AI Fake Protest Videos: The Global Slop Economy on Facebook

Cut their benefits and make them leave. It's as simple as that. >> This street interview has over a million

views on Facebook. It might not seem that weird at first. It's a British woman at an anti-immigration protest

somewhere in the UK. Here's the weird part. >> Cut their benefits.

>> Cut their benefits. >> Cut their Cut their >> Cut their benefits.

>> The system is completely broken. >> We need proper reform of the whole of the broken system.

>> It's as simple as that. >> It's as simple as that. >> Maybe you could already tell, but all of

these videos are AI generated. What's gone viral isn't just the videos themselves, but also the prompt used to

make them. It's part of a genre dominating Facebook in the United Kingdom. AI-generated content designed

to fuel anger online, particularly with anti-Muslim or anti-immigration messages. It's videos like these,

>> Why are migrants put first? [music] >> but also images like these, and news updates like these.

Maybe that's not too surprising. These are hot-button topics. Here's where it gets weirder, and you

might already suspect what I'm about to show you. These accounts are hiding a secret. Many

of them claim to be from the United Kingdom, but if you go to their page, click on their name, and click page

transparency, you can see that they aren't. Instead, they're actually based in

places like Sri Lanka, Vietnam, Cambodia, and the Philippines. And here's where it gets even weirder.

>> If we simply cut their benefits, they'll choose to leave on their own. >> These foreign-run Facebook slop accounts

are also targeting Australia and the United States with almost the same exact content.

>> Cut their benefits and they will deport themselves. >> I wanted to understand the ecosystem

that's motivating people in these countries to specifically post this type of content about these places.

>> It's a semi-dystopian system online. >> Newsfeeds are full of nightmarish slop. >> There's no sense of of the real impact

on people. >> I've spent the last few weeks deep in this strange corner of the internet

trying to understand why a certain kind of content meant for people in one part of the world has become an obsession for

people in another part of it. And the answer I found was much stranger than I thought.

One of the accounts posting these AI-generated UK protest videos was called I love UK.

I love UK is managed by people in Sri Lanka. You can see that when you look at a Facebook pages page transparency tab.

So, I went looking for more information about Sri Lankan AI slop operations, and I came across this.

Hate for profit. A 2025 [music] investigation into a Sri Lankan

commercial network on Facebook. >> [music] >> The network consisted of 128 pages and

groups which posted images like this targeting the UK [music] collectively reaching over 1 and 1/2 million

followers. >> Why are Facebook pages [music] in Sri Lanka creating this very

particular kind of content? >> Sri Lanka's a relatively small country. It seemed likely there was some sort of

coordination going on. >> Analysts at Dot Face Cut Aroma Krishnan and journalist Niamh McIntyre both

worked on the investigation. And I asked [music] them if I love UK seemed like it could be part of this network. Do you

have a record of like this particular page being part of the network? Uh is there a way to check?

>> Let me check. Yeah. Yeah. That that that's um we'd say most likely part of the group.

>> All 128 pages and groups in the network [music] linked back to an academy run by one person.

Geeth Soriya Poora. He's a creator who runs several UK politics pages and an online academy where he teaches students

how they can make this kind of content, too. >> He's been quite instrumental in, you

know, Sri Lanka being like a real hub for this because he has a an academy where he tutors and mentors other Sri

Lankans. >> On social media, Geeth projects an image of wealth and success. Proof for

customers that the business model works. And that was consistent with the little I could learn from trying to talk

[music] to the people running these accounts directly. I made a new Facebook account to ask [music] the people

running these pages a simple question. Why are you doing this? I heard back from a few of them. None were willing to

do [music] an interview. Just one account, I love UK, the same one we'd confirmed was likely part of

Geeth's network, responded with a simple explanation. They only needed money.

>> There is a huge impact of these individuals who who demonstrate you can make money doing it. And you know, you

know, have lifestyles that show that, you know, they are successful. >> So, that's how an account like I love UK

probably got started. Someone in Sri Lanka started watching Geeth's tutorials, learning the right prompts,

topics, and strategies, and started to post. But Sri Lanka is just one of many hubs

of this kind of content along with Vietnam, India, Pakistan, Bangladesh, Cambodia, Indonesia, and Nigeria. And

Geeth is just one of many tutorial creators teaching people how to make money by posting AI-generated content on

Facebook. If you watch those tutorials, you'll notice that a lot of them are teaching

people how to reach viewers specifically in the UK and US, like this one specifically showing how to make

AI-generated videos in the US politics news niche. >> At exactly 10:47 this morning

>> 10:47 this morning, a former president of the United States was one sentence away.

>> I wanted to understand why these specific countries were such a big focus for tutorial creators in a whole 'nother

part of the world. These are places with turbulent immigration debates, but so is all of Europe. These are

English-speaking countries, but hundreds of millions of people in India speak English. Why specifically get involved

with the American and [music] British and Australian immigration debates in particular?

This is Kashif Majeed. He makes tutorials about AI-generated content creation with over 2 million

YouTube subscribers. His content is mostly focused on finance and crypto, not inflammatory political topics. But I

reached out to see if he'd sit down and talk with me. And he did.

When I was a teenager and first starting to think about working in journalism when I grew up, one of the very first

tools I bought was this. A voice recorder. But voice recorders can involve a lot of steps. You got to

record, transfer files, upload to a transcription service, and then wait. [music] That's where the sponsor of this

video comes in. Pluad. Pluad makes recording devices that turn conversations into organized,

searchable knowledge with transcripts, summaries, key take aways, and action items generated automatically. I've been

using both the ultra-thin Note Pro for meetings, interviews, and calls, and the wearable Note Pin S for capturing

thoughts hands-free on the go. And after an interview, I can ask Pluad to go back through and find the information I need

without having to dig [music] through an entire recording. And they take privacy seriously. They're SOC 2, HIPAA, and

GDPR compliant. Of course, you should always make sure to follow local laws and obtain consent before recording

[music] conversations. Interviews are a big part of my job. If interviews, lectures, or meetings are a big part of

yours, this can be a really useful tool. If you want to try Plaud for yourself, go to plaud.ai/Kristof

and use the code Kristof at checkout [music] for 10% off. The link is below. Okay,

let's get back to the story. >> So, we are teaching people how to create AI videos and target US content. So,

like for USA people. >> I am always interested to see like USA niche is kind of its own thing. How

would do you describe what that means? What does it mean to make like USA content?

>> YouTube is too much low payment for Pakistani audience because their purchase power is low. So, in US and UK

and Europe countries or Australia or any other countries like white people, they have a very good

purchase power. >> We are all, for the most part, on the same internet and on that internet,

everyone's attention is for sale. But some people's attention is worth a lot more than others. The amount of

money that advertisers pay to reach a viewer varies significantly depending on where that viewer lives. This is a list

from the YouTube analytics site YT Face ranking every country by how much advertisers pay to reach viewers there.

[music] To reach a thousand YouTube viewers in New Zealand, for example, advertisers pay about $30. To reach a

thousand viewers in Thailand, they pay about 30 cents. This is called CPM or cost per thousand impressions. The

amount a creator makes is called RPM >> [music] >> or revenue per thousand impressions. And

it means that content creators around the world are incentivized to reach audiences in these [music] places,

especially if they live in these places. >> If we are creating content for Pakistan, but targeting USA, UK, Canada, it give

us too much money. Like you can say the RPM would be $6 to $30 per thousand views. And if we are creating content in

Pakistan for Pakistani audience, it's about $0.4, like 40 cents. So that's why people are

creating content for the US. >> If you were to say like, oh, this is what I've learned Americans like, what

are those categories of things? >> I would say the top three are finance, cars, reviews, and journalism, or you

can say news. >> People in the content creation automation world call these niches.

>> In finance, you can choose different niches, like crypto, like finance history, Elon Musk history, how Elon

Musk started. After finance, it's about car reviews. It's about news, like general news, like Trump news, Obama

news, and other presidents, and other USA news. >> Kashif mostly makes videos in that first

niche, finance and crypto. He told me that accuracy is an important part of his process.

>> I mentioned in all my videos that you should research before uploading. So the most most common issue is

>> [music] >> people are not so well educated. So you can say they are illiterate. They don't

research too much, and they just need views. It's a big responsibility. >> Kashif is right. Publishing content to

an audience on the internet is a big responsibility that requires doing good, accurate research.

But it did not feel like the people in this corner of Facebook were doing that. At the beginning of the story, I was

tempted to call these AI news slop channels misinformation. But as I kept talking to people, that

stopped feeling like the right name for what was going on. >> It's not that they're trying to mislead

or lie to people. It's just that they don't care about the truth.

>> Angus Bridgman is a political scientist who studies online misinformation in Canada. Angus's team fact-checked almost

6,000 posts from a network of spammy news pages targeting Canada and found that only about 54% contained false or

unsupported claims. >> So, just like a total coin flip of truthfulness on this content. Being

untruthful is not the point. >> [music] >> They they could not care less if they

are truthful or not truthful. It is simply what gets engagement. [music] >> And while making content about a

country's issues used to take a certain amount of expertise, that barrier to entry has fallen.

>> We we said the barrier to entry of engaging in mass propaganda used to be very high. Before you would have had to

hire a writer or a researcher, you'd sort of like look up what's going on in that country, let's investigate it.

You've got like a country expert. Now, you can go on [music] ChatGPT, you can go on Cloud and say, "Hey, what's going

on in that country? Write me some copy to do this thing." >> Which means that these creators don't

need to have many technical skills or be familiar with the country's politics or even speak its language to create highly

[music] specific, polarizing, and engaging content. >> It's as good as a cat video or as good

as someone doing those like AI cutting glass sculptures. It's it's just something that performs well.

>> You're just going to use AI to produce as much as possible as cheaply as possible.

>> These creators don't care about UK politics or about the immigration debate. They've just AB tested their way

to figuring out what upsets people in rich, English-speaking countries the most.

And this is what that happens to be. >> One of the pages one day for some reason put out like a refugees welcome image.

Either they were trying to do AB testing to see whether that would drive up some clickbait anger. Or, I suspect it's the

case they just because there's so little understanding, you know, these are people just pouring stuff out. They just

put that up without even recognizing what they've written. And then they went back the next day to, you know, refugees

out. It's this complete lack of concern with what's actually being produced beyond the fact that it makes

money. >> One thing that's jarring about watching tutorials from the Sri Lankan tutorial

creator Geeth is how news is just another kind of slop topic alongside cute animals, ASMR, arts and crafts,

nail art, magic, and archaeological mysteries. And as I learned more, I started to

understand just how keenly aware the platforms are of this. In 2026, it feels obvious to say that

stuff that makes people upset [music] performs well on the internet. But back in 2018, Mark Zuckerberg posted a

fascinating visual explanation of how that [music] works. He called it the natural engagement pattern.

>> I call it the Zuck curve. Um >> Jeff Allen used to work as a data

scientist [music] at Facebook and now runs the Integrity Institute, a think tank dedicated to improving the social

internet. [music] >> So, the X axis is like how bad is the content and the Y axis is like how much

engagement does it >> Past a certain point on the bad scale, content is prohibited outright. Things

like graphic content, nudity, bullying, or fraud. But what the Zuck curve showed was that

the closer content gets to that [music] line, to being prohibited, the more engagement it gets on average.

>> That chart is a perfect way to understand why any algorithm that is designed to maximize engagement will

inevitably float harmful content to the top. >> Platforms will, in theory, take down

harmful misinformation [music] or hate speech. But content that skirts that line without crossing it gets more

engagement. The numbers back this up. A 2026 report found that Facebook accounts posting low credibility content get nine

times more engagement than high credibility accounts with the same follower count. The same thing plays out

on other platforms, [music] too. On X, it's 10 times more engagement, and on YouTube, 11 times more.

Okay, so ad rates that incentivize trying to reach people in rich countries, AI tools that make it easy to

do that, and viewership patterns that reward making inflammatory content. Those are true of any platform on the

internet. But the problem of AI-generated news inflammatory slop is particularly visible on Facebook. And

the final thing I wanted to understand was why that platform. I thought that maybe

it was because of who's left on Facebook as many users have migrated to other platforms. But it turns out that the

answer might have a lot to do with how Facebook pays creators. Some of these slop channels link to an

external site where they make money on banner ads like this, but many of them don't. They don't have to because

they're making money on the platform itself. >> You can earn money using multiple types

of content once you're in the Facebook content monetization program. >> Since 2024, Facebook has directly paid

people for content via its content monetization program, which lets users share a portion of the ad revenue on

videos, photos, and even just text. The more engagement on those posts, the more money they get. Creators had previously

been able to make money on content via a few different programs, but this streamlined all of those into one

program. So even though there have always been reasons to farm engagement on Facebook, the incentives really

changed at this one >> [music] >> moment.

This is a screenshot of payments that one Indonesian creator shared. She received these for making posts to

groups supporting Albertan separation from Canada. There are now almost 18 million accounts in Facebook's content

monetization program, all shaping what kind of content and what kind of content creators Facebook is really for.

According to Facebook's policies for this program, content that provokes debate about social issues may face

restricted or reduced monetization. But, given the existence of this massively popular genre of content, that appears

to be poorly enforced. Pages also aren't supposed to be able to run meta ads that attack people on the basis of protected

characteristics like national origin. But, >> No more illegal migrants.

>> These pages seem to be doing exactly that to grow their audiences. And as meta has rolled back its

third-party fact-checking and content moderation, more of those violations may pass undetected.

>> [music] >> In the Facebook group run by Geeth, the Sri Lankan tutorial creator, one member

posted this image, personally thanking Mark Zuckerberg for how much content monetization has helped people in Sri

Lanka. You can find posts like this all over Facebook, celebrating how the program

has helped people earn a little extra income and support their families. >> This is a a story in the sense of people

in the global south making money and freeing themselves from manual labor, and it's just in the worst possible way

you can imagine it, and just spreading hate. >> There's a bizarre irony to what this

program means to people around the world and what actually gets made because of it.

>> These creators aren't anomalies or or sort of rogue actors, they are actually perfectly behaving in the way that the

[music] tech industry at large is sort of encouraging them to. >> When I look at all of the content being

made because of this program, I don't see bad actors. I see hustlers who are almost certainly completely oblivious to

the nuances of politics in Western countries, responding to the incentives that a platform has created [music] for

them. Responding to a program that will literally pay people to post almost anything so that it can sell more

eyeballs to to advertising industry, and to an advertising industry [music] that prioritizes reaching certain people on

the internet more than others. And to us, who inevitably wind up engaging the most with the content that's just barely

allowed on the platform. That confluence [music] of factors has made something designed to make people in Western

societies as angry as possible extremely profitable in the global south. Every viewer of content [music] like this is

viewing it with all of the messy, complicated political feelings they have about the country they live in, but all

of the creators creating it are coming to the table with none of that. They're just doing [music] it for money.

And every time they make money, Meta does, too. And as long as that's the case,

it's hard [music] to imagine any of this slowing down. Thank you so much [music] for watching.

This is a really fascinating episode to work on. If you have an idea or a question [music] or a topic that you

want to see covered on this channel, you should send me an email at our tip line, which is [email protected].

That's going to be the best place to send any ideas and start conversations with with me. A few weeks ago, I asked

members of our creator journalism platform New Press for their thoughts about how misinformation [music] has

changed on social media networks. Their feedback was really helpful helping me frame the big picture questions behind

this story. It's a topic that I want to keep covering a lot more of on this channel. And if you want to follow

along, you should head to newpress.com, sign up, [music] join the conversation, and keep an eye out for more

opportunities to shape my reporting. Okay, that's it. I'll see you next time. Okay, that's it. See you next time.