How to Shrink MDF Files in SQL Server 2012

To answer your question, that "all databases" selection will apply the size-monitoring to all databases, it will not shrink all databases when one grows too big.

Shrinking databases is usually not recommended, as Julien pointed out; especially on an automatic basis. I have shrunk databases before but always manually, after carefully evaluating "why" it needs to happen.

Related Q & A:

When is it OK to shrink a Database?
Should I shrink my database?

I've tried to use the GUI to shrink the database and the files but it doesn't actually change in size.

You said that you truncated large table, in such case you might not be able to shrink "immediately". From Dropping and Rebuilding Large Objects

When you drop or rebuild large indexes, or drop or truncate large tables, the Database Engine defers the actual page deallocations, and their associated locks, until after a transaction commits. This implementation supports both autocommit and explicit transactions in a multiuser environment, and applies to large tables and indexes that use more than 128 extents.

It also says that

Because the physical phase occurs after a transaction commits, the storage space of the table or index might still appear as unavailable. If this space is required for the database to grow before the physical phase is completed, the Database Engine tries to recover space from allocation units marked for deallocation. To find the space currently used by these allocation units, use the sys.allocation_units catalog view.

Other reason why you would not be able to shrink immediately because records are marked ghosted. And shrink will actually give you space backup after background Ghost Cleanup operation has completed, some more information about Ghost Cleanup Task

Initial size as such is not initial size actually, its the current size. You can shrink below initial size. What I suggest you is wait for some time till background operation finishes. Use dbcc shrinkfile tsql command to shrink the files and shrink in small chunks.

Shrinking is single threaded and takes time.

PS: If you are not aware, please note that shrinking causes massive logical fragmentation so don't forget to rebuild fragmented indexes after shrinking is done