A look at how YouTube is being used to share traumatizing content to young children automatically and at scale
even if you're a human, you have to end up impersonating the machine.” http://twitter.com/... Casey Newton / @caseynewton : Discussions about this (excellent) post have taken over our newsroom today —> http://twitter.com/... Martin Bryant / @martinsfp : The only decent solution is not letting young kids use YouTube unsupervised, which is sensible anyway. http://twitter.com/... Sarah Perez / @sarahintampa : the only solution I've found is to be pretty on top of my child's YouTube viewing. I have a few rules: http://twitter.com/... @nytimes : Youtube contains dark corners as videos that are disturbing for children slip past its filters http://www.nytimes.com/... @mikeisaac : bots slice and dice infinite combinations of YouTube videos into a fine content slurry, feeding it to yr kids until you take the ipad away http://twitter.com/... Dylan Connerton / @pragueham : The system is complicit in the abuse. http://medium.com/... A great companion to @zeynep's amazing ted talk. By @jamesbridle Kashmir Hill / @kashhill : one solution to this is for YouTube not to pop up “related” videos on child videos or to auto-play the next one http://medium.com/... James Bridle / @jamesbridle : I wrote this essay in a hurry, today, mostly to get it out of my head, and, although I didn't realise it when I started, it will be part of a forthcoming publication, so thoughts and comments are appreciated. Shuja Haider / @shujaxhaider : this is downright lovecraftian, down the author's repeated warnings about the trauma you are about to experience in the next paragraph https://medium.com/... Jacob Garbe / @logodaedalus : Damn, a *really* well-written article on a disturbing angle of automated content creation that's harmful *right now* https://medium.com/... Ian Bogost / @ibogost : There's a reason everyone is sharing this. https://medium.com/... Max Read / @max_read : youtube is probably the most undercovered and overlooked Bad Thing on the internet https://medium.com/... Josh Fruhlinger / @jfruh : man if you were creeped out by the NYT article on kids youtube vids the other day, brace yourself for this https://medium.com/... Katie Drummond / @katiedrumm : I'm taking my daughter and moving into an underground bunker disconnected from everyone and everything https://medium.com/... Allegra Rosenberg / @llegrarosenberg : “This is a deeply dark time, in which the structures we have built to sustain ourselves are being used against us.” https://medium.com/... Tom Gara / @tomgara : The nightmarish stuff going on in the worst corners of YouTube is beyond comprehension. This is so disturbing: https://medium.com/... Katie / @katienotopoulos : I truly believe YouTube will have destroyed an entire generation of young people https://medium.com/... John Arnst / @arnstjohn : When weird, human/algorithm-generated videos targeted at very young kids become ultimately vehicles for trauma https://medium.com/... pic.twitter.com/TChI2PhTJD @mathowie : YouTube's AI is like a game of telephone where things get more violent and inappropriate as they suggest new vids https://medium.com/... Robin Sloan / @robinsloan : Started reading this on the walk to the train station & it plunked me right down on a bench. Engrossing & troubling. https://medium.com/... Mary Hamilton / @newsmary : This is one heck of a rabbit hole. YouTube Kids, algorithmic content generation and a whole bag of grim weirdness. https://medium.com/... Lydia Polgreen / @lpolgreen : Get your kids off YouTube. Now. https://medium.com/... Erik Hinton / @erikhinton : welp, it's time to throw my computer out a window and forget about the internet forever https://medium.com/... pic.twitter.com/wOUzCvl6qY Robin Sloan / @robinsloan : This feels like such a telling example because no actual viewer “wants” these videos; the algorithm calls them forth https://medium.com/... @mikeisaac : also bumping google's three days of distancing itself from the “social network” label which worked well to deflect http://www.nytimes.com/... See also Mediagazer
Context & Ripple Effects
James Bridle's investigation documents how YouTube's recommendation engine and auto-play chain children from ordinary cartoons into violent, surreal, and traumatizing videos — content often mass-produced to game the same algorithm that surfaces it. The piece lands in a platform already under scrutiny for what its recommender amplifies: coverage of YouTube as a radicalizing tool and former extremists describing how the algorithm surfaced extremist videos to them as teenagers point to the same structural problem in a different audience.
What makes this episode distinct is the victim: pre-verbal or young children who cannot curate their own viewing, on a product (later the subject of an FTC settlement over child privacy) where regulators would eventually force design changes like removing comments and notifications. Bridle's framing — that humans end up impersonating the machine to feed it — anticipates the kidfluencer economy Bloomberg later flagged, where unmarked sponsored marketing targets children barred from even holding accounts.
First-order effects
- Parents and caregivers lose the assumption that YouTube Kids and auto-play are safe defaults; supervision shifts from optional to mandatory, as tech journalists like Sarah Perez describe doing with rules and constant oversight of their own children's viewing.
Second-order effects
- Advertisers and regulators gain leverage: the same recommendation system implicated here is the one later tied to radicalization and misinformation enforcement problems, pushing YouTube toward punitive policy changes against its own stars and revenue-depressing child-privacy conditions.
Third-order effects
- If the pattern holds, children's video becomes a regulated category rather than a self-policing one — with algorithmic recommendation, synthetic mass-produced content, and unmarked commercial targeting treated as systemic risks requiring design-level constraints rather than takedowns.
The trend: Algorithmic recommendation is being forced to account for its most vulnerable users, as child-facing platforms move from engagement optimization toward externally imposed safety constraints.