Corey Turner

Students just tell them to do something, and then the chatbot does it.

Here's Rebecca Winthrop.

She's one of the researchers on the study.

Ayesha, Winthrop told me if students rely on this kind of AI too much, it can actually stunt the kind of brain growth, wiring, that comes from the trying and doing and failing and trying again.

The other is social emotional growth.

So it's in childhood, right, that we learn how to get along with others, hopefully, especially people who may look and think and feel differently from us.

But these free chatbot tools are designed to be sycophantic.

What that means is they tell the user essentially whatever the AI thinks the student wants to hear.

For children and teens now, this can be really intoxicating because the user is always right.

Again, here's Rebecca Winthrop.

And Aisha, the stakes are obviously a lot higher than kids refusing to do dishes.

The stakes are children growing into adults who never learned empathy or how to relate because they spent more time engaging with chatbots than they did with other kids.

And Winthrop told me one in three teens in the U.S.

who use AI say they actually prefer talking about important or serious subjects with a chatbot than they do with other people.

So the report says AI designed for use by children and teens, for one thing, should be less sycophantic and more what they call antagonistic.

So it pushes kids' preconceived notions.

But one of the biggest recommendations they make is really for governments to do more to regulate the use of AI by children.

And in the U.S., we're at a really weird impasse right now.

The Trump administration has issued an executive order trying to prohibit states from regulating AI for themselves.

But Congress hasn't created any federal regulations so far.