Slashdot Mirror
Researchers Find Same RSA Encryption Key Used 28,000 Times
itwbennett writes In the course of trying to find out how many servers and devices are still vulnerable to the Web security flaw known as FREAK, researchers at Royal Holloway of the University of London found something else of interest: Many hosts (either servers or other Internet-connected devices) share the same 512-bit public key. In one egregious example, 28,394 routers running a SSL VPN module all use the same 512-bit public RSA key.
Wow
This is a real problem and I don't mean to minimize it. But weak encryption is infinitely better than none, and the solution to this is immensely easier than the solution to the many, many wholly unencrypted connections that are happening this very moment. I think we should prioritize getting all connections everywhere encrypted somehow.
If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
The encryption key is YouwereownedbythemotherfuckinNSAbitchandyourassesbelongtousnow.
Then it deserves at least social shaming and ostracism, if not worse than those minor responses to venial sins. Protecting the manufacturers only creates an environment where the incentives are aligned for them to do it again. If manufacturers aren't keenly aware that they need to protect their reputation, then they will cut every corner that doesn't provide them a competitive advantage.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
there is no such thing as security any more using the common models and parameters. got to step it up, without fallback to silliness like 512 bit keys. the bigger problem is nobody has been bankrupted and sent to jail yet, so the impetus is not there to fix it as the first priority of business.
if this is supposed to be a new economy, how come they still want my old fashioned money?
Recycling is a good thing, you earth rapers.
Similar mistake have been made before.
Have gnu, will travel.
First line of the article:
"What if the key to your house was shared with 28,000 other homes?"
The fact is, you very well might share the key to your house with more than 28000 other homes. Common lock brands you can buy at Home Depot, Lowe's, etc. create a surprisingly low number of different key/tumbler combinations.
Popisms.com - Connecting pop culture
*.myfoscam.org/organizationName=ShenZhen Foscam Intelligent Technology Co,Ltd
Seems to be a network enabled camera.
Create a function to generate a random key.
All commercial software is backdoored. They can't very well put in an additional user called backdoor, can they? So they create specification ambiguities, add bugs and make insecure default configurations. "But the need to create unique keys was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying 'Beware of the Leopard', so using the default configuration is your fault."
If you're looking to Open Source software to get around that, you will very likely end up using software cobbled together just well enough that it works, with next to no attention paid to actual security. If something looks encrypted, that's good enough, and you're lucky if an author of the myriads of tutorials on the internet even checked with a packet sniffer that things actually look encrypted. Download this virtual machine image and you've got yourselves a server, complete with static SSH host keys, because hey, it works and it's encrypted, isn't it? Authy, a two-factor authentication service, allowed anyone to log in with "../sms", because they relied on "rack-protection", an open source Ruby project which supposedly "protects against typical web attacks", which failed to properly escape user input. A face-palm is in order. If you believe that's just an anecdote and not symptomatic of the general state of security on the internet, you're deluding yourself.
There is no security with computers, because nobody really cares. It's all lip service. Computer geeks are just very slow to get this. To most people, security is a source of problems, not a solution. This is not going to get better. Young people all use computers and on average know less about how computers actually work than the people who were using them when computers and networks were new.
Comment removed based on user account deletion
Ahem. Unlike you, manufacturers can get keys which can sign other keys. Which means they can buy 1 key and generate as many sub-keys as they like.
Putting 1 key into a image vs having a different key go into every image.
One is cheap, the other isn't. (when pennies count).
These are not certificates. They're not validated by any trusted authority. These are host keys: you generate them yourself for the cost of electricity. You could have your router generate its own keys the first time it starts up for the cost of a couple seconds delay.
You could use your own CA and generate self signed certificates.
Except they don't have to buy them at all. They can just generate them.
Hell, *anyone* can generate random/unique keys; the only reason we buy certs is for the "trust" part of PKI, not the encryption part. Many (most?) protocols don't have a concept of "trusted" certs, so there's no benefit to buying one from a CA.
You do realize these are self signed keys? They are trivial to generate and have no cost.
No sir I dont like it.
Have a list of ALL keys ever generated, similar to the windows key service. Then you can make sure each device has its own key when manufacturing devices. Just kidding, that would require a desire to actually sell securely encrypted devices.
Security is really only for those corps and individuals alike who proactively ensure their own security. On the scale of things this is somewhat of a small #. Until the general masses embrace security as a whole, their will be none. Just the constant drivel of stories like this one.
I can hardly sit still with the desire to root my Z3 phone just for the purpose of security.
Yet another example of "security" journalism that fails to mention the name of the manufacturer. Just what junk am I supposed to throw away? Details people, details.
People are lazy and suffer from "it can't happen to me" and that's assuming they even understand what's going on. I just sent some very sensitive documents by email today encrypted with AES and a strong password. I let the financial institution on the other end know the password by SMS to the loan officer's phone. Two minutes later the idiot forwards the email to his assistant with the attached PDF and the puts the password in the body of the email. They didn't even understand what was going on when I got upset. They genuinely think I'm on the level of UFO conspiracy nuts for encrypting scans of my social security card, driver's license, paystubs, and bank statements to be sent over the Internet.
All the security problems reside in the Wetware.
If we could only take the Wetware out of the equation, we could have real and effective security.
Damn the Wetware!
You could have your router generate its own keys the first time it starts up for the cost of a couple seconds delay.
Not really. You need randomness to make a good key. Exactly how much randomness is an unconfigured router going to be able to collect in the first few seconds of it's initial power-on? Some, sure, but preferentially you want a *lot*.
Having said that, if you extended it to a couple of minutes then it should have ample time to collect enough randomness - packet arrival timings, local MACs it can see, other SSIDs it can see; perhaps the SoC even has a hardware random noise generator - to create a key that's not too non-random.
http://www.reuters.com/article... NSA ~ RSA
We should learn what we need to know about issues, before we decide what we need to feel about them.
Without SALT even encryption is bland. Try searching encrypted password hashes on google it's always interesting to see who you share the same password word. Often the userid is revealed in the search, if the userid is unique you can search the userid out and find new sites to try your new acquired user:pass combo's.
If you used your own CA, why would you bother with self-signed certificates?!
And many embedded SoC have atleast one of the following:
- HW random generator - Perfect... for more security throw in a few of the below to strengthen it...
- Embedded encryption engine with a device-specific key (set in the actual SoC during production) - Perfect... we do not really need anything else...
- A/D in one way or another.. use some static from it..
- Wifi device - other wifi devices and signal-strengths and timing between beacon-packets.
- Network interfaces - number of them, timing between packets, timing before getting link etc..
- flash-memory - may have serial-number and may have a random distribution of bad blocks.
- unique serial-number - Usually deterministic but *usually* hard to get information about remotely.
- Several clock-sources.. Use them together and be able to get some clock-drift info to throw into the pool of "random" data..
- Time from boot until a browser connected.
- Anything the client-browser throws at the router (client-ip, user-agent, registered plugins etc...)
- MAC address - yea, not really good by itself but hey... we just want as much data as possible.. and usually only possible to know if you are in range of the wifi or on the same subnet.
- IP / Netmask / gateway / gateway mac / other network-config stuff...
Throw one or more of these into a scrypt 100-1000 times and then do a sha-512 and use that as the seed for the PRNG.. Using scrypt would just be to make it much harder to guess the key if you knew the approximate values that where used for the seed initially.
Even if you would use just the MAC address for this it would be a *BIG* jump in security over using a shared certificate... But i would recommend to use at least 2 pseudo-random things from that list, preferably 4-5, before it would become fairly secure..
Once I know the key is good, I stick with it and use it everywhere. Why to mess with it if it's working?
The public keys might be identical but that does not mean that the private keys are also all the same. So the resulting encryption could well always be different despite the identical public keys.
You can make things foolproof, but you can't make them idiot proof.
Most people doing anything on the internet have no functional literacy in security. WEP, WPA, SSL, https,...it's all alphabet soup mumbo jumbo to most people.
Now, some self-appointed expert is going to chime in in a few moments and say that these ignorant fools need to educate themselves about this, and if they get pwned it's all their own fault.
The problem *is* that people need to be educated, but right now to truly understand the rudiments of security technology and the risks probably requires some night classes more suited to IT professionals. If that is the burden on the user, security will never work.
People understand locks and keys on their cars and doors, and know the consequences of not using them. The same can't be said for internet security. Most people are stuck relying on the kindness of strangers to implement security for them, and that is just asking for trouble.
The person who solves this problem, of making internet security understandable by the lay user, will be doing the world a great service.
Left MS Windows for Linux Mint and never looked back!
Vote for Bernie in 2016!
You're missing the point. Default certificates are self-signed and cost nothing. Not having unique ones is laziness and nothing else.
For use in a production environment, any WAN-facing appliance should get a unique cert ideally, but how may WAN-facing appliances do you need? I would venture a suggestion that only the VPN server and WEB server need to be WAN-facing. Everything else can be behind the VPN. So unless your company has 28394 offices, the cost is minimal. And for any company that DOES have 28394 offices, 78k/month may not be too high a price to pay for security.
How would you make a list of what are supposed to be private keys?