OO at the OS level

Object-oriented programming is all the rage these days, but we are still stuck with a procedural way of interacting with the operating system. Why? First of all what is the difference between OOP and procedural? Procedural defines an algorithm that data will follow. If the data does not work for that algorithm, the algorithm needs to be changed. OOP on the other hand defines code and data together. The algorithm is designed to go with the data, and a nice neat package is presented to manipulate. So how are we interacting with an operating system in a procedural way and how could it be object oriented? Well currently we have jpegs, mp3s, docs, and txts that all need a specific program to deal with them. Oh, what you don't have a pdf viewer? You can't open that pdf. Too bad, so sad. What I propose is to have the methods for dealing with data embeded in the files. What you want to work with that type of data? Call the 'open' command for it. Edit, well call the edit command. (Now of course these would be menues or work by clicking.) Ok problems with this idea...
Security, security, security... Who decides what code goes with what data? How do I know a pdf is not installing a trojan? Well, I suppose we can put limits on what an open command does. Shouldn't do much more than load to memory. Anything more than that and it gets the ax. We can also incorporate digital signatures to make sure the Adobe format was infact created by Adobe.
Then there is multiple copies of code... First idea is who cares? I have terabytes of storage, but yeah, no good. This one will need a trick to work.