Open TikTok for the first time and you might think the algorithm already knows you. It doesn’t. What it does know is that TikTok AI slop shown to new users fills the gap before any real personalization kicks in. According to new research from Kapwing, a video editing company that analyzed over 10,000 TikTok videos across 20 categories, nearly 60% of the first videos a brand-new account sees on the For You page are low-quality, AI-generated content. That’s not a fringe finding buried in edge cases. That’s TikTok’s default first impression.
Researchers set up a fresh account and tracked the first 500 videos served. Of those, 294 were classified as AI slop, defined as content with obvious AI-generated visuals or low-quality compilations built on synthetic scripts and voiceovers. The math is blunt. A new user is statistically more likely to see machine-made junk than anything a human actually created, and that happens before the algorithm has collected a single meaningful signal about who that person is.
Three Times Worse Than YouTube
The comparison with YouTube makes the gap hard to ignore. Kapwing ran the same fresh-account test on YouTube Shorts and found that 104 of the first 500 videos, or about 21%, qualified as AI slop. On TikTok, that number was 294. That’s roughly three times the rate, and it puts TikTok in a different category altogether when it comes to synthetic content saturation. By November 2025, TikTok had already labeled more than 1.3 billion videos as AI-generated, which gives some sense of the sheer volume of synthetic material circulating on the app.
TikTok has introduced controls that let users adjust how much AI-generated content appears in their feeds, and the company has put $2 million into an AI literacy fund. Those moves signal awareness of the problem. But Kapwing’s data suggests the measures haven’t meaningfully changed what a new account actually sees by default. The feed still leans heavily synthetic before any user preference is established.
Kids Are Seeing the Most of It
The children’s content numbers are where the report gets genuinely alarming. Of the 2,000 videos Kapwing reviewed in TikTok’s Kids category, 57.4% were AI slop, the highest rate of any category in the study. The hashtag #cartoonkids was almost entirely machine-made 97 of the 100 videos analyzed were AI-generated. Tags like #cartoons and #babysong both came in at 83%, and #forkids sat at 79%.
The pattern these videos follow is recognizable to anyone who has stumbled across them. Familiar cartoon characters appear in strange scenarios. Educational lessons contain factual errors. Characters speak with synthetic voices that don’t quite land. A counting lesson that gets the numbers wrong is easy for an adult to dismiss, but a preschooler has no frame of reference to catch the mistake. Dr. Dana Suskind, a professor of pediatrics at the University of Chicago, described the situation to Mother Jones as “toddler AI misinformation at an industrial scale.” The concern isn’t just content quality. It’s that children are being exposed to confidently wrong information at a pace no human creator could ever match, and TikTok’s recommendation engine keeps distributing it.
There’s also a feedback loop worth noting. Once an account shows any interest in AI-generated content, the algorithm doubles down and serves even more of it. That makes the initial flood of slop more than just a first-impression problem. It can shape the entire trajectory of a user’s feed, especially for younger viewers who may not know to push back against it.
Where Accuracy Matters Most, Slop Clusters Most
Beyond children’s content, the categories hit hardest by AI slop are precisely the ones where getting things right matters. Science and Education came in at 35%, Health at 33.8%, and History at 33.5%. The #healthtips hashtag alone registered 74% AI slop. At the other end of the spectrum, categories that require physical presence, like Fitness at 1.6%, Music at 1.5%, and Fashion at 1.3%, had almost none. The pattern makes sense. Generative AI can fake a voiceover and an animation far more easily than it can fake someone actually doing a workout or playing an instrument.
Some creators are using AI tools thoughtfully, as a production aid rather than a replacement for genuine content. That distinction matters. But the incentive structure on social media rewards volume, and AI makes volume trivially easy to produce. When a creator can generate dozens of videos in the time it once took to make one, platforms fill up with content that is technically watchable but offers very little of substance. TikTok’s algorithm, at least in its default state for new accounts, appears more than willing to distribute all of it. For a generation of children growing up with these feeds, that’s a problem that user controls and literacy funds alone may not be enough to solve. We’ve also covered TikTok’s broader societal concerns and the risks children face on other platforms like Roblox, where similar content safety questions keep coming up.