So when you copy an image from Finder in MacOS, if you paste it into a terminal you get the file path and if you paste it into a comment on github you get the contents of the file as a binary blob.
When pasting, an application is essentially querying the clipboard for a representation of the data it can use, and if there is none, it either defaults to something or fails.
In a well designed program, all of the translating and parsing is done before sending to the clipboard. If you copy text from MS Word and paste it into a content editable field in a website, you might expect some consistency.