GNET says AI fruit videos can mask extremism and harm children

A research analysis from the Global Network on Extremism and Technology warned that mass-produced AI "slop" videos, especially animated fruit-and-vegetable clips popular on children’s feeds, could be used to hide violent and extremist content. The group said the material is cheap to make, highly repetitive and designed to maximise clicks, views and ad revenue. The warning said the format can also feed unhealthy viewing habits. GNET argued that AI slop could contribute to cognitive decline, or "brain rot", which it defined as the cultural and psychological effects of consuming too much fast, overstimulating and intentionally meaningless media, including endless scrolling and the pursuit of instant dopamine hits. The analysis said one creator could potentially produce hundreds of short-form AI slop videos, flooding multiple platforms with massive amounts of content. The group said social media systems reward constant activity and uploads, making it possible to reach millions of children worldwide. It also said cheap, easy-to-learn tools such as PixVerse, VeoE.ai, Media.io and Sora have helped drive a sharp rise in AI-generated content. The analysis said some of the most striking examples begin as harmless or absurd clips and then move into graphic violence. It described animated fruit videos with storylines involving romance, domestic conflict, betrayal and cheating, as well as scenes of torture, mutilation, assassinations, mass-casualty events and brutal murders. One recurring character, Carrot Khalifah, is shown telling an army that "the Carrot State will rise again", a line the study said echoes Islamic State propaganda. GNET said the videos can use a cute, Pixar-like aesthetic to mask extremist messaging linked to Islamic State, extreme-right accelerationism and nihilistic violence extremism. It said the material is easily accessible, can trap young users in endless scrolling and may encourage the gradual normalisation of cruelty and emotional numbness. The authors said TikTok and similar platforms have moderation policies against extreme violence, but those safeguards are not fool-proof because AI generators can hide disturbing material behind apparently innocent content. They called on TikTok to develop countermeasures to limit the spread of such clips. GNET is the academic research arm of the Global Internet Forum to Counter Terrorism. The George Washington University Program on Extremism joined the network, which formally launched in June 2020, and the group says its work is meant to produce policy-relevant research that helps guard against real-world harms.