Despite YouTube's efforts, investigation reveals YouTube Kids is still awash with violent and age-inappropriate videos, with some discovered via recommendations
Context & Ripple Effects
This investigation lands three years after consumer groups escalated to the FTC with an updated complaint alleging adult content on YouTube Kids, and after YouTube responded with a reworked app addressing inappropriate content. The finding that violent videos persist — and are surfaced by recommendations rather than just slipping through uploads — suggests the curation fixes targeted supply, not the recommendation layer.
It also foreshadows a pattern WIRED UK itself would document a year later with YouTube monetizing sexualized videos of minors, and which researchers revisited in 2023 when they found YouTube recommending gun and school-shooting videos to young gamers. The through-line is that moderation policy and what the algorithm actually serves children keep diverging.
First-order effects
- YouTube faces renewed pressure on YouTube Kids specifically: the app marketed as a safe, curated space for under-fives is shown to be serving violent and age-inappropriate videos, undermining its core pitch to parents.
- Because some offending videos were reached via recommendations, YouTube cannot attribute the problem solely to bad uploads — its own suggestion engine is implicated, complicating the 'we removed it' defense used since the 2015 complaints.
Second-order effects
- Advertisers and brand-safety buyers get fresh evidence that kids-targeted inventory carries moderation risk, echoing the advertiser flight Epic Games' ad pause signaled during the 2019 monetization scandal.
- Regulators handed a concrete case file: the FTC complaint thread from 2015 gains a documented failure mode (recommendations) that strengthens arguments for treating kid-facing algorithmic feeds as a distinct enforcement problem.
Third-order effects
- If recommendation-driven exposure keeps recurring across years and product lines, the structural fix regulators gravitate toward is defaulting child accounts to human-curated or allowlisted catalogs rather than algorithmic feeds — a design constraint that would reshape how kids' video products are built.
- The pattern also points toward liability frameworks where platforms answer for what their recommenders surface, not just what they host, extending the accountability logic already visible in the FTC complaints and advertiser pullbacks.
The trend: Children's video platforms are being pushed from algorithmic recommendation toward curated, allowlisted delivery as repeated investigations show moderation policies failing at the recommendation layer.