You can imagine it just being easier to code indiscriminate attacks where they only review the results to pick who it is worth collecting the ransom from.

Unencrypting for vicitims in the US that couldn't pay would just add more exposure risk to them of getting caught, so they would have no incentive to actually do it. It would take a large bit of money out of the system, but it seems like you need all countries to coordinate and that one country doing so on its own, enforcing a no pay out rule, won't have much effect on non-targeted attacks.

How many of these attacks are fully automated in the initial attack/encrypt phase vs. human operators explicitly working to more fully infiltrate a target?