However, I don't think that the wording should allow silent censorship,
just because the message content is 'malicous or shows a deceitful
I respectfully disagree. My feeling is that there is no reason that the
SMTP protocol should guarantee that a notification will be sent for every
message that is not delivered, except when the MAIL FROM address has been
spoofed. I think that if we're going to change the protocol, it should
recognize the reality that sometimes messages are dropped for local policy
reasons. This practice should be discouraged, except when the server can
determine that messages that are malicious or deceitful.
I have no objections John Klensin's suggestion to add "accurately" or
"positively" in front of "determine".