How the NSA may have cracked encryption

[MartG]

Interesting article - it's amazing what can be done with a budget of several $billion

https://freedom-to-tinker.com/blog/haldermanhening...

[jbudgie]

Interesting, as you say.

But when 'the singularity ' arrives none of us will be safe.

[eldar]

jbudgie said:

Interesting, as you say.

But when 'the singularity ' arrives none of us will be safe.

Or all of us will.

[russ_a]

You could always write a letter and deliver it by hand!

[Joey Ramone]

eldar said:

Or all of us will.

It's the ultimate 'heads or tails?'. Except the winner gets utopia, and the loser gets something very, very different.

[marshalla]

Not so much cracking encryption as exploiting a weakness in the implementation. From the sound of it, they still can't crack a proper implementation in real-time (yet).

[TheExcession]

There's a nice video here explaining how the mathematics work for Elliptic Curve Diffie Hellman key exchange and cryptography.

The maths is pretty hardcore but it is easy to follow the principles of what is going on and how it works.

[0000]

marshalla said:

Not so much cracking encryption as exploiting a weakness in the implementation. From the sound of it, they still can't crack a proper implementation in real-time (yet).

Ding, we have a winner. Nothing to see here.

[deckster]

0000 said:

marshalla said:

Not so much cracking encryption as exploiting a weakness in the implementation. From the sound of it, they still can't crack a proper implementation in real-time (yet).

Ding, we have a winner. Nothing to see here.

I disagree. It's not so much exploiting a weakness in the implementation as it is a salutatory reminder that all modern encryption relies on nothing more than a very large set of potential solutions that must each be individually checked to find the correct one, and therefore we are hoping that the expected effort that needs to be spent to check these solutions is either too large to complete in reasonable time, or at least is large enough to make it not worthwhile. Therefore what was secure 20 years ago (128 bit keys for example, or even 1024 bit in this case) is no longer secure on modern hardware for a suitably motivated and well-funded attacker - and that what is secure now, will not be in 5/10/20 years. And even that is barring any genuine breakthroughs in mathematics, algorithmics, or (more likely) hardware design.

Security isn't a one-time activity; you need to be on top of things all the time to stay ahead of the game.

[0000]

I struggle to see using a predictable prime as anything other than a weakness in the implementation. Change the implementation and the problem goes away.

I'd imagine the authors behind bcrypt and the like see not adapting to increases in hardware performance as a weakness in the algorithm.

[marshalla]

0000 said:

I struggle to see using a predictable prime as anything other than a weakness in the implementation. Change the implementation and the problem goes away.

I'd imagine the authors behind bcrypt and the like see not adapting to increases in hardware performance as a weakness in the algorithm.

This.

They haven't broken the asymmetric encryption used during the shared key-exchange mechanism, they've calculated some common shared keys used by broken implementations. Change the way the shared keys are produced and the weakness is reduced.