AI Fruit Videos: The Radicalisation Risk to Children
If you’ve got a teenager in the house, or you’ve spent any time on TikTok yourself lately, you’ve probably already seen the fruit videos. Odd little AI-made dramas where a banana cheats on a strawberry, or a watermelon steals someone’s boyfriend. Millions of views. Daft, harmless, good for a laugh.
Except some of those same characters are now starring in something much darker, and researchers are warning it could be a way of nudging young people towards radicalisation.
This week, Sky News reported that violent AI-generated clips of fruit and veg characters could help radicalise young people, according to researchers looking into the trend.1 Reports elsewhere have already documented the shift, cutesy AI “fruit slop” videos increasingly dipping into domestic violence, sexual assault and even murder plotlines, all wrapped in the same bright, Pixar-style packaging that made the original trend so shareable.2
Let me explain the terms first, because you must understand them. “AI slop” just means content churned out by generative AI tools, fast, cheap, and with no editorial oversight. A “microdrama” is one of those short, episodic videos, usually under a minute. Put them together and you get an entire soap opera cast of fruit characters, produced faster than any human writer could manage, with nobody checking what’s actually in the script.
Why this matters more than it looks like it should
It’s not really about fruit. It’s about how easily something disturbing can be smuggled past a young person’s guard when it’s wrapped in something familiar and non-threatening. I’ve spent years looking at how digital manipulation actually works, and the pattern is always the same, get the audience comfortable first, then move the line.
The research backs this up. A team at the University of Copenhagen has built a model showing how generative AI can accelerate the path to radicalisation, not just by recommending extreme content, as the old YouTube rabbit-hole did, but by creating new content that reflects a person’s anger or grievance back at them.3 Generative tools add a layer recommendation algorithms never had. They don’t just find the extreme content. They can manufacture it, endlessly, tailored to whatever keeps someone watching.
This isn’t theoretical. Researchers at GNET (the Global Network on Extremism and Technology) have already documented extremist groups using AI video generation tools to glorify real terrorist attacks, reusing exactly this kind of stylised, viral-friendly format.4 The fruit videos are a much softer version of the same technique. Familiar, funny, low-stakes content is the perfect cover for something that shouldn’t be low-stakes at all.
It sits within a regulatory gap. Under the Online Safety Act, Ofcom’s Protection of Children Codes require platforms likely to be accessed by children to protect them from “priority content” this includes violent material.5 Those duties came into force in July 2025. Fast-moving, AI-generated trends like this are exactly the kind of content those rules were meant to catch and exactly the kind that’s hardest to catch in time, because a video format can go from funny to violent within the same trend, on the same accounts, in a matter of weeks.
⚡Please don’t forget to react & restack if you appreciate my work. More engagement means more people might see it. ⚡
For Parents
1. Ask what they’re watching, calmly. If your child shows you a fruit video, don’t panic-react. Ask them to talk you through it. Curiosity opens a conversation. Alarm closes one.
2. Explain how these are made. A teenager who understands that a video is AI-generated, with no editor or fact-checker anywhere near it, is better placed to question what they’re seeing next time.
3. Use the controls you already have. Restricted Mode, screen time limits, and reporting tools exist on every major platform. They won’t catch everything, but they slow the drift.
4. Watch for desensitisation, not just content. The bigger risk isn’t one video. It’s a young person becoming comfortable with escalating violence because it kept arriving in a format that felt harmless.
For Teachers & DSLs
1. Add AI-generated content to safeguarding training. Most existing training assumes a human posted the harmful content. Increasingly, nobody did, and that changes how you talk about it with pupils.
2. Fold it into media literacy, not just online safety. Pupils need to understand that “AI-made” doesn’t mean “harmless” or “not real content.” That distinction is getting harder for adults to make, let alone children.
3. Log it against your Prevent duty where relevant. If a pupil’s viewing pattern shows escalation from this kind of content towards more extreme material, treat it as you would any other radicalisation indicator, and record it under KCSIE as usual.
If your child, or a child you work with, has seen something online that’s worrying them, Childline is there, day or night, and they don’t have to give their name.
0800 1111 | childline.org.uk
As always, thank you for your support. Please share this across your social media, and if you do have any comments, questions, or concerns, then feel free to reach out to me via the Social page, as I am always happy to spend some time helping to protect children online.
Remember that becoming a paid subscriber means supporting a charity very close to my heart and helping it do amazing things for people. Childline, I will donate 100% of paid subscriptions collected every six months, as I don’t do any of this for financial gain. Or to make a one-off donation, go to my Just Giving page.
If you or a child you know needs support:
Childline: 0800 1111 | childline.org.uk
Available 24/7, 365 days a year. Free, confidential, and here for every child.
Sky News, “Violent AI clips of fruit and veg characters could help radicalise young people, researchers claim,” news.sky.com, accessed 10 August 2026. Note: this piece reports researchers’ claims; the specific study behind the headline was not independently accessible from this environment at time of writing.
CBC News, “’It’s absurd; it’s brain rot’: Welcome to AI fruit slop, where fruits cheat on each other,” cbc.ca, accessed 10 August 2026.
University of Copenhagen, “Artificial intelligence may accelerate the path to radicalisation,” news.ku.dk, accessed 10 August 2026.
GNET (Global Network on Extremism and Technology), “An ‘Ode to Violence’: Extremist Exploitation of Viggle AI,” gnet-research.org, accessed 10 August 2026.
Ofcom, Protection of Children Codes of Practice under the Online Safety Act 2023, ofcom.org.uk, accessed 10 August 2026. Note: violent content is classed as “priority content” harmful to children; duties took effect 25 July 2025.