When trying to prove security of asymmetric signature, for instance for existential (or strong) unforgeability against chosen messages attack, do we need to consider the signer as a possible adversary?

-