I understand what the OP is out after:
In RSA, theres something called blind signing.
Applying the PRIVKEY to a plaintext, the resulting chipertext can only be decrypted by applying the PUBKEY to the text.
Applying the PUBKEY to a plaintext, the resulting chipertext can only be decrypted by applying the PRIVKEY to the text.
Then blind signing is applying a factor X to a key, so the signer does not know the contents of the message.
If the message is M*X, the signature is S*X provided that S is a signature of M.
If E is a encrypted message encrypted with keypair consisting of PUB A and PRIV B it will be:
Apply A to P and gain E.
a adversiary can fool the receiver to decrypt the message as:
Send to owner of B.
Owner applies B to E*X and yeld P*X.
Adversiary removed X by dividing P*X with X, and yelds the plaintext P.
More info: http://en.wikipedia.org/wiki/Blind_signature
The OP wonders if there is similiar risk with signing a adress with its own key and risking leaking the key or something.Can say that since the adress is a hash of the pubkey, its NO risk whatsoever to sign the adress.
There MIGHT be riskes with signing public/private keys, but I don't know enough about ECDSA to prove it false or true.