Password Hashes on AG Nodes

I'm comparing the password hashes of SQL logins on AG nodes. Of the twenty SQL logins 15 have the same hashes over the 2 nodes and 5 have different hashes. Is a different hash an indicator of a different password, or could a salt have been added to some passwords and not others. Ideally we wouldn't be using SQL logins at all, but our hand is forced in some instances