> I had unrestricted internet access as a child and turned out fine.

The statistical power of an n=1 study applied to a population many orders of magnitude larger is not very strong.

That aside, many, if not most children below a certain age lack the requisite ability to discern danger/non-danger with a fidelity that would satisfy their parents who have moral and legal responsibility in that domain. I admit there's a tension between privacy and the duty to protect.