From my understanding domain validation certificate are verified by receiving an email on some special email addresses on the domain (e.g. postmaster@example.com, admin@example.com) or by sending on the designated email address on the domain name registration record (the information published on WHOIS).

Since many emails servers from non major email service providers do not encrypt and only few email providers even supports STARTTLS. Therefore, it is possible to intercept emails sent to these vulnerable email providers (how hard would this be?). Is domain validated certificate somewhat useless if the website's designated contact email address is hosted in less than secure email provider (e.g. self managed email servers in particular)?

What would prevent me from being issued a valid SSL certificate signed by a recognized CA for a domain I do not own by doing a passive/active MITM on an email service provider?

I'd assume that if you control the authoritative DNS, then you also control the Domain Validation by definition, so that's not realty a vulnerability. However, DNS caches is another matter, how could a CA make sure that its DNS entries comes from a legitimate source especially if the site owner don't make use of DNSSEC?

  • Just to be clear: you're considering a case where an attacker eavesdrops on (or impersonates) email between the domain's WHOIS contact email address and the CA, yes? Based on your comment below, it seems like the two parts to your question are: (1) Does the security of SSL reduce to the security of the email communication between the domain owner and the CA? (2) If so, how secure is that, exactly? – apsillers Feb 16 '15 at 13:29
  • @apsillers: yes, that is correct, I've clarified the question a bit to that effect. – Lie Ryan Feb 16 '15 at 13:36
  • 4
    So the whole question is about How realistic is it, that I can man-in-the-middle email traffic of a sysadmin? – sebix Feb 16 '15 at 13:47
  • @LieRyan Does the linked topic answer your question? If yes, you can mark this one as duplicate. If not, please explain in more detail. – sebix Feb 17 '15 at 9:30

So yes, if the operator of www.dodgy-domain.elbonia lists a WHOIS contact address of admin@dodgy-domain.elbonia, or if a CA can be convinced to send a verification e-mail to postmaster@dodgy-domain.elbonia, and that operator has not set up SSL support in e-mail yet or the CA doesn't have outgoing SSL support for e-mail, several parties could potentially intercept that e-mail, including the Corrupt Communications Ministry of Elbonia or someone with a tap on the party-line modem link out to the operator's site.

There's a chicken-and-egg problem too, because you can't receive STARTTLS-protected e-mail without a certificate on your mailserver.

However, that isn't necessarily a problem for the Internet as a whole, or for the majority of domains, or for a typical domain. Google, Yahoo!, Microsoft, mail.com, Apple all support STARTTLS, and a lot of newly established domains, personal domains, or small-business domains would have WHOIS addresses from one of those services rather than from the domain itself. Other organizations use Google Apps or hosted Exchange for their incoming mail, which means they get the same STARTTLS protection. Notably, markmonitor.us, which handles domain registration on behalf of a lot of Fortune 500 firms and other large organizations, uses Google Apps hence supports STARTTLS.

Also, a passive attacker (who could read the message but not block its delivery) would leave the "confirmation" e-mail in the target mailbox, which gives some risk that the real administrator will notice the e-mail, contact the CA, get the certificate cancelled, publicly warn his users, and report the matter to the police.

Note that at least one CA does not issue DV certificates in part because of concern for possible man-in-the-middle attacks.

  • I disagree with your attacker model. Advocating for STARTTLS as a solution here suggests a passive attacker (indeed, you mention potentially delivered emails as red flags), but when the traffic is passing through (or routed through) adversary-controlled infrastructure (such as Elbonia's ISP), I see no reason not to assume active capabilities. I'm not aware of any countermeasures against an active attacker between the CA and the server, and that's a scary thought. – Joost Jul 6 '16 at 14:41

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.