SafeShare

A lot of parents and educators assume that if the video they picked for their kid is real, appropriate, and from a channel they trust, whatever plays next is roughly the same kind of thing. That assumption used to be reasonably safe. It isn’t anymore. A recent investigation found that after watching a popular preschool show like Cocomelon on YouTube, more than 40% of the recommended videos that followed contained AI-generated content, often bizarre, poorly made, and with no actual connection to the video that was chosen.

That’s not a video-selection problem. The chosen video can be exactly right. What YouTube serves up next, automatically, is a separate and increasingly unpredictable layer.

Quick Takeaways

Following a popular kids’ show on YouTube, more than 40% of recommended follow-up videos have been found to contain AI-generated content, per a recent investigation covered across multiple outlets (Digital Trends).
Fairplay for Kids, a children’s advocacy group, identified 278 channels dedicated to AI-generated content drawing an estimated 63 billion views a year, and found that top AI-slop channels targeting children have collectively earned over $4.25 million in annual revenue (Fairplay for Kids).
YouTube does require creators to disclose “meaningfully altered or synthetic” content, but for most videos that disclosure sits in the expanded description, not somewhere a parent or child would see it before or during playback (YouTube Help).
SafeShare removes the recommendation sidebar and autoplay queue entirely, which is where this specific problem concentrates.

How Much AI-Generated Content Is Actually Showing Up?

More than most parents would guess. One channel identified in recent reporting, posting under a name like “Jo Jo Funland Car Ride Song,” published more than 10,000 videos since August 2025, averaging around 50 new videos a day. For comparison, Sesame Street has published roughly 3,900 videos to YouTube across its entire 20 years on the platform. That production speed is only possible because the videos are AI-generated, not filmed or animated by a team.

A separate analysis of YouTube Shorts found approximately 21% of sampled Shorts contained AI-generated content. None of this requires a parent to have picked a bad video. It’s a description of what the recommendation and autoplay systems are increasingly likely to serve up next, regardless of how carefully the first video was chosen.

Doesn’t YouTube Already Label AI-Generated Content?

Partially, and not in a way that changes what shows up in front of a child. YouTube’s disclosure policy requires creators to label “realistic altered or synthetic content,” and failing to do so can trigger enforcement action. But for most videos, that label appears in the expanded description, a part of the page a parent has to click to open and a young child never sees. Only content on sensitive topics like health or news gets a more prominent, on-video label.

In practice, that means the labeling system exists, but it doesn’t function as a warning a parent glances at before a video autoplays. A 2023 investigation found AI-generated videos containing false science information being recommended to older kids as “educational,” despite YouTube’s policies. Some AI-generated content aimed at children has gone further than low-quality or inaccurate, including disturbing or violent material discovered by researchers investigating channels with innocuous-sounding names.

Why Does This Matter Specifically at the Moment a Video Ends?

Because that’s exactly when autoplay and the recommendations sidebar take over, and neither one is filtered by which video came before it. A parent who spent time finding a good, age-appropriate video has no control over what the platform decides to show next once it ends. The sidebar and autoplay queue are populated by engagement signals, not by any judgment about whether the next video is real, accurate, or appropriate, and the research above shows that a meaningful share of what fills that space right now is AI-generated content produced faster and cheaper than anyone is currently able to review.

Everything, but it’s worth being precise about what actually changes. SafeShare doesn’t scan a video for AI-generated content, and it doesn’t claim to. What it does is remove the recommendation sidebar and the autoplay queue entirely from the page your child watches. Since the AI-slop problem specifically concentrates in that layer, the video you chose plays, and when it ends, nothing else loads automatically, because there’s no sidebar or autoplay queue on the page for anything to load into.

Here’s the setup, under a minute:

Step 1. Copy the YouTube or Vimeo link for the video you’ve chosen.

Step 2. Paste it into safeshare.tv.

Step 3. SafeShare generates a clean viewing page: the video plays, and when it ends, the page stops. No sidebar, no autoplay, no next video.

Step 4. Optionally trim the video to a specific start and end point, and set a custom title so the link reads clearly for next time.

A free account at safeshare.tv keeps a link like this permanent, so you’re not recreating it every time the same video comes up again.

Frequently Asked Questions

Is this only a problem for very young kids?
The research cited here focuses on preschool and toddler content specifically, since that’s where the volume and the financial incentive are highest. Older kids aren’t immune to encountering AI-generated content in recommendations generally, but the concentrated, high-volume “slop” problem described in this piece is most documented in content aimed at the youngest viewers.

If YouTube already requires AI disclosure labels, why does this still happen?
Because the label mostly lives in the expanded video description, which a young viewer isn’t reading and a parent isn’t necessarily checking before every autoplay. The policy exists; it doesn’t function as something a child sees before a video plays.

Does trimming or a custom title help with this specifically?
They don’t address AI content directly, but they mean the link plays exactly the segment you intended and nothing else.

If autoplay deciding what your kid watches next feels like a bigger gamble than it used to be, a free account at safeshare.tv removes that layer entirely, no scanning, no detection, just nothing left to autoplay into.