Irrespective of any MITM issues
What on earth persuaded the makers that software to interrupt your browsing to insert advertisments was a good idea?
(Yes, this is my rant of the month.)
The US Department of Homeland Security's cyber-cops have slapped down PrivDog, an SSL tampering tool backed by, er, SSL certificate flogger Comodo. Comodo, a global SSL authority, boasts a third of the HTTPS cert market, and is already in hot water for shipping PrivDog. What is PrivDog? Let's allow the US Computer Emergency …
It's very cunning how Comodo put a positive spin on replacing ads that pay for content with ads that line their own pocket.This is massively underhanded.
Does anybody know if they *only* replace malicious ads, or do they replace them all?
Given that many websites currently detect AdBlock, I presume that it will soon be just a matter of time before many more websites will detect (and flag) the use of Ad/content replacement/substitution tools. Certainly where a site earns revenue from the Ad's served I would be looking into this as a revenue protection issue.
Isn't this essentially theft of service? A website provides useful content and uses ads to support that content but if a MITM is blocking the ads that pay the website and substituting ads that pay the MITM it seems to be a clear case of theft.
It was only nasty cheapo SSL certificates which were busted. Up until now this was fixed by extra-cost extra-trustworthy certificates which displayed a green splodge for higher security.
Unfortunately the extra-cost certificates are now busted as well, but don't worry as you will soon be able to get extra-extra-cost certificates with a gold splodge.
I never got why we need a CA anyway.
If I trust Facebook, then I trust Facebook. I don't necessarily trust every website ever created by anyone who's bought a certificate from the supplier that Facebook's bought their certificate from.
The CA is merely convenience in the process so that the first time I go on B&Q.com, I don't have to guess whether or not I can trust the certificate.
But a browser that "accepts" the first certificate it sees for a domain and remembers it forever after, that will flag if the cert changes or expires, more than fulfills most of the issues. We already do this for SSH, for instance.
Tie it in with a DNSSEC system where the authoritative, untamperanle DNS response for a website includes a proof-of-certificate and you pretty much wipe out the CA's function entirely. We already do this for email (DKIM, etc.).
To do this for websites isn't that much more of a push forward, and now we have the impetus.
Key security should be in the DNS, and should be tied - the .uk root should be saying THIS is the cert for the .co.uk TLD and it's the only one I specify. And then when asked, .co.uk will say THIS is the cert for the facebook.co.uk site (and here's the IPv4 and IPv6 addresses). And then Facebook can specify what THEY want under that domain as required. All signed, all authorised, back to the root.
"I never got why we need a CA anyway.
If I trust Facebook, then I trust Facebook. I don't necessarily trust every website ever created by anyone who's bought a certificate from the supplier that Facebook's bought their certificate from.
SNIP
Key security should be in the DNS, and should be tied - the .uk root should be saying THIS is the cert for the .co.uk TLD and it's the only one I specify. And then when asked, .co.uk will say THIS is the cert for the facebook.co.uk site (and here's the IPv4 and IPv6 addresses). And then Facebook can specify what THEY want under that domain as required. All signed, all authorised, back to the root."
Actually, I don't think security by network (which seems to be your point) is enough. After DNS is secured, you have to secure the transport, and the physical layer etc ... Long efforts before you're SURE www.facebook.com is really facebook.
SSL took the approach of end to end cryptography, which is desirable and good.
The only problem is there is a gap: all CAs are hard coded in the browser and no user ever look at them (they're so obscure ...) and there is no secure directory service. That's the current security hole exploited by superfish and its siblings and the real short-coming of SSL.
As long as people use this practice of 'breaking the chain of trust' there are bound to be some who implement it utterly wrong.
In other words, as long as it is possible to exploit this, someone will. Not exactly deep wisdom, but true nonetheless. It seems it is past time to plug this gap.
You just hit a HARD problem in computer security. It basically boils down to a question of "Who can you trust?" Because the first rule to having ANY form of trust system (chain, web, et al) is the need to trust someone or something; IOW, someone has to play the role of Trent. Only problem is, given sufficient resources, Mallory (or Gene) can impersonate anyone: including Trent. So ask yourself, "What now?"
Other ad blockers block ads. When you go to your bank's website with other ad blockers you're pretty sure you're connected to your bank as the bank's certificate is trusted. This acts as a proxy and uses its trusted certificate to tell you that the bank's certificate is legit. The problem is that it doesn't actually care if the certificate is actually legit. If it's not your bank's website and is actually a clone made by some guy sitting in Lagos with a laptop that you're giving your bank credentials to, well the proxy is ok with that because it didn't check to see if Mr. Lagos' certificate was valid.
This post has been deleted by its author
Isn't it funny how anti-crap software always morphs into the opposite?
And SSL is so awesome, it couldn't prevent this either.
And the company pushing SSL the hardest, which everyone loves and can do no evil, has, like Comodo, morphed into a self-perpetuating robot army that feeds on sleazy advertising.
*sigh* Why can't we have an internet that favors decent ads, and honest search results, and privacy, and...?
Let's be fair here. Implemented properly a local mitm does not need to constitute a security risk. But the software doing it needs check the validity of certificates and either block untrusted ones and inform the user or pass untrusted certificates straight to the browser for it to handle.
Then that should tell you you shouldn't be banking on the company network. Their network, their rules. Pretty sure they spelled that out when you started working there. Indeed, given the Verizon deal, since none of us lay down our own cabling direct to where we communicate on a regular basis, we are basically at the mercy of whoever provides the pipes.
There are lots of banks. There are even more providers for whom you are arguing should be blacklisted.
So who maintains the list? That list will get big for an enterprise running in 50 countries.
Much easier to just tell people it is not permitted and may be monitored. If you know the risk and do it anyway, that is your fault.
It depends on what country you are in but that could be illegal. In some countries your company is not allowed to MitM certain sites like banking, healthcare, etc. But SSL scanning is important - there's malware that uses HTTPS simply because so many organisations do not MitM it for malware scanning. SSL scanning on sites like dropbox, etc, is critical.
Their tools their rules.
If I want to create my own fake root cert and install it on my own box and inform anyone who uses my box that i can record any traffic going to the web, that is my prerogative.
I don't see any difference between me doing it for personal reasons and a company doing it for security reasons.
If you want to use a personal service, use hardware you own, not mine.
I still can't understand why this isn't passing off. IANALBIPOOTI and I believe that Reckitt & Colman Products Ltd v Borden Inc established that there are three criteria: the trader enjoys some goodwill; there is some misrepresentation by a third party; and the trader suffers damage to the aforementioned goodwill.
OK, so HSBC has some goodwill. Yes they launder drug money; help people evade taxes and charge you a fortune for incurring a small overdraft. But nevertheless, you trust them not to hand over your account to a third party for them to do as they will.
When you https:// to hsbc.co.uk, you are using their goods and services, to wit, their online banking facility. Part of that service is your assurance that you are connected to HSBC before you start typing the sort of stuff you really want to remain private. When you see the padlock in the address bar, you believe that you have a connection to an entity whose identity is assured by another entity that can be trusted to assure that identity. Any software that presents you with an MITM certificate for hsbc.co.uk signed by one of these dodgy outfits who have installed a bogus (yes I really think that's the word) root cert is surely passing off their own certificate as the certificate which has been presented to you by the organisation you think you are connected to?
But... would deleting all the certs (except the require MS ones) and starting over create problem? I just found a pile of Comodo certs.... <sigh> Not sure how they got there, but this does raise my curiosity level. I found these in both Firefox and IE... I also see them at my work place.
I seem to have 3 Comodo certificates. All 3 are "COMODO x Certification Authority"s, where x is either empty, "ECC" or "RSA".
I certainly wouldn't suggest deleting all of your certificates except the MS ones. Most of the certificates that you encounter online are signed by someone besides MS. I seem to have a lot more TRCAs than I remember (100-200 now) and a lot of them look particular to certain countries. (Just glancing through, I see a Greek one, a Finnish one and a couple Chinese ones.) If you're really worried about having some bad certificates, be sure to export them before deleting them. That way when you discover that you really needed one of those, you can add it back.
This post has been deleted by its author
Hmmm, no comments above about Comodo being hacked a couple of years back when the perps made off with amongst other things, an unknown number of legal SSL Certs, which have been used to by pass browser security.
Did they ever find out who did this ?
Here they go Comodo issuing useless certs ?
Comodo inserting adds !
Why on earth anyone is still using the company and their products defeats my imagination.