The most-watched videos on my niece’s tablet last month starred a sentient watermelon working through a love triangle. Three weeks later the same feed served up a strawberry being mutilated by a character coded as a cartel enforcer.
That drift, from harmless brainrot to something uglier, is what a new analysis from the Global Network on Extremism and Technology says is happening at scale.
The gore is the part nobody’s watching
GNET‘s researchers argue the AI-generated fruit videos flooding social feeds work as a delivery mechanism for covertly spreading “violent and extremist content.”
Most of the reels look like nothing. Colorful fruit characters, sometimes pornographic ones, moving through dramatic setups: love triangles, job loss, brotherhood. Standard algorithmic filler.
But the analysts tracked a rising number of these reels sliding into what they call “AI fruit/vegetable gore” — extreme scenes of torture, mutilation and other cruel acts of violence.
Cartels and far-right coding, in fruit form
Here’s the detail that separates this from garden-variety shock content. The violence isn’t generic.
Many of these reels depict violent characters as belonging to specific groups or ideologies, the authors observe: criminal cartels, far-right groups, extreme nihilism. The ideology is baked into the character design.
“Violence-based AI slops reinforce a fascination with violence for violence’s sake, thereby allowing the so-called online ‘ode to violence’ to progressively gain ground among young people within the digital space,” the analysts write.
They go further, arguing this material can “lay the basis for the consumption of further violence-based audiovisual material, making young users fall into the trap of online radicalization and audiovisual violence consumption.”
Read the report with a hand on your wallet
Now the skepticism, because this report earns some.
Whether AI is actually driving kids toward violence remains to be seen in academic studies. GNET’s report is largely a surface-level analysis. It also inadvertently references at least one AI-generated editorial, which is a rough look for a document about the dangers of synthetic media.
And the provenance matters. GNET is the research arm of the Global Internet Forum to Counter Terrorism, a group funded and run by representatives of the tech industry, and one that’s been criticized for its role in advancing digital censorship. An industry-funded body warning about industry-made content deserves a second read.
The radicalization angle is the smaller problem
Set the extremism question aside for a second, because it’s beside the point.
This content is being churned out by bot farms at an industrial rate. All of us get dragged into it, and kids get dragged furthest, since their media literacy is still developing.
What you end up with is a recursive nightmare of algorithmically-driven slop. It can be made to glorify violence just as easily as electoral fraud, racism, or any other social ill you care to name.
That’s the actual finding. Not that fruit gore turns children into extremists, but that a machine capable of generating infinite fruit gore on command is equally capable of generating infinite anything else, aimed at whoever the algorithm decides is watching. The depravity on display is interchangeable. The limitless potential behind it is what makes gen-AI slop dangerous, full stop.
If you want a practical takeaway, it isn’t a parental control setting. It’s that the feed itself has no floor, and nothing in the current system is building one.
