Even if there are such changes, it seems unclear to me that they can be attributed to kids playing outside unsupervised.
But my understanding is that most child molestation is by trusted adults known to the family (or family members), most kids nowadays (certainly by age 12 or 14) have plenty of access to porn (probably much easier than in the past), kids still fight during and after school, and injury rates aren’t especially changed. I could be wrong about though, I haven’t looked up precise figures.
Just making up a list of scary sounding hypothetical situations and anecdotes isn’t the best way to judge risk.
The biggest causes of death for children in the US are cancer, car collisions, drowning, gun homicides, and suicide. If we want to make life safer for children we should be focusing on reducing car use and using traffic engineering to slow cars down (ideally <20 miles/hour) in populated areas, pay more attention to home swimming pools, get rid of guns to the extent possible, and figure out what in the society is making teenagers so unhappy they would want to kill themselves (maybe has something to do with their lack of autonomy).
At the end of the day it's just weighing a bunch of different risks and outcomes against each other. Different people come to different conclusions about "what is best" and there can be no definitely right or wrong approach - not unless people could agree on what they actually wanted to achieve/raise.
Citations needed.
People need to stop knee-jerk reacting to things and understand trends often happen in context to previous happenings. The helicopter movement arose after we realized in the 70s and 80s that free-range childhood often led to some really dangerous things happening to kids, that unsupervised time actually helped cover up things like sexual abuse or violence. It isn't just millenials being millenial or something.