Suicide tips and other dark and dangerous content was found spliced in kids' cartoon videos on YouTube and YouTube Kids
Free Hess, a pediatrician and mother, had learned about the chilling videos over the summer when another mom spotted one on YouTube Kids.
Context & Ripple Effects
This is the third act of a problem YouTube has been chasing since late 2017, when [[a:923842|James Bridle documented how traumatizing clips were being served to young children automatically and at scale]]. YouTube answered with a stated crackdown — ramped-up removals and halted monetization for exploitative channels (its November 2017 policy detail) — but a WIRED UK investigation months later found YouTube Kids still awash with violent, age-inappropriate videos, some surfaced by recommendations themselves.
What is new here is the method and the messenger: Free Hess, a pediatrician, surfaced cartoons with suicide guidance spliced into otherwise innocent footage — content engineered to slip past both moderation and parents — on the main service and the Kids app alike.
First-order effects
- YouTube must remove the flagged videos and re-run its 2017 playbook — mass takedowns plus demonetization — against channels whose entire model is hiding harmful frames inside children's content.
- Parents and pediatricians like Hess get fresh evidence that the Kids app's curation cannot be trusted unsupervised, pressuring YouTube to tighten what its recommendation system surfaces to child accounts.
Second-order effects
- Every new exposé compounds the credibility damage from prior findings — conspiracy theories suggested in YouTube Kids search results and an algorithm later shown curating home movies into content sexualizing children — making advertisers and app-store gatekeepers harder to reassure.
- The splicing technique forces moderation investment toward frame-level and audio-level detection rather than metadata or title screening, raising the cost floor for any platform hosting user-generated kids' video.
Third-order effects
- The cycle — exposé, crackdown, recurrence — points toward algorithmic curation of children's media becoming a standing regulatory liability rather than a self-policing product choice, a trajectory the corpus extends all the way to a 2026 analysis finding YouTube Shorts still pushing bizarre, undisclosed AI-generated videos at kids.
The trend: Child-facing recommendation systems are stuck in a recurring failure loop where each moderation crackdown is followed by new evidence that harmful content reaches kids through the same pipes.