If your threat model includes the U.S. government you are in the very, very, very, very, very minority of the population of selfhosters.

If you’re 100% sure that the dupes are only between your source of truth and “everything else”, you can run fdupes then grep -v /path/to/source/of/truth/root the output - all the file paths that remain are duplicate files outside your source of truth, which can be deleted.