TikTok serves up AI-generated junk at an astonishing rate. A fresh report reveals that 59% of the first 500 videos shown to a brand-new account come laced with low-quality synthetic visuals, robotic voiceovers and bizarre animations designed to farm views. That’s nearly three times the proportion found on YouTube Shorts in the same test.
The findings come from video creation platform Kapwing, which set up clean accounts on both platforms and scrolled their algorithmic feeds. On TikTok the slop piled up fast. Of those initial 500 videos, 294 qualified as AI slop. The rate climbed higher the deeper the feed went. And the problem hits children hardest.
Kapwing defined the term clearly. AI slop means careless, low-quality content generated using automatic computer applications and distributed to farm views and subscriptions or sway political opinion. Think distorted cartoon characters in nonsensical scenarios. Synthetic narrators stumbling through basic facts. Visual glitches that break immersion the moment attention lingers.
Children’s content stands out as the worst offender. Some 57.4% of videos in TikTok’s kids category counted as slop. The hashtag #CartoonKids proved even more extreme. Only three of 100 videos examined appeared human-made. The rest? Nearly all synthetic. Digital Trends reported on the study yesterday, highlighting how preschoolers lack the tools to separate polished educational clips from confident but inaccurate imitations. A counting video that gets numbers wrong might amuse adults. For young viewers it plants confusion.
Other categories showed heavy contamination too. Science and education hit 35%. Health reached 33.8%. History sat at 33.5%. Contrast that with fitness, music and fashion. Those areas stayed clean, with slop rates below 2%. The pattern suggests algorithms push synthetic material hardest where engagement comes cheap and production costs near zero.
But why TikTok in particular? The platform’s For You Page optimizes aggressively for immediate watch time. Creators churn out dozens of videos in the hours once needed for one. Generative tools make that possible at scale. Kapwing’s earlier work on YouTube, referenced in the new study, found only 21% of Shorts served to new accounts qualified as slop. Tubefilter noted the three-to-one ratio and observed that slop density appeared to grow as users scrolled further. “TikTok has a slop problem,” the report states bluntly.
The numbers reflect a broader shift. By November 2025 TikTok had labeled 1.3 billion videos as AI-generated. That figure captures only disclosed content. Undisclosed slop likely adds more. And once an account lingers on synthetic videos the algorithm doubles down. Interest in one piece of AI content signals appetite for similar fare. Feeds turn quickly into echo chambers of machine-made material.
Platforms aren’t blind to the issue. YouTube has cracked down on repetitive, low-effort channels and introduced tools allowing users to limit exposure to such content. TikTok offers an option to see less AI-generated material and has announced funds for AI literacy. Yet researchers question whether these steps suffice. Tubefilter argued that more active moderation will prove necessary if the company wants real change. User toggles alone probably won’t stem the tide.
Concerns run deeper than mere annoyance. Pediatric experts worry about impacts on developing brains. Repeated exposure to distorted facts or malformed narratives can erode trust in information itself. Young children absorb patterns from what they watch. When much of that input comes flawed or surreal the effects compound. One expert cited in coverage warned that the sheer volume changes the baseline experience of childhood media.
Creators face their own pressures. Genuine producers compete against accounts pumping out hundreds of clips daily with minimal oversight. Some legitimate users experiment with AI as an assistive tool. They add original scripting, editing judgment and personal voice. The report distinguishes those efforts from pure slop. The distinction matters. Quality depends on human direction, not the mere presence of machine assistance.
Still the incentives favor volume. Ad revenue scales with views. Platforms reward retention. Synthetic content often hooks initial curiosity even if it fails on closer inspection. And detection remains difficult. Many clips mimic human styles just enough to pass casual review. Only sustained attention reveals the tells: unnatural motion, factual slips, repetitive phrasing.
Recent coverage reinforces the urgency. Futurism reported two days ago that TikTok appears completely taken over in the eyes of new users, with kids’ content hit hardest. The piece echoed Kapwing’s data and warned that the default experience for anyone downloading the app now skews heavily artificial. Similar discussions have circulated on X in recent days, with users sharing examples of bizarre AI kids’ videos dominating feeds.
What comes next? TikTok could tighten labeling requirements, boost human review for high-engagement categories or adjust algorithmic signals to penalize low-effort synthetic output. Yet each fix carries trade-offs. Overly strict rules might discourage creative experimentation. Weaker measures allow the flood to continue. The company must balance growth, user satisfaction and responsibility toward younger audiences.
For now the data paints a clear picture. New users stepping onto TikTok encounter a feed where synthetic content dominates from the start. Children searching for cartoons or learning videos wade through even higher proportions of machine-made material. The gap versus YouTube is wide. And without stronger intervention that gap may define the platform’s reputation in the years ahead.
Industry watchers will track whether TikTok’s response matches the scale of the problem. Kapwing’s report offers a benchmark. Future tests can measure progress. Or decline. The first few scrolls on a fresh account tell the story before any personalization kicks in. Right now that story features far more slop than substance. And for a generation raised on these feeds the long-term effects remain unknown.
WebProNews is an iEntry Publication