Well, it depends on where the interop DLL came from. They can be built
by hand, or generated automatically, or something in between (start with
one built automatically and edit the code to match specific requirements).
If you wind up compiling code at some point to generate the interop
assembly, you can adjust the target CPU at that time. Otherwise, you are
stuck with whatever the target CPU was set to.
In any case, even if you can get the target CPU to match on the
assemblies, there is still the issue of the native library being
referenced through the interop assembly. The target CPU for _that_
library will also have to match the run-time JIT-compiled code for the
managed assemblies in use, and you will not get a compile-time warning
about that if it is wrong. it will just cause the process to exit when the
mismatched assembly tries to load.