Why the internet is filling up with nonsense ‘AI slop’

Elon Musk riding a UFO, “Shrimp Jesus” and tiny children baking impossibly perfect birthday cakes.

These are just three examples of bizarre AI-generated imagery that has taken over the internet in the past year.

Across Facebook, X, Instagram and TikTok, surreal AI images have been plaguing news feeds, typically accompanied by nonsensical captions in broken English.

Dubbed “AI slop”, there is no hiding from the manipulated images and videos, with their prominence even leading to the Oxford University Press (OUP) naming “slop” as a contender for its word of the year.

Slop, the OUP writes, is “art, writing, or other content generated using artificial intelligence, shared and distributed online in an indiscriminate or intrusive way, characterised as being of low quality, inauthentic, or inaccurate”.

There have long been fears that AI image tools could be used to create highly convincing “deepfakes” and images or videos that could help spread disinformation or influence elections.

But so far, the most viral AI images have been obviously fake and downright strange. Free software for creating images and videos has led to a surge in slop, driven by scammers, spammers and the occasional genuine user seeking to go viral.

Facebook’s most recent report from its Transparency Center reveals that two of the top five most widely viewed images on the network in the three months to September were AI-generated – viewed 38.6m and 35.8m times respectively.

Matt Navara, a social media consultant, says the novelty of these posts has quickly worn thin for many of Facebook’s billions of users.

“The proliferation of low-quality or spammy AI content risks cluttering feeds and diminishing overall user satisfaction,” he says.

“I’m seeing little evidence yet that users are actively demanding such content in their feeds.”

In March, Facebook users began to notice a flood of images featuring Jesus Christ mixed with a crustacean rising from the sea. These “Shrimp Jesus” posts garnered hundreds of thousands of interactions and tens of millions of views.

Other weird AI trends soon emerged, including African women building fruit sculptures and young children showing off elaborate birthday cakes with the caption: “This is my first cake”.

On Facebook, these posts were published on pages with innocuous names such as “Easy Recipes”, “Interesting Planet” or “Life Nature”. A study in March by researchers at the Stanford Internet Observatory found that many pages filled with AI content were likely attempts by scammers to cash in on virality.

They found 120 pages posting improbable AI images with clickbait headlines, many of which shared the same creator or published multiple posts with the same captions.

Some garnered a huge response despite their bizarre nature. One popular “Crab Jesus” had 209,000 engagements and 4,000 comments.

“We suspect these high levels of engagement are partially driven by the Facebook recommendation algorithm,” the report said.

The AI images were “captivating visually, easy to create, cheap to generate in large numbers [and allowed the pages] to engage in high rates of posting of new content that might grab people’s eyes”, says Josh Goldstein of Georgetown University, one of the report’s authors.

He called this phenomenon a new kind of “engagement bait”.

Among the pages highlighted by the report were many that appeared to have been hacked and turned into AI content farms. These pages sought to direct users away from Facebook to websites that could then be used to eke out a trickle of advertising income.

Meanwhile, on YouTube and Telegram, a network of influencers – many from India or the Philippines – are busy coaching their followers on how to make money from slop.

In one video, an influencer tells followers to create viral images depicting an old man being eaten by insects.

“The Indian audience is very emotional, after seeing photos like this, they like, comment and share them,” he says, claiming that users can make money through a Facebook scheme called “Performance Bonus”, which provides a tiny fee to creators of viral content. His video promises followers they can make 4 lakh (£3,700) per month.

Other AI images have also been shared widely during some of the biggest news events of the year – and not just by grifters and scammers.

In May, a pro-Palestinian Instagram template created using AI, originally designed with Microsoft’s Image Creator, was shared more than 50m times, including by celebrities such as Bella Hadid. The template featured a crude refugee camp disappearing towards the horizon with the words “All Eyes on Rafah”.

In October, Republicans in the US took to sharing AI-generated images in the wake of Hurricane Helene to criticise Joe Biden’s response plan.

Amy Kremer, a Republican activist, shared a post on X of a crying girl holding a puppy during a flood. “Y’all, I don’t know where this photo came from and honestly, it doesn’t matter,” she said.

Then, as the US election approached, a new surge of AI images featuring Republican donor Elon Musk spread rapidly across Facebook.

This included posts claiming that the Tesla founder had created a flying saucer.

However, it is not just images that are the problem.

Newsguard, the news rating service, has identified more than 1,000 websites pumping out AI-generated fake news stories in 16 languages, many of which appear to have little or no human oversight.

McKenzie Sadeghi, AI editor at Newsguard, says: “Our tracking of AI-generated news websites has found that it shows no signs of slowing down. The barriers to creating AI-generated content remain low and the incentives – programmatic advertising revenue, site traffic, engagement – remain.”

Tech giants have taken some measures to separate fake from legitimate posts, including Facebook, which labelled some posts with the tag “AI Info”.

But the industry has also been aggressively pushing more users to adopt their own AI tools, such as Meta’s AI assistant or X’s Grok.

Meta has also started encouraging users to create their own fake images, while also promoting other AI-generated content on Facebook and Instagram feeds.

However, with a flood of more low-quality content on the horizon, some technologists are urging Meta to reconsider the prominence given to these posts.

“If platforms value human creativity, they should label and downrank outputs in which human creative involvement is minimal,” says Ed Newton-Rex, founder of Fairly Trained, a non-profit advocating for tech companies to respect creative rights.

He adds: “I think the latent demand for spaces that are human-first is higher than many companies realise.”

Some AI experts have suggested that over time, Meta and others will adjust their algorithms to deprioritise the most egregious AI slop – much like spam filters. However, Meta’s own stated plans suggest otherwise.

Data from the company’s Transparency Center shows that in the three months to September, more than 31pc of content viewed on Facebook was from accounts “unconnected” to the user – meaning it has been served up by an algorithm.

This is up from 8pc in 2021 and 24pc this time last year. Meanwhile, on Instagram, a growing number of AI videos have been populating users’ feeds and Reels.

Mark Zuckerberg has also made it clear that he expects the internet to gradually become populated with more AI posts, believing that is exactly what his users want.

“I think there’s been this trend over time where the feeds started off as primarily and exclusively content for people you followed, your friends,” he told The Verge.

“We’re also going to show you content that’s generated by an AI system that might be something that you’re interested in … how big it gets is kind of dependent on the execution and how good it is.”

This is a far cry from Zuckerberg’s previous goal of spreading “meaningful social interactions”, as it seems apparent that Facebook and other online platforms show no sign of slowing up in displaying mindless scrolling content.

Broaden your horizons with award-winning British journalism. Try The Telegraph free for 1 month with unlimited access to our award-winning website, exclusive app, money-saving offers and more.