As social media feeds fill with AI-generated images, even the tech savvy find it difficult to distinguish between what's real and what's fake, writes Associate Professor Wendy Sloane
Date: 10 September 2026
I admit my online reading habits are a little unconventional. I’m a longtime admirer of conjoined twins Abby and Brittany Hensel, and I’m fascinated by the resilience and determination of 26-year-old Loren Shauers, whose body was cut in half following a grisly accident. After years writing “triumph over adversity” articles for women’s magazines, I’m often drawn to bizarre stories that don’t captivate others.
That may explain why my social media feeds are packed with an endless stream of extraordinary human-interest stories, accompanied by striking videos and images. Increasingly, though, it is impossible to tell which are genuine and which are AI-generated.
According to Ofcom, more than half the population now gets its news from social media, rising to around 75 per cent among 16 to 24-year-olds. At the same time, those platforms - TikTok, Instagram and YouTube - are being flooded with a new kind of content: “AI slop”; fake images, videos and posts designed to look authentic and provoke strong emotional reactions to maximise – and monetise – online engagement, regardless of their authenticity.
AI slop does not simply spread falsehoods; it changes the environment in which we encounter information. When journalism, satire, advertising, propaganda and AI-generated fiction appear together in the same endless feed, the boundaries between truth and fabrication begin to blur. The result is a paradox: we may become less able to recognise falsehoods while also becoming less willing to believe anything at all - even the truth.
My own social media feeds offer a revealing case study of how quickly these systems learn what captures our attention - and how readily AI-generated content exploits it. Knowing my fascination with eclectic human-interest stories, I'm regularly served a smorgasbord of fabricated photos and surreal reels.
Doesn't have to be believable
AI slop doesn't need to be consistently believable. It depends on keeping us scrolling, reacting and sharing. The more convincing it becomes, the harder it is to separate authentic reporting from synthetic fiction. Even when users suspect the content is AI-generated - or it is labelled as such - millions still engage with it.
The stakes become considerably higher, however, when AI slop moves beyond entertainment and begins to shape public life. Political deepfakes are among the most powerful examples: the widely circulated image of Trump depicted as Jesus on Truth Social, shared not long after he posted a video portraying the Obamas as apes; a fabricated video of Volodymyr Zelenskyy surrendering to Putin, and fake images showing a York Labour politician handing out cash to men in balaclavas.
Every convincing deepfake makes authentic photographs easier to dismiss, credible reporting easier to question and legitimate evidence easier to undermine. Making matters more complicated, AI’s large language models, or LLMs, are becoming a major source of “truth”, although they are trained on an information environment already distorted.
Media literacy experts argue that the best defence may not be better AI detectors, but better digital habits: pausing before sharing emotionally charged posts, checking whether stories appear in multiple reputable outlets, and tracing claims back to their original source. Yet these habits run directly against the incentives of social media, where speed, outrage and novelty are rewarded far more than caution.
The greatest danger of "slopaganda" is not that it fools us once. It's that it trains us to stop believing anything at all. In an internet where everything can be faked, nothing must be proved. And when engagement matters more than evidence, truth doesn't simply lose the argument. It loses the algorithm.
Read a longer version of this article in the British Journalism Review.
Dr Wendy Sloane is an Associate Professor on London Met's Journalism – BA (Hons) course and Deputy Dean of the School of Computing and Digital Media. She started her career in 1988 as a reporter-researcher for Time Magazine, based in New York and then Vienna and Moscow.