A Kapwing study found nearly all cartoon videos on TikTok are machine-made. YouTube's numbers are bad too, but the gap points to a structural difference.
Open a fresh TikTok account and scroll for a few minutes. Roughly six in ten videos that surface will be AI-generated, low-effort synthetic filler optimized to game recommendation algorithms, according to a study by the video editing platform Kapwing. On YouTube, that figure drops to around one in five.
The gap reflects how each platform's recommendation engine weights novelty and volume against quality signals, and how each has chosen to intervene or declined to.
Children's content is the sharpest edge of the problem
The research found that 97% of videos tagged #CartoonKids on TikTok are AI-generated. Across children's content more broadly, 57.4% of videos served to young users are machine-made. Pediatricians have warned that exposure to AI-generated misinformation at this scale poses risks to developing brains, though the specific mechanisms of harm remain a subject of ongoing research.
Get the latest model rankings, product launches, and evaluation insights delivered to your inbox.
Florida has sued TikTok over what the state alleges is misleading communication to parents about what children actually encounter on the platform.
TikTok has labeled over a billion AI videos and is now testing a filter
TikTok has labeled more than 1.3 billion AI-generated videos on its platform. That figure also shows the limits of disclosure: labeling a video does not remove it from a child's feed or reduce its algorithmic reach.
TikTok is now testing an opt-in slider that would let users filter synthetic content out of their feeds. YouTube and Meta have not shipped equivalent tools, relying instead on disclosure badges and watermarks. If the slider reaches general availability, it would be the first major platform feature to give users direct control over the ratio of AI content they see, rather than simply identifying it after the fact.
Whether TikTok deploys the filter broadly, and whether regulators treat voluntary tools as sufficient given the children's content figures, will determine whether the next round of platform accountability focuses on disclosure or on feed-level control.