RSA Question

[Aldo]

Considering these retrospective accounts published in May 2021 about events in 2011:

The Full Story of the Stunning RSA Hack Can Finally Be Told

In 2011, Chinese spies stole the crown jewels of cybersecurity—stripping protections from firms and government agencies worldwide. Here’s how it happened.

wired.com

The Story of the 2011 RSA Hack - Schneier on Security

Really good long article about the Chinese hacking of RSA, Inc. They were able to get copies of the seed values to the SecurID authentication token, a harbinger of supply-chain attacks to come.

schneier.com

If anyone is now selling/promoting new service with RSA keys, is that legit? Are RSA key devices for logging in with one-time codes still- or newly-viable after that root breach back then?

 

[PHolder]

There is some confusion here. RSA corporation had a brand. RSA the protocol is generic. "RSA keys", of the branded hardware device, were broken back then and I think they've not gone back to that business. RSA keys of the security protocol kind are still very much in use. 1024 bit RSA keys are no longer considered secure enough, and 2048 bit ones are still in use but should probably soon be marked as not recommended. 4096 bit RSA keys are still completely secure for now, but probably will face risks if quantum computing comes of age.

 

[Aldo]

Thank you, this helps refine my question.

At first glance, if the current vendor of a device that uses the RSA security protocol fails to identify themselves as NOT RSA the corporation, then they are automatically suspect of taking advantage of the confusion that exists, and are thus not trustworthy of providing an actually-secure service.

OTOH websearch says SecurID "an RSA Business" is the corporation's current brand name for hardware token generators, software token generators, and various other methods of delivering "OTP authentication" ... leading me to think the company is yes in the business at this time.

Now I see in browsing history that a bank I use is offering an RSA SecurID device for $25, to provide "an additional layer of security [for] certain transactions and sensitive information ... to further verify your identity when accessing certain accounts or services. SecurID® is a registered trademark of RSA Security, Inc." So this marketing effort is yes properly identified, however this is a bank that I trust the least, among the banks I use.

In general I do yes want this level of 2FA security available from all financial accounts, to avoid the non-security that is inherent in relying on cellphones or email. However the hack-history (of RSA the company's implementation of RSA the protocol, I guess?) made me think twice. Like, if SecurID devices from 2012 were hackable in eighteen minutes back then, how are they better now - is it just stronger tactics via longer key length? Or is it something better because of using a different strategy?

As I understand it, the Rivest-Shamir-Adleman scheme relies on keeping digital secrets "up the chain". But secret numbers can become non-secret, as by leak or brute-force crack. As cracking compute-power increases, key length must increase also, to ensure the compute-time for cracking remains prohibitively long. This is how even 'perfect' encryption-etc algorithms have only a temporary window of usefulness, beyond which they become insecure, by virtue of crack-power having caught up with their key-length. Is this a reasonable way to explain it to people? Not too tech-y and not too long

 

[PHolder]

The premise with ANY of these 2FA type tools is that someone else holds a secret that matches the one in your device. For example, the Google Authenticator type 2FA is just a secret string that they show you at the beginning as a 2D barcode (QRCode.) If that secret is ever exposed (as was done with the hacking of the RSA corporation secrets) then you're basically totally hosed, from a security perspective. Yubikeys arrive with the secret preloaded into the key and also into their database, but you are relying on them to keep their database very secure. (There is a way to reload a new secret into the key and database at the same time, but of course you need to do that online and some may see that as a risk.)

There is no easy way to explain cryptographic strength. It depends on a lot of factors, not the least of which is how powerful is your "enemy". AES256 is so secure that it will generally NEVER be broken. I've done the math for many people many times before, but they always come back with "what if..." There is no what if. It's the law of large numbers. 2^256 is very, very, very large. There aren't that many atoms in the known universe. If you have the most powerful computers in the world, running forever on cracking ONE AES256 encrypted secret, it would take trillions of centuries or more.... it's just never happening.

It's more likely that espionage, or force, will be brought upon the target. You can get the secret from someone with force or money for a lot less effort than with computer password cracking.

The "difficulty" of RSA and other asymmetric key algorithms usually relies on the fact that there are problems that computers just can't currently solve efficiently. (Known as NP Hard problems... they take non-polynomial time to solve... and as the problem size grows (the key size) the size of the attack needed to solve the problem grows non-linerarily... i.e. exponentially.) There are charts you will find, that based on current state of the art, compare the effort between symmetric and asymmetric keys. These stem from the idea that usually bulk encryption is done with an symmetric key, and then that key is secured using asymmetric encryption. You might find this interesting in that regard: https://www.keylength.com/en/4/

 

[Lob]

The (company) RSA hack was actually a phish that exploited Excel IIRC that then got the bad guys to establish a foothold in the RSA company environment which then allowed the threat actors a presence in the environment, essentially a back door. They got to the "seed warehouse", the most trusted element of the whole SecurID universe.

Cleaning up from that would have been in the following order:
- rid the network of the presence
- set up a new "seed warehouse" (better protected than the last one)
- re-issue 752 million RSA SecurID tokens ($$$$$$$!!)
- hope that nobody noticed

The world is a better place today with SoftTokens (or not, that's a whole other thread) as if this happens again, fewer physical tokens would have to be replaced.

Bear in mind that SecurID tokens come with non-replaceable batteries and a limiting shelf life anyway, they will expire by design. The hack just expired them all, at once. Perhaps having relatively few competitors helped RSA survive this, it's a difficult market to get into which perhaps saved them.