A cartoon car-ride song that's been making the rounds looks like any other toddler time-killer—until the kids float outside a moving vehicle, ride on the hood going backward, and chirp that "red means stop, and green means right." Children's media experts say clips like "Vroom Vroom! Car Ride Song" are part of a fast-growing wave of low-quality, artificial intelligence-built videos for kids that can be misleading at best and "downright dangerous" at worst, as one expert puts it to Mother Jones. One YouTube channel behind such content has pushed out more than 10,000 videos in seven months, a pace that human creators can't match. A Kapwing analysis estimates roughly one-fifth of YouTube's feed is now what's being labeled "AI slop." Researchers warn that for very young viewers, this isn't just junk—it's potentially rewiring brains.
Pediatric surgeon Dana Suskind calls it "toddler AI misinformation at an industrial scale," pointing to clips that show babies eating whole grapes (a choking hazard), bloody-looking apples, and honey, which carries botulism risks. There are also bungled lessons on vowels that inexplicably contain consonants; US states, with the states' names spelled incorrectly ("Oklolodia" for Oklahoma?); and continents, with directional compasses featuring more than four points. Kids-media veterans argue that platforms, AI companies, creators, and parents themselves all share responsibility for the phenomenon, but with limited regulation and automated moderation, the burden is landing on caregivers to spot the slop. "We're at the beginning of a monster problem," says child-development expert Kathy Hirsh-Pasek, "and we have to get hold of it quickly." More here.