[cryptography] Auditable CAs

Marsh Ray marsh at extendedsubset.com
Wed Nov 30 14:30:40 EST 2011


On 11/30/2011 12:01 PM, Ben Laurie wrote:
> On Wed, Nov 30, 2011 at 5:16 PM, Marsh Ray<marsh at extendedsubset.com>  wrote:
>>
>> Perhaps you define this category of "publicly visible certs" as "certs
>> which display without warnings on default-configured browsers when
>> presented by the correct site".
>>...
>> On the other hand, one could interpret this category of "publicly
>> visible certs" as "certs visible to the public", i.e., "certs served by
>> legitimate servers on routable IPs located via public DNS". But this
>> interpretation would be much weaker (and I don't think that's what you
>> mean).
>
> Although I rather like your first definition, this one seems closer to
> the truth: it may be that some sites which currently validate
> correctly in default-configured browsers would choose not to in our
> system.

The certs I am worried about though are the certs that were issued in 
secret (e.g. Comodo and friends) and are never "publicly visible" until 
they are used in an attack.

If the attack is sufficiently targeted, it may be the case that no one 
other than the victim ever sees the cert at all. In the event of a mass 
MitM attack (e.g. *.ir), the attacker would likely have free use of his 
previously-hidden cert for at least as long as the combined reporting, 
reaction, and revocation latency.

It's not clear how this proposal is actually an improvement on the 
current system in this regard.

On the other hand, if you *did* engage the CAs and get their buy-in, 
they could pledge to update the log promptly with every cert they 
issued. Specific CA certs could be configured with this flag in the 
browser's trusted store. This would allow a missing audit proof to be 
treated as a hard stop and would seem to be a more meaningful 
distinction among CAs than the current EV scheme. (The few CAs I've 
spoken were really grasping for ways with which the 'better' CAs could 
distinguish themselves in the industry.)

Additionally, such a flag could be added to HSTS. Rather than pinning to 
a specific CA ("I will only use this one CA ever"), a site could pin 
itself to the use of a CA that promised to participate in the auditing. 
This would reduce some of the DoS potential inherent in CA pinning yet 
still allow browsers to catch that critical transition from "provably 
logged cert" to "non-logged cert".

>> But the proposal does nothing _directly_ to prevent a CA from issuing a
>> cert, right? And since browsers aren't logging the certs as they find
>> them, this doesn't inform the owner of the domain either.
>>
>> Instead it seems to be a hoped-for effect of "default-configured
>> browsers will raise hell if they are presented with a non-logged cert
>> and CAs will feel compelled to go along with the audit logging".
>
> CAs do not have to go along with anything, the log will accept a cert
> from anyone - which obviously includes the owner of the cert.

There would need to be a way for end-users to report new certs via their 
browser, much like they report browser crashes today. Wouldn't some 
users want it? I think it would be good to involve the users in this 
process as much as is practical.

>> they'll have to put the certs they
>> issue in the logs too, right?
>
> Someone will, yes.
>
>> Wouldn't they have to put the certs they sign in the public log? They
>> don't have to do this today.
>
> No, but their certs are already publicly visible today.

I don't believe this is the case. It's one of the big problems with the 
system we have today.

Consider a sub-CA which is issed for the purpose of a company's 
deep-inspecting firewall (e.g., a BlueCoat). The device will use the 
sub-CA to issue new certs on-the-fly for each new website that the 
internal network clients browse to. The rest of the world (hopefully) 
never sees those certs.

Yet this log of the certs that it has generated is highly confidential. 
It contains info about the browsing history of the entire company, e.g., 
parts suppliers, financial institutions, use your imagination.

The current crop of "trusted" CAs refuse to give the names or even the 
count of the sub-CAs they've sold. They only require that the party to 
which they sell them agree in a contract to use them accordingly.

I'm all for saying that these sub-CAs need to be put on the boat to the 
island of lost toys like the toxic plastic that they are. But I wouldn't 
expect the parties that currently enjoy this privilege to go quietly. :-)

- Marsh



More information about the cryptography mailing list