Which “Format As” output are you using in the compiler? The only ones that will invoke the embedded MultiMarkdown post-processing engine are those that are tagged as being MultiMarkdown. If you use one of the regular outputs, no post-processing will be done and you’ll just get the text as you typed it, codes and all.
To be clear, Scrivener doesn’t really “recognise” the text as being anything—all of this is handled by MMD itself, we’re just making that part of the job easier for you, so you don’t have to go and mess around with complicated command prompt utilities to create an HTML file, etc. We also do generate a little MMD where we can—you can use the “Title” checkboxes in the Formatting compile pane to automatically produce headings at their appropriate level of depth, for example, rather than having to sort out all of those hashmarks yourself. The chapter on MultiMarkdown in the user manual PDF covers what we support for output, but at its most basic usage you can just feed a Markdown file into the editor and then compile using one of the MMD->FORMAT options to get a nice result.