6 ms·

A fellow from Kazakhstan here. Banning this certificate or at least warning the users against using it WILL help a lot. Each authoritarian regime is authorita

by nurbo 7y ago

A fellow from Kazakhstan here. Banning this certificate or at least warning the users against using it WILL help a lot. Each authoritarian regime is authoritarian in its own way. Kazakhstan doesn't have a very strong regime, especially since the first president resigned earlier this year. When people protest strongly against something, the government usually backs down. For example, a couple of years ago the government withdrew their plans of lending lands to foreign governments after backlash from ordinary people. If Kazakhs knew about the implications of installing this certificate, they would have been on the streets already. If Firefox, Chrome and/or Safari block this certificate, the people will show their dissatisfaction and the law will be revoked. Sometimes the people in authoritarian countries need a little bit of support from organizations to fight for their rights. I really hope the browser organizations would help us here.

ftl64 7y ago
Couldn't agree more.
mikorym 7y ago
What is the extent of the MITM attack that you can do with this certificate? Can you intercept all https traffic?
- noselasd 7y ago
  Yes
- alerque 7y ago
  Yes, having a root CA certificate like this installed in a client allows the certificate issuer (so in this case KZ government and anything they authorize ISPs to do) to impersonate any and every other domain. So yes, ALL https traffic to and from that client to be subject to intercept.
  - amelius 7y ago
    Why does such a certificate exist in the first place?
    - darklajid 7y ago
      Because it takes nothing but an openssl installation (or similar) to generate one? Anyone can control a (root) certificate - the problem is getting others to trust it. Legitimate use cases might be: You want to intercept (and decrypt) traffic going from your local computer to SSL/TLS endpoints (affects only you) for example. Less clear cut / nice example: Company wants to read your traffic and therefor deploys a cert like this on your computer, now can snoop on anything you do, https or not.
    - alerque 7y ago
      Anybody can make one. You can make one on your own computer, give the result to your friends, tell them to connect through you as a proxy, and intercept everything. The tricky part is that browsers are hard coded with a list of a few trusted root certificates to trust. In order for the home baked certificate you just made to do any good, people have to explicitly install it and mark it as trusted. That means you have to distribute your newly minted root certificate and get every end point device to accept it manually. That's what's so sinister about Kazakhstan's approach: by issuing a governmental mandate for citizens install the certificate they generated, and restricting their internet if they don't, they are effectively bypassing the Internet's current trust system entirely and granting themselves cart-blanche access to all their traffic.
    - rndgermandude 7y ago
      It exists to intercept https and potentially other TLS traffic. It exists because everybody can make such a certificate. I made such a CA certificate for my personal use, not to MITM myself, but to issue certificates for some internal services that are out of scope of letsencrypt. Every major desktop OS comes with tools that let you make a CA certificate, Windows does, macos does, linux distro usually ship openssl/gnutls/nss tools (as installable packages). The challenge is not to make it but to get it trusted by OS and software. The Kazakhstan government solved it by having the ISPs just tell people to install the thing themselves into each and every device you own. Why does the government want this? To snoop on people. Usually framed as "We need to be able to fight terrorists, criminals and/or foreign enemies who 'abuse' encryption to hide their malicious activities". Tho, a lot of times the government will say all people are potential terrorists, and you just don't know if they are until you start snooping on them. It's not only a thing with just authoritarian regimes, either. Australia passed a law which basically forces Australian companies and citizens to add backdoors in any products using end-to-end encryption (thereby effectively disabling end-to-end encryption) so the government can read communication if they want to. The UK has a law ("snooper charter") that requires companies to "remove or disable" encryption when the government shows up with a warrant. The US similarly are looking into end-to-end encryption busting legislation. And they already compelled companies to effective disable encryption systems, e.g. when a judge ordered lavabit (then the email provider Edward Snowden used) to hand over their encryption keys and install a government provided device capable of logging all traffic. And let's not forget that for a long time US law classified strong encryption as a "weapon" which meant you could not export encryption easily. Or the NSA e.g. pushing their backdoor encryption-busting PRNG (Dual_EC_DRBG) and weak encryption schemes (Speck, Simon). German politicians recently started demanding end-to-end encryption busting legislation too, except they said "we do not want to make encryption weaker or insecure, we just want that the companies give us the plaintext data", which once more shows that they didn't thought it was necessary to do the most basic research into how this stuff works before talking.
- tialaramex 7y ago
  If a user trusts this root CA (~ "installs the certificate") then someone who controls the root can now make their MITM look like the real deal, because it's trusted. After all you've said you trust them. Whether you _should_ trust the authoritarian government of Kazakhstan is a policy issue. On its own the root does not magically intercept the traffic, so Kazakh ISPs will need to do a bunch of (potentially quite expensive) work to actually MITM traffic for the Kazakh government, but with the root once that work is done it doesn't get flagged as a problem. Because this exact same strategy (root that is not trustworthy is installed) is used in corporate setups to do anti-exfiltration, porn filtering and dozens of other things of dubious value, browsers are designed to let you, or the computer's administrator, choose to trust root CAs and indeed lots of counter-measures that protect ordinary users from bad guys are deliberately _disabled_ in the scenario where you've told it to trust some third party. You know best. If you imagine a hypothetical system which just doesn't trust this root, say somebody has a Raspberry Pi they smuggled across the border, or more prosaically, they just said "No" and refused to install the root certificate -- such a system just will treat the MITM as an error, your secure web browsing won't work because it can't make a secure connection. Or contrariwise, suppose you install the root in an otherwise ordinary PC in New York connected to AT&T, it will have no effect because the Kazakh government obviously isn't in New York MITMing your connections to other stuff.
  - neves 7y ago
    How do I know that NSA does not have one root certificate pre-installed in my browser?
    - corndoge 7y ago
      Review and compile the browser yourself, or just trust that someone would have found it by now and trust that your browser vendor knows that and would never do it in the first place unless they wanted to kill their browser instantly
    - _nickwhite 7y ago
      The NSA doesn't need a root certificate installed on your computer if they already have the private keys to the root and/or intermediate CA providers.
jacquesm 7y ago
I wished that all individuals in all countries would have such an attitude towards their governments.
JoshTriplett 7y ago
Please do post this feedback in the bugzilla bug and the linked discussion thread; this is the kind of thing that helps developers make a more informed decision rather than just speculating on what would help people more.
vbezhenar 7y ago
I really don't like the idea that some third forces would interfere with internal politics of my country. Browser should work according to technical standards, not according to what US citizens decided to be good or bad. If Firefox wants to forbid locally installed roots, I'm all for it, but implement it for everyone. That said, I don't see how government would step back. People are uninformed and generally passive, they wouldn't care enough. So, sadly, it might be the only way to push back that decision. But I still don't like it.
- oytis 7y ago
  Privacy is not 'what US citizens decided to be good', it's fundamental human right. I'm not a Kazakh, I'm Russian, but our politics has a lot in common. And I would say, third forces interfering with politics is not as bad (if at all), as local governments interfering with people's privacy.
- pornel 7y ago
  Encryption is political. The technical standards around HTTPS are politically motivated. The push for HTTPS everywhere, de-facto required TLS in HTTP/2, eSNI & DoH, were largely a response to the US government's mass surveillance. However, who makes these changes is interesting. It happened to be mostly developers in Five Eyes' countries acting against Five Eyes. Do we need to have elected representatives in browser vendors and encryption standards bodies? Or given that the elected representatives are for mass surveillance, would that be any good?
  - martingxx 7y ago
    I disagree that encryption is political. Fundamentally, it's a privacy and security mechanism and on its own it's no more political than locks, safes, paper shredders or curtains. Because of the complex, un-intuitive nature of encryption, it mixed particularly badly with politics, and we're still suffering from the fallout of that now. (crypto wars 1) Firefox and other application vendors who use those standards do end up being unavoidably slightly closer to politics (as demonstrated in this particular issue now), but I think Mozilla would do well to keep their goal simple - Protect user privacy where they can and explain to users when they can't. A notice to the user like "You are using a key known to allow access to third parties" is just a fact and no more political than "the site you are visiting uses weak crypto standards" or similar.
winter_blue 7y ago
> If Firefox, Chrome and/or Safari block this certificate For users who are already subject to the MITM, when they try to download a new version of their browser, couldn't the ISP simply serve them a slightly modified version of the new/updated Chrome/Firefox binary which does not blacklist their certificate? It would require a high level of technical expertise, but what could prevent this sort of thing?
- maxheadroom 7y ago
  >...but what could prevent this sort of thing? If it's on Windows, probably the digital signature[0] of the installer bundle will flag for invalid/unknown signature. For nix you could verify checksums but for Apple, if it's not in the store, I think it's nigh* impossible to do. [0] - https://docs.microsoft.com/en-us/windows/uwp/packaging/sign-app-package-using-signtool https://docs.microsoft.com/en-us/windows/uwp/packaging/sign-...
reeves23423 7y ago
In the meanwhile consider using Tor https://www.torproject.org https://www.torproject.org . It has different transports transport plugins available if that will make the traffic look like regular traffic and not Tor traffic.
- noragami 7y ago
  Tor has been blocked here for the last 10 (or so) years. You have to find and add a couple of bridge servers to get it working, until they get blocked as well, of course.
  - tgragnato 7y ago
    This is correct, but there are a couple of caveats. As a reference - relays https://metrics.torproject.org/userstats-relay-country.html?start=2016-01-01&end=2019-07-19&country=kz&events=on https://metrics.torproject.org/userstats-relay-country.html?... - bridges https://metrics.torproject.org/userstats-bridge-combined.html?start=2016-01-01&end=2019-07-19&country=kz https://metrics.torproject.org/userstats-bridge-combined.htm... - ticket https://trac.torproject.org/projects/tor/ticket/20348 https://trac.torproject.org/projects/tor/ticket/20348 Noteworthy "I messaged the user my unpublished obfs4 bridge whose OR port is firewalled and the bridge is working so far". Also, the boxes don't seem to perform a good job in inspecting for protocols tunnelled over HTTPS (https://gitweb.torproject.org/pluggable-transports/httpsproxy.git/tree/README.md https://gitweb.torproject.org/pluggable-transports/httpsprox...).
gnull 7y ago
I totally agree with your point. Blocking this certificate in major browsers or warning users would be of great use. But there is a minor remark I have to make here. I'm sure you were talking about citizens of Kazakhstan, when you used the word "Kazakhs". The word "Kazakh" actually denotes an ethnicity, not a citizenship. And although Kazakhs are predominant ethnicity in Kazakhstan (~65% of the population), there are many others, and it's incorrect to call them Kazakhs. Wikipedia suggests "Kazakhstani" as a English term for Kazakhstan citizens. I also saw other people using the word "Kazakhstanians". Maybe one of these two would be a better choice than "Kazakhs".
- wnscooke 7y ago
  But he actually could have been talking primarily about Kazakhs in Kazakhstan, so nothing wrong there.
- huttarla 7y ago
  While technically this is true, it is rare (and sounds nerdy) for English speakers who are familiar with Kazakhstan to actually use the word "Kazakhstani" or especially "Kazakhstanian." The word "Kazakh" is used informally to refer to natives or citizens of Kazakhstan when the meaning is clear from the context. Similar to how "Americans" (which could arguably refer to any North or South American) is used instead of "U.S. citizens." Or "Saudis." At least that was my experience from living in K'stan for 4 years in the '90's.
  - gnull 7y ago
    I didn't know that. I'm not a native English speaker. If this is actually an established way to informally address Kazakhstan citizens in English, then it's OK, I do not object. But I'm still skeptical about it. I think in my own speech I'll stick to "Kazakhstanian" (despite the word "Kazakh" is so much easier to type). Because I'm just used to make distinction between words "Kazakh" and "Kazakhstanian" when I speak Kazakh and Russian languages (most popular ones in Kazakhstan).

by nurbo 7y ago

ftl64 7y ago

Couldn't agree more.

mikorym 7y ago

What is the extent of the MITM attack that you can do with this certificate? Can you intercept all https traffic?

noselasd 7y ago
Yes
alerque 7y ago
Yes, having a root CA certificate like this installed in a client allows the certificate issuer (so in this case KZ government and anything they authorize ISPs to do) to impersonate any and every other domain. So yes, ALL https traffic to and from that client to be subject to intercept.
- amelius 7y ago
  Why does such a certificate exist in the first place?
  - darklajid 7y ago
    Because it takes nothing but an openssl installation (or similar) to generate one? Anyone can control a (root) certificate - the problem is getting others to trust it. Legitimate use cases might be: You want to intercept (and decrypt) traffic going from your local computer to SSL/TLS endpoints (affects only you) for example. Less clear cut / nice example: Company wants to read your traffic and therefor deploys a cert like this on your computer, now can snoop on anything you do, https or not.
  - alerque 7y ago
    Anybody can make one. You can make one on your own computer, give the result to your friends, tell them to connect through you as a proxy, and intercept everything. The tricky part is that browsers are hard coded with a list of a few trusted root certificates to trust. In order for the home baked certificate you just made to do any good, people have to explicitly install it and mark it as trusted. That means you have to distribute your newly minted root certificate and get every end point device to accept it manually. That's what's so sinister about Kazakhstan's approach: by issuing a governmental mandate for citizens install the certificate they generated, and restricting their internet if they don't, they are effectively bypassing the Internet's current trust system entirely and granting themselves cart-blanche access to all their traffic.
  - rndgermandude 7y ago
    It exists to intercept https and potentially other TLS traffic. It exists because everybody can make such a certificate. I made such a CA certificate for my personal use, not to MITM myself, but to issue certificates for some internal services that are out of scope of letsencrypt. Every major desktop OS comes with tools that let you make a CA certificate, Windows does, macos does, linux distro usually ship openssl/gnutls/nss tools (as installable packages). The challenge is not to make it but to get it trusted by OS and software. The Kazakhstan government solved it by having the ISPs just tell people to install the thing themselves into each and every device you own. Why does the government want this? To snoop on people. Usually framed as "We need to be able to fight terrorists, criminals and/or foreign enemies who 'abuse' encryption to hide their malicious activities". Tho, a lot of times the government will say all people are potential terrorists, and you just don't know if they are until you start snooping on them. It's not only a thing with just authoritarian regimes, either. Australia passed a law which basically forces Australian companies and citizens to add backdoors in any products using end-to-end encryption (thereby effectively disabling end-to-end encryption) so the government can read communication if they want to. The UK has a law ("snooper charter") that requires companies to "remove or disable" encryption when the government shows up with a warrant. The US similarly are looking into end-to-end encryption busting legislation. And they already compelled companies to effective disable encryption systems, e.g. when a judge ordered lavabit (then the email provider Edward Snowden used) to hand over their encryption keys and install a government provided device capable of logging all traffic. And let's not forget that for a long time US law classified strong encryption as a "weapon" which meant you could not export encryption easily. Or the NSA e.g. pushing their backdoor encryption-busting PRNG (Dual_EC_DRBG) and weak encryption schemes (Speck, Simon). German politicians recently started demanding end-to-end encryption busting legislation too, except they said "we do not want to make encryption weaker or insecure, we just want that the companies give us the plaintext data", which once more shows that they didn't thought it was necessary to do the most basic research into how this stuff works before talking.
tialaramex 7y ago
If a user trusts this root CA (~ "installs the certificate") then someone who controls the root can now make their MITM look like the real deal, because it's trusted. After all you've said you trust them. Whether you _should_ trust the authoritarian government of Kazakhstan is a policy issue. On its own the root does not magically intercept the traffic, so Kazakh ISPs will need to do a bunch of (potentially quite expensive) work to actually MITM traffic for the Kazakh government, but with the root once that work is done it doesn't get flagged as a problem. Because this exact same strategy (root that is not trustworthy is installed) is used in corporate setups to do anti-exfiltration, porn filtering and dozens of other things of dubious value, browsers are designed to let you, or the computer's administrator, choose to trust root CAs and indeed lots of counter-measures that protect ordinary users from bad guys are deliberately _disabled_ in the scenario where you've told it to trust some third party. You know best. If you imagine a hypothetical system which just doesn't trust this root, say somebody has a Raspberry Pi they smuggled across the border, or more prosaically, they just said "No" and refused to install the root certificate -- such a system just will treat the MITM as an error, your secure web browsing won't work because it can't make a secure connection. Or contrariwise, suppose you install the root in an otherwise ordinary PC in New York connected to AT&T, it will have no effect because the Kazakh government obviously isn't in New York MITMing your connections to other stuff.
- neves 7y ago
  How do I know that NSA does not have one root certificate pre-installed in my browser?
  - corndoge 7y ago
    Review and compile the browser yourself, or just trust that someone would have found it by now and trust that your browser vendor knows that and would never do it in the first place unless they wanted to kill their browser instantly
  - _nickwhite 7y ago
    The NSA doesn't need a root certificate installed on your computer if they already have the private keys to the root and/or intermediate CA providers.

jacquesm 7y ago

I wished that all individuals in all countries would have such an attitude towards their governments.

JoshTriplett 7y ago

Please do post this feedback in the bugzilla bug and the linked discussion thread; this is the kind of thing that helps developers make a more informed decision rather than just speculating on what would help people more.

vbezhenar 7y ago

I really don't like the idea that some third forces would interfere with internal politics of my country. Browser should work according to technical standards, not according to what US citizens decided to be good or bad. If Firefox wants to forbid locally installed roots, I'm all for it, but implement it for everyone. That said, I don't see how government would step back. People are uninformed and generally passive, they wouldn't care enough. So, sadly, it might be the only way to push back that decision. But I still don't like it.

oytis 7y ago
Privacy is not 'what US citizens decided to be good', it's fundamental human right. I'm not a Kazakh, I'm Russian, but our politics has a lot in common. And I would say, third forces interfering with politics is not as bad (if at all), as local governments interfering with people's privacy.
pornel 7y ago
Encryption is political. The technical standards around HTTPS are politically motivated. The push for HTTPS everywhere, de-facto required TLS in HTTP/2, eSNI & DoH, were largely a response to the US government's mass surveillance. However, who makes these changes is interesting. It happened to be mostly developers in Five Eyes' countries acting against Five Eyes. Do we need to have elected representatives in browser vendors and encryption standards bodies? Or given that the elected representatives are for mass surveillance, would that be any good?
- martingxx 7y ago
  I disagree that encryption is political. Fundamentally, it's a privacy and security mechanism and on its own it's no more political than locks, safes, paper shredders or curtains. Because of the complex, un-intuitive nature of encryption, it mixed particularly badly with politics, and we're still suffering from the fallout of that now. (crypto wars 1) Firefox and other application vendors who use those standards do end up being unavoidably slightly closer to politics (as demonstrated in this particular issue now), but I think Mozilla would do well to keep their goal simple - Protect user privacy where they can and explain to users when they can't. A notice to the user like "You are using a key known to allow access to third parties" is just a fact and no more political than "the site you are visiting uses weak crypto standards" or similar.

winter_blue 7y ago

> If Firefox, Chrome and/or Safari block this certificate For users who are already subject to the MITM, when they try to download a new version of their browser, couldn't the ISP simply serve them a slightly modified version of the new/updated Chrome/Firefox binary which does not blacklist their certificate? It would require a high level of technical expertise, but what could prevent this sort of thing?

maxheadroom 7y ago
>...but what could prevent this sort of thing? If it's on Windows, probably the digital signature[0] of the installer bundle will flag for invalid/unknown signature. For nix you could verify checksums but for Apple, if it's not in the store, I think it's nigh* impossible to do. [0] - https://docs.microsoft.com/en-us/windows/uwp/packaging/sign-app-package-using-signtool https://docs.microsoft.com/en-us/windows/uwp/packaging/sign-...

reeves23423 7y ago

In the meanwhile consider using Tor https://www.torproject.org https://www.torproject.org . It has different transports transport plugins available if that will make the traffic look like regular traffic and not Tor traffic.

noragami 7y ago
Tor has been blocked here for the last 10 (or so) years. You have to find and add a couple of bridge servers to get it working, until they get blocked as well, of course.
- tgragnato 7y ago
  This is correct, but there are a couple of caveats. As a reference - relays https://metrics.torproject.org/userstats-relay-country.html?start=2016-01-01&end=2019-07-19&country=kz&events=on https://metrics.torproject.org/userstats-relay-country.html?... - bridges https://metrics.torproject.org/userstats-bridge-combined.html?start=2016-01-01&end=2019-07-19&country=kz https://metrics.torproject.org/userstats-bridge-combined.htm... - ticket https://trac.torproject.org/projects/tor/ticket/20348 https://trac.torproject.org/projects/tor/ticket/20348 Noteworthy "I messaged the user my unpublished obfs4 bridge whose OR port is firewalled and the bridge is working so far". Also, the boxes don't seem to perform a good job in inspecting for protocols tunnelled over HTTPS (https://gitweb.torproject.org/pluggable-transports/httpsproxy.git/tree/README.md https://gitweb.torproject.org/pluggable-transports/httpsprox...).

gnull 7y ago

I totally agree with your point. Blocking this certificate in major browsers or warning users would be of great use. But there is a minor remark I have to make here. I'm sure you were talking about citizens of Kazakhstan, when you used the word "Kazakhs". The word "Kazakh" actually denotes an ethnicity, not a citizenship. And although Kazakhs are predominant ethnicity in Kazakhstan (~65% of the population), there are many others, and it's incorrect to call them Kazakhs. Wikipedia suggests "Kazakhstani" as a English term for Kazakhstan citizens. I also saw other people using the word "Kazakhstanians". Maybe one of these two would be a better choice than "Kazakhs".

wnscooke 7y ago
But he actually could have been talking primarily about Kazakhs in Kazakhstan, so nothing wrong there.
huttarla 7y ago
While technically this is true, it is rare (and sounds nerdy) for English speakers who are familiar with Kazakhstan to actually use the word "Kazakhstani" or especially "Kazakhstanian." The word "Kazakh" is used informally to refer to natives or citizens of Kazakhstan when the meaning is clear from the context. Similar to how "Americans" (which could arguably refer to any North or South American) is used instead of "U.S. citizens." Or "Saudis." At least that was my experience from living in K'stan for 4 years in the '90's.
- gnull 7y ago
  I didn't know that. I'm not a native English speaker. If this is actually an established way to informally address Kazakhstan citizens in English, then it's OK, I do not object. But I'm still skeptical about it. I think in my own speech I'll stick to "Kazakhstanian" (despite the word "Kazakh" is so much easier to type). Because I'm just used to make distinction between words "Kazakh" and "Kazakhstanian" when I speak Kazakh and Russian languages (most popular ones in Kazakhstan).