibly dark skin. And I guess I just feel that it's really is using technology to create these people that cannot exist. But,

of course, like, also then sets an existing beauty standard despite that standard being impossible. Yeah, I got so frustrated. And that's why I actually did the whole thing about actually detecting the undertones to prove that these skin tones were impossibly black. Because people would be like, oh, no, there really are people that with that dark a pigmentation. I'm like, no, there aren't. Like, they have no undertones. And I actually, like, I actually found the people they were talking about and I put them on vector scopes. I'm like, I can tell there's a scientific difference here.

There is a difference. And, you know, it's like the models that would create these are open source models, so they don't have the guardrails to prevent you from doing this anyway. I don't know if there are guardrails. I haven't tried to make many impossibly black characters. I'm not very interested in doing it. I don't know if you're allowed to do it on like Google's AIs, for example. I could probably test it.

But the ones that they were using are open source anyway that can create sexual content. Like Sora, which is closing down soon, VEO, even like Mid Journey. These like big names have guardrails. So they're using open source models that are a little bit less high quality but can make sexual content. So they're much harder to stop. More after a quick break. Another podcast from some SNL late night comedy guy, not quite. Unhumor me with Robert Smygel and friends, me and hilarious guests from Jim Gaffigan to Bob Odenkirk to David Letterman, help make you funnier.

This week, my guest, SNL's Mikey Day and head writer Streeter Seidel, help an acapella band with their between songs banter. The worst singer in the group? The worst? Yeah. Me. Is there anything to the idea that because you're from Harvard, you only got in because your parents made a huge donation. The yard herds, right? That's the name.

The Harvard Yard, but they're open. Do you have a name suggestion? We're open. Since you guys are middle-aged, one erection. Listen to humor me with Robert Smigel and Friends on the I-Heart Radio app, Apple Podcasts, or wherever you get your podcast. Humor me. I need some jokes to make me seem funny. Run a business and not thinking about podcast.

Podcasting, think again. More Americans listen to podcasts than ads supported streaming music from Spotify and Pandora. And as the number one podcaster, IHearts twice as large as the next two combined. So whatever your customers listen to, they'll hear your message. Plus, only IHeart can extend your message to audiences across broadcast radio. Think podcasting can help your business. Think IHeart. Streaming, radio, and podcasting.

Let us show you at IHeartadvertising.com. That's IHeartadvertising.com. What's up, fam? This is Isaiah Thomas. And I'm C.J. Toledano and our podcast Point Game is about defying the odds. Like LeBron heading into the playoffs without Luca and Austin Reed. And finding ways to win no matter what. He's the smartest player to ever play the game.

His IQ is at a level that we've never seen before. And he knows without Luca and Austin Reeves, I got to manipulate the game. We get a player's perspective on the challenges of the playoffs. I think Joker's going to be exhausted this series because when they don't have Rudy in the lineup, he has to really guard guys like Nas Reid. He has to guard Julius Randall. And then he has to give us everything he gives us on the night-to-night basis on offense.

And when IT's friends stop by, like Quentin Richardson, we dive into some playoff history too. Steve Nass would get that thing. That man, hell get the flying. He run up the court licking his fingers while he got the ball. After you go through a training camp with that, Isaiah, you figure it out real quick.

Get your ass up and down the court, and you're going to get the ball. So listen to Point Game on the, I Heart Radio app, Apple Podcasts, or wherever you get your podcasts. Will Ferrell's Big Money Players and IHart Podcasts presents Soccer Moms. So I'm Leanne. Yeah. This is my best friend, Janet.

Hey. And we have been joined at the hips since high school. Absolutely. Now a redacted amount of years later, we're still joined at the hip. Just a little bit bigger hips. Wider. This is a podcast. We're recording it as we tailgate our youth soccer games in the back of my Honda Odyssey.

With all the snacks and drink. Sidebar. Why did you get hard seltzer instead of beer? They had a bogo. Well, then you got it. Do you want a white collar something here? Just hit it. Oh, what are y'all doing?

Microphones? Are you making a rap album? Oh, I would. Come on. How did you believe? I would buy it. Cuts through the defense like a hot knife through sponge cake. That sounds delicious.

Oh, you're lucky. I'm not a drug addict. You're lucky I'm not an alcoholic. You are. I'm lucky I'm not a killer. I love this team, and I'm really trying to be a figure in their lives that they can rely on. Oh. Oh, oh.

Listen to soccer moms on the IHeartRadio app, Apple Podcasts, or wherever you get your podcasts. Let's get right back into it. The last time I did a deep dive into AI-generated content on social media, I noticed just how often people who make this kind of content insist that you two can make passive income just by setting up an AI influencer. Now, I had no proof, but the whole thing. just had the whiff of a scam to me. Like when a girl you went to high school with DMs you about a business opportunity

where you can make lots of money and be your own boss from home. You know, when we first did our episode when V-O-3 and Sora were first, like, becoming things, one of the things that I... It was early days that I haven't really revisited it since, but one of the things I saw a lot about this kind of content was it had a lot of almost like an emmer

MLM quality to it where people would be making kind of not all racist, but some really racist depictions of black people using AI. And then there'd be sort of a like, oh, and you can, I make so much money from this. I can teach you how to do it too. I can teach you how to create an AI influencer and start making cash. And at the time, I had no idea. But my, my thought was, I was like, I think that this is one of those things where I don't know that a lot of AI influencers are making a lot of money because if they were, they probably wouldn't also be selling the like course on how to make money because that seems like the grift here.

I don't know. What are your thoughts on that? It is both. So some of the most racist ones were what you were talking about where they were primarily selling courses to people. And the reason that those are the most racist is because they're mostly trying to grab people, usually men, who are creating these. And so it's like the shock value that gives you to look and then say like,

oh, maybe I could do this, which is saying a lot about like the type of people that oftentimes create this. And I mean, yeah, anyway. So the, the ones that are making money are oftentimes like your bigger accounts. And a lot of those aren't selling like AI making courses. So like they are actually making money primarily on the off site porn sites that they direct people to. They're, their media and I have not bought any of this.

to be clear, but like the prices for the AI content they're selling are like really ridiculously high. So you can tell what that means they're trying to get like what betters would just call whales. Like they're trying to just get some people who just have a bunch of money to spend and like that's how they make it. So, you know, they get their big audience on Instagram and they just hope that, I don't know, maybe a hundred people spend a bunch of money on their off sites. And that's how they're doing it. But there's a bit of both. What role do you see AI having on the influencer economy? I'm pretty worried about it lowering rates for everyone. I don't know too much about the influencer economy, ironically, because I don't take sponsorships, so I haven't seen it firsthand. But I think that the concern is that a lot of those direct consumer style advertisements can just be made using AI now. Whether or not they will is a different question. It is,

still something that is pretty easily caught and pretty embarrassing, but a lot of brands aren't doing their own marketing. They're going to agencies and they might be pretty hands off. For some reason, I've seen a lot of AI advertisements for like Cetafil, the skincare company. I use that every day. Yeah, but I don't think that they're doing it. I think that someone is making the ads and then doing an affiliate link on Amazon. So it's like, that's the other tricky thing is you don't always know if they're actually associated with it or it's just like an Amazon affiliate thing. So it's tricky to tell, but I mean, obviously a lot of the AI companies are selling this as, yeah, you can just have AI influencers and not communicate with regular influencers. But I also think that the reason

people like regular influencers is because they have a relationship with them. So I think they're also missing the point. Yeah, we talk about that a lot. You know, we, the reason why, Like, as a podcaster, people are often like, oh, you know, are you worried about AI podcasters coming for your job? And your whole podcast could be AI one day. Not really. I'm not really that worried about it. I don't like the idea of it. But I know that the reason why people come to podcasts and my podcasts, but podcasts in general, is for things that AI is not so good at fostering.

Trust, authenticity, a connection with a real human. I just don't, I mean, if people wanted to listen to AI podcasts, there are plenty of AI podcasts that exist already. You could go listen to them right now. People aren't listening to them right now. And I think there's a reason for that. It's just a general misunderstanding. And I think a little bit, like, people don't think much of social media or they think, they don't think about like why people are actually on social media or why people are listening

to podcasts instead of the other types of media that they know people are watching less of. And it's generally because it's more personal or it feels more authentic to them. So I don't think that's changing anytime soon. there will be people who don't value that as much who might switch more to the AI versions. I'm worried about that, again, like maybe being a negotiating tactic if you're an advertiser, like,

well, I can just get an AI to do it? And, you know, it's like, well, can you? I don't know, but that could at least be a tactic. So I think that it still comes with risk, but I don't worry about it as much as I would if I were like a purely transactional TikTok shop creator, maybe. Then I might be a little bit worried.

Yes, and I have found that so much, at least in the podcast space, so much of the conversation of like, oh, AI can replace humans is being, it's hype that is being promoted by people who are selling AI tools. And so they are invested in people believing, especially like decision makers and people who are writing the chats, they have an investment in those people believing that that is what people want, even if maybe it isn't. Yeah, absolutely. And so many people are selling it. It does feel a little bit NFT, a little bit, a little bit, you know, web three. era where it's just like you feel like you have to get people in on it. Ironically, you really don't need, with crypto, you really did need other people to buy in for it to go well. You don't need to do that here, but a lot of those people are coming over to the AI product space and vibe coding apps and stuff like that. So yeah, I absolutely agree. That's the kind of vibe it has to me. And of course, it always comes down to the thing that you flagged as the piece that worries you is how can we underpay and undercut human labor. How can we disrespect human labor? I hate that that's always

what it comes back to, but that does seem to be what it always comes back to. Yeah. And again, I would just remind people that like people do like working with other people and like I think small businesses are, you know, going to keep working with other people. A lot of the bigger corporations I'd be probably a little bit more worried about if I were them. But I mean, it's going to happen, like natural, like technology will keep happening. And I think it is unclear if this is actually just to be clear, moving faster than any other like normal technology would do. And so far as like how it's impacting labor,

a lot of like layoffs that are being done are done in the name of AI, but are really efficiency cuts maybe from over-hiring of COVID. Like I don't know these things. They're not transparent about it. I think it's very likely that AI will have a massive impact on labor. I think that that is, I mean, almost guaranteed at this point, if I'm being honest. And yet, like, I think that it will also go by many names. They might be more excited to call things AI because that does have usually positive benefits to their stock values.

So, like, we just have to be careful not to, like, overhype it, even when it is a real risk. Very well put. And I've heard you mentioned that you think that AI detection will eventually be outpaced by AI quality. the improvements. I don't know if this is a question that you can answer, but do you, how urgent is that window? What needs to happen before that window closes? So I don't think that software detection is going to save us here, at least not in the next few years. And I think that people assume it will. I don't know why, except for the fact that they're used to software solving their problems, but there's no indication that we're really that close here,

if I'm being honest. Like, AI photo detection is not that good. AI video detection is very bad right now. You still have media literacy tells right now that are very helpful. And if you're a platform, like, you have access to that data and you should probably take it more seriously, in my opinion, but their moderation is so under-resourced that they're not doing that. So there's, like, a lot of problems kind of all converging at once. But I think that we have an opportunity right now to figure out what is real or not,

like a little bit more on a day-to-day basis. and as AI video and photo keeps improving, it's just going to keep getting harder. I'm just trying to be realist about that. So I would like platforms to take it more seriously now because it is technically still a little bit possible and we don't know where it's going to be in the future. I don't think that it's going to be completely undetectable anytime soon,

but that's not really what I'm worried about. I'm more worried about it, making people believe that real things aren't real en masse than anything else. Yeah. I mean, I guess to sort of wrap it up here, you, I think that you often are kind of seen as like the deep, the AI deep fake spotter debunker guy on the internet. But as you mentioned, it's really more of a broader set of media literacy tools. Do you have tips for what folks should be looking out for? Full disclosure, I've admitted it on the podcast. It turns out I have a real blind spot for AI generated. animal videos. And I've kind of been put in group chat time out because my friends are like, it's AI, girl, are you not seeing this? Those bunnies are not jumping on this guy's tummy. It's fake. So give us some tips for what we can do to be a little bit sharper when it comes

to media literacy, given all of this. So the first is, I think reposts accounts are pretty much dead to me. I just don't think that they're useful anymore. I think they're really likely to repost AI content right now. And part of the reason is that if you're not getting, like, algorithms make it possible to just get things from original sources. Like, you don't need another reposter to curate your algorithm for you. That's one thing. The second is that, um, this goes to a little bit of like why right now is important is even

if, uh, AI videos get much better in the coming months, that means that accounts are going have to keep updating their content, deleting their old stuff that's more obviously AI. So as of now, a lot of deceptive AI accounts are less than four months old, or you can scroll back to their old content and be like, okay, that's more clearly AI generated. If you were ever good at spotting it before, if you weren't good at spotting it before, then that might not be that helpful for you. And the other thing is just really spend more time watching, like, people and creators you trust rather than having the algorithm decide for you because the algorithm can't figure out

what's real any better than you can right now. So I think like I spend probably less time watching AI videos than other people because I watch more long form content and like less long form content is AI generated unless it's faceless YouTube. Like there are exceptions. Unless it's like your AI generated podcast that we are talking about. But, you know, I'm really looking for people I trust and following people I trust and trying to tailor my recommendation algorithms to that as well. If you think about it as more of a people problem, it becomes more manageable. It's like, do I trust this source? Do I trust this person?

You just, you're not going to be able to just spot your way out of this. It's just not possible. It's going to be exhausting at the very least. I find that more and more people are tuning out before they are trying to actually figure it out. So take that burden off yourself, like spend more time watching people you trust and less time dune scrolling and trying to figure out what's real. It's just not going to be a thing we can do for that much longer. Jeremy, you are a person I trust.

Where can folks follow you and all of your content? I'm at Jeremy. Jeremy finds AI, my handles on all social media. And then riddance.aI is our substack and where we do investigations. So check us out there. Thank you so much for being here. You are such, I really, I get tagged in your content a lot. And I'm always like, yes, that's AI. We're like, oh, interesting.

I haven't seen this. So thank you so much for what you're doing. I feel like it makes me smarter and like a better digital citizen, I guess. Got a story about an interesting thing in tech or just want to say hi? You can reach us at hello at tangoody.com. You can also find transcripts for today's episode at tangoody.com. There are no girls on the internet was created by me, Bridget Todd. It's a production of IHeartRadio and Unbossed Creative.

Jonathan Strickland is our executive producer. Tari Harrison is our producer and sound engineer. Michael Amato is our contributing producer. Edited by Joey Pat. I'm your host, Bridget Todd. If you want to help us grow, write and review us on Apple Podcasts. For more podcasts from IHeartRadio, check out the IHeart Radio app,

Apple Podcasts, or wherever you get your podcasts. Another podcast from some SNL, late-night comedy guy, not quite. Unhumor me with Robert Smigel and Friends. Me and hilarious guests from Bob Odenkirk to David Letterman help make you funnier. This week, my guest, SNL's Mikey Day and head writer Streeter Seidel

help an a cappella band with their between songs banter. Where does your group perform? We do some retirement homes. Those people are starving for banter. Listen to humor me with Robert Smigel and Friends on the IHeart Radio app, Apple Podcasts, or wherever you get your podcasts. Hey, it's Edwin Castro, also known as Castro 1021. And I'm Kunky, his best friend and business manager. And we've got a new show called The 1021 Podcast.

I'm taking you behind the scenes on how I became one of Twitch's most popular streamers. We also love sports. And with the World Cup right around the corner, we'll be breaking down the biggest storylines ahead of the big. tournament here in the USA. Listen to the 1021 podcast on the Iheart radio app, Apple Podcasts, or wherever you get your podcasts. What's up, fam? It's Isaiah Thomas.

And I'm C.J. Toledano. It's our favorite time of the year on our podcast point game, the playoffs. We're digging into the biggest surprises of the season. And I'm looking back on some of my greatest playoff moments. If we didn't talk ever again, I was calling you. You just understood. That's how personal it got. Wow.

Then after that game seven, Marquis come until he's like, you. know I love you, dog. You know, it's all love. This was just playoffs. This was just basketball. So listen to Point Game on the IHeart Radio app, Apple Podcasts, or wherever you get your podcasts. Will Ferrell's Big Money Players and IHeart Podcast presents soccer moms. So I'm Leanne.

Yeah. This is my best friend, Janet. Hey. And we have been joined at the hips since high school. Absolutely. A redacted amount of years later, we're still joined at the hip. Just a little bit bigger hips. This is a podcast.

We're recording it as we tailgate our youth soccer games in the back of my Honda Odyssey. With all the snacks and drinks. Why did you get hard seltzer instead of beer? They hit a bogo. Well, then you got them. Listen to soccer moms on the IHeart Radio app, Apple Podcasts, or wherever you get your podcasts. This is an IHeart podcast. Guaranteed human.