Wandering Thoughts archives


The conflict between wildcard TLS certificates and Certificate Transparency

Certificate Transparency is an increasingly important part of the modern TLS world, especially for website certificates (which I believe are still the dominant use of TLS certificates). One part of Certificate Transparency is monitoring for certificates issued for your own sites and domains, but that's not the only use; another one is looking for certificates issued to suspicious names. For instance, a bunch of people would probably be interested if someone issued a certificate for www.really-its-paypal-login.com or geeglemail.com or any number of other TLS certificate names like that.

(This is not just of interest to the real Paypal and Google, it's also of interest to things like email spam filtering systems, intrusion detection systems, and the systems that help provide browser warnings of suspicious websites.)

The examples I've shown here are top level domain names, but that doesn't have to be the case. Often it's going to be easier to get an alarming name into a subdomain instead of a domain; for a start, you don't have to worry about a domain registrar alerting the moment something that matches *paypal* shows up in a new domain registration. When an attacker embeds the alarming name as a subdomain, one of the few ways that outside people can spot it is when the TLS certificate information shows up in the Certificate Transparency logs, because the TLS certificate exposes the full host name.

Well, at least until wildcard certificates come along. When combined with CT, the effect of wildcard certificates is to hide from scrutiny all of the names that can be put into the wildcarded portion. People monitoring the CT logs no longer see 'login.paypal.really.somedom.wat'; all they see is '*.somedom.wat' or '*.really.somedom.wat', which of course means that they basically see nothing.

(There are good aspects of this as well as bad ones, since CT with full host names exposes internal host names that you may not want to have known for various reasons.)

As a result, I'm not particularly surprised that Let's Encrypt doesn't support wildcard certificates. Let's Encrypt is intended for public hosts, and with automated issuance I feel that Certificate Transparency is especially important in case something goes wrong. Not issuing wildcard certificates maximizes public visibility into what LE is actually doing and issuing.

With all of this said, Let's Encrypt's FAQ says that their primary reason for not issuing wildcard certificates is the question of automated issuance (which I suspect partly means automated proving of control), not any philosophical reason. It's possible that LE would decide they had philosophical reasons too if people came up with a good technical solution; I guess we may find out someday.

web/CertificateTransparencyVsWildcardCerts written at 21:57:12; Add Comment

Page tools: See As Normal.
Login: Password:
Atom Syndication: Recent Pages, Recent Comments.

This dinky wiki is brought to you by the Insane Hackers Guild, Python sub-branch.