I agree. People who send this data externally are ultimately responsible. Hypothetically, If I use the internet to steal data from my employer it’s not the network teams fault for allowing it to happen. I’m just a thief in position of trust exploiting my capabilities.

You can't have a walled garden and not be responsible for the things that take place in it

However you have to be much more intrusive on developers if you are going to require semantic analysis of all data being sent to see if it was justified and whether it was mentioned in a (plain text, localized potentially into many languages) privacy policy.