Sender Policy Framework is a system to detect spoofed emails.
From http://www.libsrs2.org/overview.html :
WHAT EMAIL FORWARDING SERVICES NEED TO KNOW ABOUT SPF
To combat spams, worms, and viruses, the new anti-forgery standard called SPF adds a layer of protection to SMTP. When an SPF-aware MTA receives mail (from, say, username@aol.com ), it asks the envelope sender’s domain ( aol.com ) if it recognizes the IP address of the SMTP client. The domain publishes SPF records in DNS describing its outbound servers, as a sort of “Reverse MX” record. If those records do not describe the client IP, the MTA may reject the SMTP transaction as a forgery attempt.
How can SPF help in detecting and preventing spam and worms?
I would point out that perfectly legitimate emails can be erroneously caught by SPF too, > say registration confirmation or password reset email sent direct from webserver that has > different IP address than the mail server). – ewanm89
So have the web server talk to the mail server to send and not send directly. slow, more resources but ultimately a good cause?