By default the "Auto Detect Unicode" only works for UTF8 files with BOM. More precision needs to be added in the Help for this part, telling people to either manually change encoding or add a BOM.
PS: I haven't got the time to do test with UTF16 or UTF32 so I have no idea.