Security Collapse in the HTTPS Market
queue.acm.org
queue.acm.org
Weakest link. A crucial technical property of the HTTPS
authentication model is that any CA can sign certificates
for any domain name. In other words, literally anyone can
request a certificate for a Google domain at any CA
anywhere in the world, even when Google itself has
contracted one particular CA to sign its certificate.
It's the elephant in the room that everyone tries to ignore. Google's Chrome browser tries to mitigate the problem, but only for select sites.Firefox has made countless GUI tweaks and other useless changes over the years but has ignored this problem until very recently.
Plus, as the article makes clear, everyone and his brother can be a Certificate Authority. The first CA I see when I view certificates in Firefox is:
TÜRKTRUST Elektronik Sertifika Hizmet Sağlayıcısı
The authorities in Turkey are delighted to have that entity there. But WTF? Why should my security be dependent on their non-malevolence? I sure don't trust them. Not after this: http://nakedsecurity.sophos.com/2013/01/08/the-turktrust-ssl...When I was looking at the default root certificate repositories that the major browsers and operating systems use. They are mostly your regular list of CAs and governments, but there's one name that popped out as unique: AOL.
America Online has two legacy certificates[1] in the Microsoft[2], Apple[3], NSS[4], and Android[5] default list of root CAs. I'm assuming this is from back when AIM was all the rage, but remarkably AOL has been keeping up the audits[6] for them. Does anyone have any more info on the history of these certs?
I think might be a great opportunity to address the two problems above. Could AOL start offering free SSL Certificates?
Pros:
1. Their root certificates are already in everyone's list (backwards compatibility).
2. Their core business model is not issuing certificates (not seen as a racket).
3. They would get a huge press coverage for being a "savior of HTTPS" or some such spin (positive spotlight for AOL).
4. There would now be competition in the free SSL cert market (maybe other CAs would start offering free options, too).
Cons:
1. This would be a cost for AOL. Perhaps other tech companies could partner with them to subsidize the cost of issuing the certificate? Perhaps there could be kickstarter to pay for the costs? Perhaps AOL could spin off a non-profit foundation or donate the certificates to Mozilla?
2. Unforseen technical problems associated with starting to chain to a certificate that hasn't been in active use for a long time. I have no idea what these could be. Thoughts?
3. SSL certs would likely be issued with no warranty (since they are free). Not a deal breaker in my opinion, because the scope for these could be for non-commercial use.
Anyway, just tossing out this idea for feedback. There's no sense in pursuing this further if there's technical reasons making this impossible.
Again, does anyone know anyone who works at AOL? I would love to talk to someone there about this.
[1] - https://pki-info.aol.com/AOL/
[2] - https://social.technet.microsoft.com/wiki/contents/articles/...
[3] - http://support.apple.com/kb/HT5012
[4] - https://www.mozilla.org/en-US/about/governance/policies/secu...
[5] - https://android.googlesource.com/platform/libcore/+/master/l...
[6] - https://pki-info.aol.com/AOL/2013_AOLRoot_Audit_Attestation....
We don't have a web of trust, where my distrust of the Chinese government can be indicated and influence the measure of trust afforded, or where I can revoke my computing platform's implicit trust of the Chinese government. I might not trust their intentions, or I might not trust their security invulnerability.
If you are suggesting that AOL becomes the sole CA, I see how this clears a lot of bad stuff, but I'm not sure how it fixes anything. You seem to want to me trust AOL, but maybe I don't. Furthermore, I may trust them today yet have concerns that absolute power corrupts absolutely.
HTTPS design depends critically on absolute systemic integrity -- like a benevolent monarchy, it's great when it works, but a single failure to exclude malicious entities is presumably and likely fatal, at least as regards design's intent. The system keeps "working" without meeting its goals, and any security system that does this is actively harmful and is better off halting.