King of slop: How anti-migrant AI content made one Sri Lankan influencer rich

By Meriem Mahdhi , Niamh McIntyre, Mark Sellman

LONDON – Sri Lankan influencer Geeth Sooriyapura’s Instagram feed shows off his lavish lifestyle. He and his friends flash expensive-looking watches, dine at a five-star seafront hotel, and film videos at a sleek modern apartment with a pool.

He claims to have made his money – $300,000 of it – by running Facebook pages, including some which push racist, Islamophobic and anti-migrant posts aimed at British audiences.

He said he targets “old people … because they are the ones who don’t like immigrants,” in an interview with The Bureau of Investigative Journalism (TBIJ).

A joint investigation by TBIJ, the Institute for Strategic Dialogue (ISD) and the Sunday Times reveals that the self-proclaimed ‘Facebook ads king’ gets paid to teach others the secrets of his success, promising that they too can turn Facebook into a “money bank”.

In total, TBIJ and ISD found 128 Facebook pages and groups targeting UK audiences – with a combined following of 1.6 million users – linked to Sooriyapura and his students.

Sooriyapura said that he did not encourage his students to “spread violence”.

“We just educate people on Facebook monetization and audience targeting, so after that, students target an audience of their own interests,” he said.

Meta said it removed the pages flagged by TBIJ that violated its policies on inauthentic behaviour.

Turning hate into gold

In March, one of Sooriyapura’s pages posted a false claim, misrepresenting Sadiq Khan’s pledge to build 40,000 new London council homes by saying the homes would only be available to Muslims, so they could be “near Mosques and Halal food shopes [sic]”.

Outraged commenters on the post called for the Mayor of London to be deported or hanged; some raised the possibility of civil war. It is unclear if any of these users know that the page was set up to blog about food in Sri Lanka and is still run from the country, profiting from their outrage.

A spokesperson for Khan said, “Fake and racist content online has real-world consequences. It can stoke hatred as well as fuel prejudice and division.”

“The rise of AI is supercharging the spread of disinformation, making it easier to create highly convincing content at scale and increasingly realistic deepfakes.”

“Social media companies must work harder and faster to protect users and prevent this kind of content from proliferating on their platforms.”

The page is typical of those run by Sooriyapura and his students, many of which share AI-generated content about UK politics, often focusing on inflammatory topics.

In a video reviewed by TBIJ, Sooriyapura recommends that his students target the UK because audiences are highly engaged with political topics, particularly immigration, which he calls a “strong trigger” for engagement.

“The UK is an important audience,” he says. “They don’t really like people from our countries living there – not just Sri Lankans, but even more so Indians.”

Tapping into this market is easy, he claims: there’s no need for complex video creation, and often a photo combined with emotionally charged music can be enough to attract attention and boost reach.

One post which used this simple format received 50,000 views by turning a photo of mostly Black men in an overcrowded boat off the coast of Italy into a 15-second video with melancholy background music. The image text reads “Share if you want Britain to put its OWN PEOPLE before foreigners looking for handouts!”

The pages run by Sooriyapura and his students, many of which have generic patriotic names, also post content made with a new generation of AI video tools, which allow users to create videos using written prompts. In one video, Sooriyapura tells his students that AI-generated videos can help political content go viral up to 10 times faster.

One AI-generated video shows a large naval ship colliding with two inflatable dinghies full of people – representing those crossing the Channel to seek asylum in the UK – who are then thrown out of the boats and into the sea. The caption reads: “Who really wants to see something like this?”

A previous TBIJ investigation found TikTok content creators using similar AI-generated racist videos – including one of people on small boats being threatened with a rocket launcher – to amass followers, and even make money through the platform.

Sooriyapura and his students also use generative AI to brainstorm content for the pages: in an instructional video, one student explained how he has used ChatGPT to research UK politics and identify “trigger points,” such as ongoing protests.

On the Facebook pages for his academy, which claims to have taught 2,500 students, several posts brag about the money the UK pages, which post generic patriotic content and news headlines alongside misinformation and hate, bring in. In one video, Sooriyapura hosts a ‘dansala’– traditionally a celebration where Sri Lankan Buddhists give out free food to passers-by from a roadside stall – for a group of students who focus on engaging a British audience, toasting to their success with chocolate cake.

Sam Stockwell, a researcher at the Centre for Emerging Technology and Security, said: “We’ve seen the rise of generative AI tools that now obviously allow a lot more users … to create very realistic, tailored content that can be easily scaled.

“While typically we’ve found that this is often done in relation to trying to manipulate voters … This is an interesting case where we’re starting to see more of a convergence of financial motives and disinformation.”

Siddharth Venkataramakrishnan, an editorial manager and analyst at ISD, said: “Commercial disinformation has received less scrutiny than politically-motivated content, but the lack of an ideology doesn’t make it less harmful.”

Gaming the algorithm

Sooriyapura’s method involves creating pages and groups catering to a particular content “niche”: some he recommends include lifestyle, cats and wildlife, religion, flowers, nail art or politics. He says the latter, which includes generic patriotic content and news headlines alongside misinformation and hate, gives content creators a high chance of going viral.

In another instructional video, Sooriyapura shares his screen, which appears to show earnings of $1,400 (£1,060) in a month from just one Facebook page – around a third of an average annual income in Sri Lanka.

He says he makes money through Meta’s “in-stream ads” program, which allows advertisers to place adverts before, during or after videos, and between short-form reels.

His videos also recommend allowing Facebook users to pay for subscriptions to popular pages, and taking advantage of Meta’s performance bonuses – direct payments to creators of viral content.

His methods appear to yield results, at least for some of his students. In a Facebook group run by Sooriyapura, one student posted an AI-generated image of Meta CEO Mark Zuckerberg with a large pile of cash. “Thank you very much Mr Zuckerberg, for helping people in Sri Lanka,” he wrote.

Meta disputes how lucrative the pages in the network were.

In 2024, Meta estimated that it would make $16bn – 10% of its annual revenue – from adverts promoting scams and banned goods, according to a recent Reuters and Indicator investigation. TBIJ reporting raises further questions about the nature of user-generated content on Facebook that Meta and its creators both profit from.

Taking sides

The UK politics pages in Sooriyapura’s network have a strong anti-Labour bias, with Khan and Keir Starmer singled out for criticism. Numerous posts call for them to be arrested or share AI-generated images and videos of them in jail.

The pages have also shared multiple AI-generated images and videos of fake mass protests calling for the mayor of London and the prime minister to resign.

Anti-Labour posts often have an Islamophobic bent. One falsely claims Starmer, an atheist, will introduce sharia law, another features an AI-generated image of the Prime Minister in Islamic dress, which the Prime Minister has never been photographed wearing, with text which reads: “Ostarmer Bin Laden Wanted For Terrorist Crimes Against White Christians Pensioners and the Disabled.”

By contrast, the pages are supportive of far-right British parliamentarians, like Reform UK leader Nigel Farage. An AI-generated image of Farage shared by multiple pages was accompanied by a pledge that under a Reform government, there would be “welfare for British citizens only.” Though the image is fake, Farage has pledged to halt benefits for non-citizens if he becomes prime minister.

Several Reform UK figures and official pages have also engaged with posts shared by the network, apparently unaware of its financial motivations.

Bill Piper, a Reform UK councillor in Leicestershire, commented, “I’ve been saying it for ages” on an AI-generated image of a man wearing a t-shirt printed with the slogan “support the country you live in or live in the country you support”.

Nigel Clarke, a former Reform UK parliamentary candidate who came third in Warwick and Leamington in the 2024 general election, also reshared a post from one of the pages.

Emily Darlington, a Labour MP who sits on the Commons science, innovation and technology committee, said she is planning amendments to the forthcoming Elections Bill to crack down on misinformation.

“We need to look at how we improve and tighten up the regulation regime here in the UK,“ she said. “We need to find a way to stop this monetisation of hate.”

– thebureauinvestigates.com