One in Five Websites Still Use Outdated SHA-1 Encryption Algorithm

[Zarathustra[H]]

SHA-1 has been known to be potentially insecure since 2005, but this wasn't proven in practice, at least not publicly, until Google recently announced they executed a successful collision attack, breaking it. The collision attack is notable as it takes about one 100,000th of the time to crack SHA-1 as it would with a full brute force attack.

The Register is reporting that one in five websites out there are still relying on SHA-1, despite its known vulnerabilities. This is an improvement over November last year when a third of all sites were relying on SHA-1.

Google, Microsoft and Mozilla have all set deadlines for early 2017 for websites to migrate away from SHA-1. It is unclear exactly what actions will be taken at that time, but presumably browsers will start issuing security warnings, and sites using SHA-1 will be down-ranked in search results.

This does all need to be put into perspective though. Even with Googles collision attack, it requires approximately 110 GPU's to break it in about a years time, so your average Russian script-kiddie isn't going to be doing it any time soon, even if he is lucky enough to have a few Pascal Titan's. It does, however, mean that SHA-1 is vulnerable to state actors with access to supercomputing resources, and potentially people with some money looking to rent supercomputer time. They would have to try to hide their true goals, but in a world where rented AWS clusters have been used for massive DDOS attacks, this doesn't seem infeasible.

Kevin Bocek, chief security strategist for Venafi, commented: "Even though most organisations have worked hard to migrate away from SHA-1, they don't have the visibility and automation necessary to complete the transition. We've seen this problem before when organisations had a difficult time making co-ordinated changes to keys and certificates in response to Heartbleed, and unfortunately I'm sure we are going to see it again."

 

[ZeqOBpf6]

One in five, but which one in five? I wouldn't be surprised if 20% of all websites account for less than a tenth of a percent of all traffic.

 

[Gigus Fire]

isn't md5 a hashing algorithm and not an encryption method? Why put it on the image as 30 seconds on a smart phone when it's disingenuous to compare that to encryption algorithms?

 

[Zarathustra[H]]

isn't md5 a hashing algorithm and not an encryption method? Why put it on the image as 30 seconds on a smart phone when it's disingenuous to compare that to encryption algorithms?

That was googles work, not mine. I thought it was a little odd too, but I assumed it was relevant. Maybe if you can circumvent MD5 you can use it to trick systems into accepting an altered file, and that - in some way - is relevant to the security discussion here?

 

[SvenBent]

isn't md5 a hashing algorithm and not an encryption method? Why put it on the image as 30 seconds on a smart phone when it's disingenuous to compare that to encryption algorithms?

That was googles work, not mine. I thought it was a little odd too, but I assumed it was relevant. Maybe if you can circumvent MD5 you can use it to trick systems into accepting an altered file, and that - in some way - is relevant to the security discussion here?

SHA-1 is a hash too
https://en.wikipedia.org/wiki/SHA-1

They are both in the same catagory of cryptographic hash functions
https://en.wikipedia.org/wiki/List_of_hash_functions

 

[Ryokurin]

isn't md5 a hashing algorithm and not an encryption method? Why put it on the image as 30 seconds on a smart phone when it's disingenuous to compare that to encryption algorithms?

SHA-1 is encryption, but it is also used as a hash. It's full name is Secure Hash Algorithm 1.

When a secure connection is made using a SSL certificate, it's hash is also checked, that can be with MD5, SHA-1, (but neither has been suggested for that purpose for a very long time), SHA-256, etc to verify that the certificate has actually been signed by the correct authority.

 

[Zarathustra[H]]

SHA-1 is a hash too
https://en.wikipedia.org/wiki/SHA-1

They are both in the same catagory of cryptographic hash functions
https://en.wikipedia.org/wiki/List_of_hash_functions

SHA-1 is encryption, but it is also used as a hash. It's full name is Secure Hash Algorithm 1.

When a secure connection is made using a SSL certificate, it's hash is also checked, that can be with MD5, SHA-1, (but neither has been suggested for that purpose for a very long time), SHA-256, etc to verify that the certificate has actually been signed by the correct authority.

Thanks for that info. I am definitely a layman when it comes to encryption techniques.

 

[lcpiper]

One in five, but which one in five? I wouldn't be surprised if 20% of all websites account for less than a tenth of a percent of all traffic.

Exactly.

You tell me one in five online banking websites still use SHA-1 you'll raise an eyebrow. Shit, one in five don't even offer a user a chance to build an account and aren't selling anything so ... "Which ones is entirely pertinent.

 

[Rahh]

TLS 1.0 is also now unsupported and with methods to crack it out as well. Bundle sha and tls 1.0 and see how many are still running that. It's pretty scary how many businesses are running on outdated servers and security. This is why we'll continue to see stories about cyber threats and hacks. Getting worse everyday until it's all government regulated...

 

[Zarathustra[H]]

Exactly.

You tell me one in five online banking websites still use SHA-1 you'll raise an eyebrow. Shit, one in five don't even offer a user a chance to build an account and aren't selling anything so ... "Which ones is entirely pertinent.

With people sharing passwords across sites, that could still be a problem.

Break SHA-1, get a plain text password, then use it to gain access to another, more critical site.

Yes, I know, this is a PEBCAK problem, not a problem of the system, but when 99% of users do the same PEBCAK thing, it still has a huge potential for disruption and damage, even for those who are smart about their passwords.

For instance, per federal law, the Bank is responsible for any losses involved with illicit access of a personal account. (business accounts are different and don't have this protection)

If you are a customer of a small bank, and many of the other customers wind up being victims of fraud due to password sharing, this can certainly impact you. The bank could wind up in financial trouble.

Or even a larger bank. The bank may be responsible legally for these losses, but they are not in the business of losing money. You bet this is factored into the cost in the form of fees and higher rates than it would be in a low fraud environment. Even a relatively small amount of clueless password sharing victims could be enough to impact the calculus of a large financial institution such that they feel the need to raise rates.

 

[lcpiper]

With people sharing passwords across sites, that could still be a problem.

Break SHA-1, get a plain text password, then use it to gain access to another, more critical site.

Yes, I know, this is a PEBCAK problem, not a problem of the system, but when 99% of users do the same PEBCAK thing, it still has a huge potential for disruption and damage, even for those who are smart about their passwords.

For instance, per federal law, the Bank is responsible for any losses involved with illicit access of a personal account. (business accounts are different and don't have this protection)

If you are a customer of a small bank, and many of the other customers wind up being victims of fraud due to password sharing, this can certainly impact you. The bank could wind up in financial trouble.

Or even a larger bank. The bank may be responsible legally for these losses, but they are not in the business of losing money. You bet this is factored into the cost in the form of fees and higher rates than it would be in a low fraud environment. Even a relatively small amount of clueless password sharing victims could be enough to impact the calculus of a large financial institution such that they feel the need to raise rates.

I'll say it again in case you missed it Z;

lcpiper said:

Shit, one in five don't even offer a user a chance to build an account and aren't

There are many many websites that don't even have customer user accounts, that is my point. These people claim one in five are still using SHA-1, but what if two in five don't even have users create an account? The user is simply consuming the info posted, as in a blog, but not creating accounts in order to interact beyond simple consumption?

It changes things greatly. Now I pulled that number out of my ass, it's just to make the point that not every website is a potential risk in this case.

 

[Zarathustra[H]]

I'll say it again in case you missed it Z;


There are many many websites that don't even have customer user accounts, that is my point. These people claim one in five are still using SHA-1, but what if two in five don't even have users create an account? The user is simply consuming the info posted, as in a blog, but not creating accounts in order to interact beyond simple consumption?

It changes things greatly. Now I pulled that number out of my ass, it's just to make the point that not every website is a potential risk in this case.

That is true. Googles announcement that they were de-prioritizing sites that don't use SSL a while back means that a lot of sites where SSL may not make sense, like strictly public information sites with no user interaction or accounts now have one form of SSL or another.

I hadn't thought of that. Still, there is a potential that many of these SHA-1 sites are more than just passive user consumption sites, as most sites have at least a comments section, or something like that these days, but without further data we have no way of knowing.

 

[lcpiper]

That is true. Googles announcement that they were de-prioritizing sites that don't use SSL a while back means that a lot of sites where SSL may not make sense, like strictly public information sites with no user interaction or accounts now have one form of SSL or another.

I hadn't thought of that. Still, there is a potential that many of these SHA-1 sites are more than just passive user consumption sites, as most sites have at least a comments section, or something like that these days, but without further data we have no way of knowing.

Maybe one of these days we'll have an OS and Browser with a "Security Widget", that has little "lights" that display the security posture of the connection to each page we have up and are accessing.

So when I go to the [H] I actually have a little panel that shows me all the details about my connection and not just a stupid "lock symbol".

 

[Zarathustra[H]]

Maybe one of these days we'll have an OS and Browser with a "Security Widget", that has little "lights" that display the security posture of the connection to each page we have up and are accessing.

So when I go to the [H] I actually have a little panel that shows me all the details about my connection and not just a stupid "lock symbol".

Sounds like a good idea for a browser plugin.

Do we have any programming entrepreneurs in here?

 

[SvenBent]

Maybe one of these days we'll have an OS and Browser with a "Security Widget", that has little "lights" that display the security posture of the connection to each page we have up and are accessing.

So when I go to the [H] I actually have a little panel that shows me all the details about my connection and not just a stupid "lock symbol".

i dont know if its my cyberfox or my httsp everywher plgun but i i click that little lock icon i get alle the security info. like encrypting and signing algo.
it even tell me what cookies right i gave the site.
or did you want it more out on the front of the browser ?

 

[lcpiper]

SventBent, that is exactly the right kind of info but I wish it were displayed in a dashboard style right from the browser's main page instead of something you have to go check.

Like a submarine's christmas tree;

SSLv1 X
SSLv2 X
SSLv3 O
TLS X


So as I am showing it, the connection is using SSLv3 but although I am configured for TLS, TLS is not being used, perhaps because the other site is not configured for it. If I wasn't configured for it maybe it would be grey or black. If the connection is pretty insecure, say using SSLv1 it shows up as a warning.

SSLv1 X
SSLv2 X
SSLv3 O
TLS X

Turn the Xs and Os into green, red, and yellow lights, etc and you get the idea.

Or, Fields that are labeled and show what the connection is.

CON TLSv1.2
ENCYP AES
KeyStr 128 bit

anyway;



etc