Open a brand new YouTube account today, and roughly one in five of the first videos it pushes at you will have been made by AI for one reason: to farm views. That is not a guess. A study by video-editing firm Kapwing tried exactly that, and 104 of the first 500 Shorts were AI slop.
Now picture that same feed on a tablet in front of a four-year-old. Parents are asking about it, and the platforms’ answers are thin. In April 2026, more than 200 child-development groups and experts signed a letter asking YouTube to act.
I went through the studies, the investigations and the expert letters myself for this guide. The goal is to separate what is documented from what is only being shouted online. Where a claim is bigger than the evidence, I say so.
This article summarizes published reports and expert warnings for general information. It is not medical or psychological advice. If you are worried about your child’s development, talk to a pediatrician.
Table of Contents
- What “Brain Rot” Means and Why Oxford Picked It
- Fact 1: AI Slop Channels Earn an Estimated $117 Million a Year
- Fact 2: 1 in 5 Videos Shown to New YouTube Users Is AI Slop
- Fact 3: One Third of a New Feed Is “Brain Rot”
- Fact 4: The New York Times Found 40% AI Visuals After One CoComelon Video
- Fact 5: 200+ Groups Told YouTube and Google to Act
- Fact 6: What Science Does and Does Not Show About Young Brains
- Fact 7: Violent “Fruit Slop” and the GNET Warning
- Fact 8: Fake “Educational” Channels Aimed at Toddlers
- Fact 9: What YouTube Says It Is Doing
- 7 Warning Signs Parents Can Watch For
- A Parent’s Action Plan for 2026
- What Platforms and Regulators Should Do
- Conclusion
- Frequently Asked Questions (FAQ)
What “Brain Rot” Means and Why Oxford Picked It?
Oxford’s dictionary team named “brain rot” its Word of the Year for 2024, after usage of the phrase climbed 230% between 2023 and 2024. More than 37,000 people voted in the public round.
Oxford defines it as the supposed decline of a person’s mental state from overconsuming trivial or unchallenging online content. The word “supposed” matters. Brain rot is a cultural label, not a medical diagnosis.
The phrase is also older than the internet. It was first used by Henry David Thoreau in his 1854 book Walden, long before anyone scrolled a feed.
When researchers use the term today, they usually mean fast, fragmented, overstimulating video that offers almost nothing to learn. AI has made that kind of video nearly free to produce. That cost drop is the heart of this story.
Related read: Not sure what counts as slop? Our guide on the signs you are looking at AI slop shows the patterns to watch for.
Fact 1: AI Slop Channels Earn an Estimated $117 Million a Year
Kapwing looked at the top 100 trending YouTube channels in every country, 15,000 channels in total. It found 278 channels that posted nothing but AI slop.
Together those channels had collected more than 63 billion views and 221 million subscribers. Kapwing estimates they earn about $117 million a year. The company also reports that South Korea’s trending slop channels alone passed 8.45 billion views, with Pakistan and the United States next.
Spain led on subscribers, with more than 20 million people following these channels. That is close to half the country’s population.
Two cautions are fair here. First, $117 million is an estimate built from public view counts, not audited payouts. Second, these 278 channels are only the ones that made the trending lists. The real market is certainly larger.
The business logic is simple. A video made in minutes costs almost nothing, so even a modest ad payout is profit. One person can run dozens of channels and upload all day.
Related read: For the full money trail, see how the AI slop economy actually makes its cash.
Fact 2: 1 in 5 Videos Shown to New YouTube Users Is AI Slop
To test what a brand new viewer sees, Kapwing created a fresh YouTube account with no watch history. That matters because there was no personalization to blame.
Of the first 500 Shorts the feed served, 104 were pure AI slop. That works out to 21%.
The finding landed at an awkward time for the platform. In his annual letter in January 2026, YouTube CEO Neal Mohan used the phrase “AI slop” himself and promised to build on the company’s spam and clickbait systems.
YouTube also makes a fair point of its own. A YouTube editorial lead has said that using AI is not banned, and that creators who use AI tools can still earn money. The platform’s target is low-quality, mass-produced content, not AI as a tool.
For parents, the takeaway is practical. A shared family account or a child’s profile is not a clean slate. The same recommendation engine is running underneath.
Related read: More numbers are in our breakdown of what the data says about YouTube Shorts and AI slop.
Fact 3: One Third of a New Feed Is “Brain Rot”
The same test produced a second number. Of those 500 Shorts, 165 were classed as brainrot, or 33%.
It is easy to misread this figure. Brainrot is a broader bucket than AI slop. It includes AI slop plus other low-effort content built to hold attention, such as repetitive loops, meaningless challenges and absurd characters.
Here is how the numbers line up:
| What was measured | Result | Who found it |
|---|---|---|
| Channels that were AI slop only | 278 of 15,000 | Kapwing |
| Combined views of those channels | 63 billion+ | Kapwing |
| Estimated yearly revenue | About $117 million | Kapwing (estimate) |
| AI slop in first 500 Shorts | 104 (21%) | Kapwing |
| Brainrot in first 500 Shorts | 165 (33%) | Kapwing |
One honest limit applies. Deciding what counts as slop or brainrot is a judgment call by Kapwing’s researchers. It is a useful snapshot, not a scientific measure of harm.
So what does this mean for a child? The numbers show how much of the feed is built to grab attention. They do not, on their own, prove what that does to a developing brain. That second question is where the real expert debate sits, and it is what we take up next.
Fact 4: The New York Times Found 40% AI Visuals After One CoComelon Video
In February 2026, The New York Times published an investigation that changed how many parents saw YouTube. Reporters reviewed more than 1,000 videos recommended to young children. They found the algorithm pushing odd, often nonsensical AI-made clips from channels that claimed to teach toddlers and preschoolers the alphabet and animals.
The test that grabbed headlines was simple. A reporter watched a legitimate CoComelon video, “Wheels on the Bus,” then scrolled the Shorts that YouTube recommended in a 15-minute session. More than 40% of what appeared next showed signs of AI-generated visuals. The team used an AI detector to confirm its judgment, and in some cases the music and voices were synthetic too.
The reporters described what the clips looked like. Faces were warped. Animals and people had extra body parts. On-screen text was garbled. Many stories had no logical flow, and some carried plain misinformation.
One caveat belongs here, because it is easy to get wrong. The 40% figure comes from one session that began with a very popular preschool video. It is not the share of all YouTube Shorts, and it is not the average for every child. It shows how fast a feed can tilt, not the typical feed.
The Times also sent YouTube three examples of hyperrealistic AI videos found on YouTube Kids. YouTube removed them. That tells you the system can catch problems once someone points at them. The question parents are asking is why they needed pointing out.
There is also a rules gap. YouTube asks creators to disclose AI use in realistic content. Cartoon-style videos for children fall outside that rule, so a large amount of AI kids’ content carries no label at all.
Related read: Our report on the disturbing truths about AI kids’ videos walks through real examples parents have flagged.
Fact 5: 200+ Groups Told YouTube and Google to Act
On April 1, 2026, the children’s advocacy group Fairplay released an open letter addressed to YouTube CEO Neal Mohan and Google CEO Sundar Pichai. The press release says more than 200 experts and organizations signed it, including social psychologist Jonathan Haidt.
The letter asked for six changes:
- Clearly label all AI-generated content on YouTube.
- Bar AI-generated content from YouTube Kids.
- Prohibit AI-generated “Made for Kids” videos on the main site.
- Stop recommending AI-generated content to anyone under 18.
- Give parents a switch that turns AI content off by default.
- Halt investment in AI-generated videos for children.
That last point is aimed at a specific event. Fortune’s coverage of the campaign notes it followed Google’s $1 million investment in Animaj, an AI animation studio that makes children’s videos. Critics saw a mismatch between funding the genre and being asked to restrain it.
Rachel Franz, who runs Fairplay’s Young Children Thrive Offline program, wrote that AI slop hypnotizes young children and makes it hard for them to move on to play, sleep and time with other people. She also argued the recommendation system makes it nearly impossible for kids to avoid this content. Those are Fairplay’s words and Fairplay’s position, not a finding from a clinical study. They are worth reading with that in mind.
YouTube’s answer was measured. A Google spokesperson said YouTube Kids has high standards and limits AI content in the app to a small set of high-quality channels. The company said it requires creators to disclose realistic AI content, and that it is working on AI labels for YouTube Kids. No launch date was given.
Both sides have a case. Parents see a flood. The platform says the youngest viewers are in a more protected space. The truth sits in the gap between the main app, where most children actually watch, and the curated one.
Related read: For the full story on the letter and what followed, read our post on the 200+ groups pushing YouTube on kids’ AI videos.
Fact 6: What Science Does and Does Not Show About Young Brains
This is the part most viral posts get wrong, so it deserves care.
Many headlines say AI slop causes cognitive decline in children. That claim traces back to a report from the Global Network on Extremism and Technology, or GNET. Euronews summarized it with a headline that linked AI slop to cognitive decline and violent tendencies in kids.
A closer look at the report is more modest. The Next Web’s analysis points out that the document was written by three researchers at ITSTIME, an Italian center that studies terrorism and online security. The report itself contains no cognitive testing, no sample of children and no control group. “Brain rot” in the report is a cultural phrase, not a diagnosis.
So where does that leave parents? Here is what I can say fairly:
- What is documented: large amounts of low-quality AI video are being recommended to young children, as the Times and Kapwing both found.
- What experts worry about: the Times reported that specialists warn these videos can contain conflicting information and lack a clear plot, and can be cognitively overwhelming for very young viewers.
- What is not yet proven: I found no long-term study that measures the effect of AI-made kids’ videos on brain development. The harm is a reasonable concern, not an established fact.
- What is the precautionary argument: Fairplay’s letter points to the absence of evidence that these videos are safe. That is a call for caution, not proof of damage.
The most defensible worry may be the simplest one. Time spent on endless autoplay is time not spent on play, sleep and talking with people. That concern does not depend on any one dramatic claim about neurons.
The American Academy of Pediatrics has also advised parents to steer away from AI-generated and sensationalized content, according to reporting on the Times investigation. For current age-by-age screen-time guidance, check the AAP’s own pages rather than relying on numbers passed around on social media.
If a post tells you AI videos are “proven to destroy” a child’s brain, it has gone further than the evidence. If a post tells you there is nothing to worry about, it has ignored what experienced reviewers saw in those feeds. The honest answer is in the middle: real exposure, real expert concern, and unfinished science.
Fact 7: Violent “Fruit Slop” and the GNET Warning
Not all slop is harmless nonsense. In August 2026, the Global Network on Extremism and Technology, known as GNET, published a report on a strange corner of AI video: colorful fruit and vegetable characters acting out dramas. Some of those dramas turn violent.
GNET is a research initiative backed by the Global Internet Forum to Counter Terrorism. Its researchers described a subgenre they call AI fruit and vegetable gore. Cartoon-like produce characters are put through graphic harm, and the videos can look innocent in a quick scroll. According to Cybernews’ summary of the study, the researchers say that is how some of it slips past TikTok’s rules on violent content.
The report flagged a few examples that echoed real extremist imagery. One clip appeared to mimic the style of terrorist propaganda, and another used a character name that alluded to a known extremist slogan. I am deliberately not repeating the details, because describing them in depth adds nothing for parents.
The concern GNET raises is about normalization. Familiar cartoon characters can make harsh violence feel funny or cool, which may lower a young viewer’s guard. Dexerto’s report describes the clips as attention traps dressed up in kid-friendly animation.
There is a limit on this claim, and it matters. The Cybernews piece itself notes that the idea of AI videos pushing young users toward violent extremism has not yet been proven by academic studies. GNET’s report is largely a surface analysis. It documents that the content exists and reaches kids. It does not show how many children are affected or what happens to them afterward.
Treat it as an early warning, not a verdict. The practical point for parents is easy to act on. If a video features cartoon food, animals or characters, the cartoon style tells you nothing about whether the content is gentle.
Related read: The fruit-video trend has a bigger backstory. See what happened when the viral Fruit Love Island series got cancelled.
Fact 8: Fake “Educational” Channels Aimed at Toddlers
One of the most worrying patterns in the reporting is how these channels present themselves. Many do not call themselves entertainment. They call themselves learning.
The Times found channels that market themselves to parents as teaching toddlers and preschoolers the alphabet and animals. The videos then show things like a rhino turning into a dump truck to represent a letter. A parent glancing at the title would assume it was a normal early-learning video.
Bloomberg Businessweek reported on why this keeps happening. One creator, who runs a channel about making money online, showed her audience a simple method. She asked ChatGPT for repetitive children’s song lyrics full of nonsense words, fed them into an AI video generator, and got a colorful animated reel within minutes. She said the approach could earn hundreds of dollars a day and that the creator only does about 5% of the work.
That is the whole economic engine in one example. Cheap production meets a young audience that never skips, never complains and replays the same clip.
The accuracy problem is also real. Beyond the surreal visuals, some videos have given unsafe guidance. An op-ed by one of the signers of the Fairplay letter describes reporting from Mother Jones that found clips showing toddlers swallowing whole grapes, which is a choking risk, infants eating honey, which carries a botulism risk, and children riding unrestrained in the front seat of a moving car.
This is a stronger and more specific worry than “brain rot.” A video does not need to be scientifically proven to rewire anything to be a problem. It only needs to show a toddler something unsafe as if it were normal.
Here is a quick way to size up an unfamiliar kids’ channel:
- Upload volume: dozens of videos a day is not something a small human team does.
- Visual consistency: characters that change shape, color or voice mid-video are a red flag.
- Story logic: a lesson that does not connect to the letter, number or animal it claims to teach.
- Channel identity: no named creator, no website, no about page.
- Safety content: anything showing food, roads or water should match real safety rules.
No single sign proves a video is AI made. Several together are worth a closer look.
Fact 9: What YouTube Says It Is Doing
It would be unfair to tell only one side. YouTube has said a lot in 2026, and some of it is concrete.
The company’s CEO named managing AI slop a priority for the year in his January letter. The platform removed the three hyper-realistic examples the Times sent it from YouTube Kids. And according to coverage of the Times investigation, YouTube announced new parent controls that let families set time limits on Shorts.
After the Fairplay letter, a Google spokesperson told reporters that YouTube Kids has high standards and limits AI content in the app to a small set of high-quality channels. He added that YouTube labels content from its own AI tools and requires creators to disclose realistic AI content. Tubefilter reported that YouTube is working on AI labels for YouTube Kids, though no schedule was given.
Fairplay’s reply to those steps was that they do not go far enough. Franz has argued that the platform tends to act on channels after press coverage rather than fixing the root cause. Another expert wrote that putting the burden on parents to block channels one by one protects only the children of parents with the time and digital skills to do it.
So who is right? Probably both, in part. YouTube is moving, and the labeling and removal efforts are real. But the feed is huge, the tools are opt-in, and the cartoon-style rule gap means a lot of AI kids’ content still arrives unlabeled.
For a family, the lesson is not to wait. Whatever the platform ships next year, the choices you make this week about what your child watches matter more.
Related read: To see how YouTube has been cracking down more broadly, read our breakdown of YouTube’s AI slop purge.
7 Warning Signs Parents Can Watch For
Before the action plan, a quick note. None of the signs below is a diagnosis. Children get cranky, fixated and tired for countless reasons. Think of this list as a prompt to look more closely at what your child is watching, not as a way to judge their development.
- They repeat nonsense phrases or character names over and over, and can list far more of them than anything from a book or show you chose.
- They get very upset when a video is stopped. Some frustration is normal at any age. A meltdown every single time is worth noticing.
- They struggle to switch to other activities like play, meals or bedtime once a screen is on. This echoes Fairplay’s concern that the videos make it hard for young children to move on to essential activities.
- They copy strange sounds, voices or movements from videos during everyday tasks.
- They seem confused about what is real, for example asking whether animals can really do what a cartoon animal just did.
- They show a new interest in rough or violent play that looks borrowed from a clip.
- They lose interest in books, conversation or toys that used to hold their attention.
If several of these show up together, the first step is to watch their feed with them for a few days. If you are still worried, a pediatrician is the right person to talk to.
A Parent’s Action Plan for 2026
You cannot fix YouTube’s recommendation system from your living room. You can change what reaches your child. These steps are ordered from quickest to most involved.
1. Watch with them for a week.
Sit beside your child and note what the feed actually serves. It is the fastest way to see whether the patterns in this article show up in your home. A colorful thumbnail and a cheerful title tell you nothing about safety.
2. Turn autoplay off.
Autoplay removes the natural stopping point after each video and is a major reason a quick clip turns into an hour. YouTube’s parental controls help pages explain how to adjust these settings and set limits.
3. Curate instead of searching.
Make a short list of channels and videos you have checked yourself, and let your child pick from that list. Platforms with human-made children’s programming are a safer base. PBS Kids and Khan Academy Kids are two well-known options built by named educators and producers.
4. Check reviews before adding a new show.
Independent reviewers can tell you whether something is age-appropriate. Common Sense Media rates children’s content and explains why.
5. Use a monitoring tool if it helps your family.
Software like Qustodio can track screen time and filter content. It is a supplement. Experts quoted in this story have been blunt that parental controls alone cannot solve a feed-wide problem.
6. Teach simple media questions.
Even young children can learn to ask, “Does that look real?” and “Could that really happen?” Tell them in plain words that computers can make videos that are not real. This is the skill that lasts, long after any one app changes.
7. Model the habit.
Children copy what they see. If a phone is always in your hand at dinner, a rule about their tablet will not land.
8. Replace, do not only remove.
Taking a screen away leaves a gap. Fill it with something concrete: a book, a walk, a craft, a game that needs two people. The goal is not zero screens, it is better screens and more of everything else.
9. Use the platform’s own feedback tools.
Reporting a video and using the “Not interested” option gives the system a signal. It is not a complete fix, but it costs ten seconds.
For a wider look at risks across platforms, see our guide to the hidden dangers of AI on social media.
What Platforms and Regulators Should Do?
Parents cannot carry this alone, and the expert groups in this story agree. Based on the Fairplay letter and the reporting above, the main asks are clear:
- Label all AI-generated content, including animated and cartoon-style videos that currently fall outside YouTube’s disclosure rule.
- Keep AI-generated content out of the youngest viewers’ recommendations.
- Give parents a simple switch to turn AI content off by default instead of making them hunt through settings.
- Stop investing in AI-made videos aimed at babies and toddlers.
- Add more human review. As several commentators point out, AI is poorly placed to police AI by itself.
It is only fair to repeat that platforms are responding. Labels for YouTube Kids are in development, parent time limits on Shorts have been added, and the CEO has made this a stated priority. The open question is speed and whether those tools will be on by default.
Conclusion
Here is what the evidence supports. AI-made slop is a real and profitable industry, with an estimated $117 million a year flowing to the top slop channels alone. A fresh YouTube feed showed about one in five slop videos and one in three brainrot videos in Kapwing’s test. The New York Times found that one preschool video could lead to a Shorts feed that was more than 40% AI-looking. And more than 200 experts and organizations asked YouTube to change how it treats children.
Here is what the evidence does not yet support. There is no long-term study showing that these videos damage brain development, and the loudest headlines about cognitive decline rest on a report with no children in its sample. The strongest, safest conclusion is a precautionary one: these videos are low quality, sometimes unsafe, and built to hold attention, so they deserve a real gatekeeper.
That gatekeeper is you. Watch what they watch, choose a few trusted sources, keep autoplay off, and fill the freed hours with real things. You do not need to panic. You do need to pay attention.
This article is for general information and does not replace advice from a pediatrician or child development professional. Statistics come from the cited studies and news investigations and may change as new research appears.
Frequently Asked Questions (FAQ)
What is AI brain rot content?
It is a popular label for fast, repetitive, overstimulating video that offers little to learn, often made with AI tools. Oxford chose “brain rot” as its 2024 Word of the Year to describe the supposed decline in a person’s mental state from too much trivial online content. It is a cultural term, not a medical diagnosis.
How much of YouTube is AI slop?
In Kapwing’s test of a brand new account, 104 of the first 500 Shorts were AI slop, which is 21%, and 165 were brainrot, which is 33%. That is a snapshot from one test account, not a figure for all of YouTube.
Does AI slop damage children’s brains?
Nobody has proven that. The Times reported that specialists worry about confusing, overwhelming content for very young viewers, and Fairplay points out there is no evidence these videos are safe. But I found no long-term study measuring brain development effects. The responsible reading is concern and caution, not a confirmed finding.
Is YouTube Kids safe from AI slop?
Not entirely. The Times sent YouTube three hyperrealistic AI videos found on YouTube Kids, and YouTube removed them. Google says it limits AI content in the app to a small set of high-quality channels, but critics say the safeguards are not yet enough.
How can I tell if a kids’ video is AI-generated?
Look for characters that change shape or voice, garbled text, impossible physics, a story that does not connect to the lesson, and channels that upload dozens of videos a day with no named creator. No single sign is proof. Several together are a reason to look closer.
How much screen time is right for my child?
It depends on age, and guidance is updated over time. Check the American Academy of Pediatrics’ current recommendations for your child’s age rather than relying on numbers shared on social media, and talk to your pediatrician about your own situation.
What should I do if my child saw something disturbing?
Stay calm, ask them what they saw and how they felt, and listen first. Then report the video to the platform, block the channel, and review the feed together. If your child is distressed or the behavior changes noticeably, contact their pediatrician.
Will YouTube label AI videos for kids?
YouTube has said it is working on AI labels for YouTube Kids but has not given a launch date. Today, creators must disclose realistic AI content, while cartoon-style videos for children are outside that rule.
Keep reading:
How Facebook’s AI slop problem spiraled
Spotify’s AI music purge explained
The AI slop books climbing Amazon’s bestseller lists