News for hackers. Stuff that matters.
Re:In Cryptography, Advances in Program Obfuscation by Zikes on Thursday January 30, 2014 @06:13PM (#7154041)

    But if you have a black box obfuscator, creating a
    public key encryption protocol becomes a simple matter
    of choosing your favorite secret-key encryption scheme,
    expressing its workings as a computer program, 
    obfuscating the program, and making the obfuscated
    version widely available. Anyone can then use it to
    encrypt a message to send to you, but no one can tease
    the decryption key out of the obfuscated software.
I cannot imagine downloading and running such an application for which nobody could feasibly determine it was safe, except to run it in some sort of secure sandbox.
Re:In Cryptography, Advances in Program Obfuscation by joe_the_user on Thursday January 30, 2014 @10:43PM (#7155111)
Ah,

Further, how could any of us determine that any given program we happened to have downloaded wasn't built with something like this?

If true, this seems to open ever nefarious possibility one might imagine. The evil could be in the executable but it could be in weird point arithmetic in the source code.

I hope this is just hype or something. (reading the article, I'm supposing the main problem is that so far any program like this will be huge).

Re:In Cryptography, Advances in Program Obfuscation by Sniffnoy on Thursday January 30, 2014 @11:04PM (#7155164)
Note that this is if you have a black-box obfuscator, and black-box obfsucators are impossible. (The proof is actually pretty simple; see [0] ("Theorem 4") for a short version.) Although, according to the article, apparently something similar is possible with an indistinguishability obfuscators; I'm not too clear on how that works. In that case, though, where you use an indistinguishability obfsucator, then yes there is the problem of having a huge program (which is probably also slower).

[0] http://www.scottaaronson.com/blog/?p=392

Re:In Cryptography, Advances in Program Obfuscation by tptacek on Thursday January 30, 2014 @06:44PM (#7154198)
Then you're not being very imaginative. Instead of assuming that this some giant EXE, assume instead that the (open source, carefully vetted) host for this code exposes a small VM, with no linkage to system libraries or direct access to memory, whose sole purpose is to unwrap small encrypted blobs.
Re:In Cryptography, Advances in Program Obfuscation by bdamm on Thursday January 30, 2014 @06:58PM (#7154268)
Hence a secure sandbox, as the parent suggested. Careful when you insult people.
Re:In Cryptography, Advances in Program Obfuscation by tptacek on Thursday January 30, 2014 @07:45PM (#7154491)
Hm. You're right. That came out way snippier than I meant it too. Thanks.
Re:In Cryptography, Advances in Program Obfuscation by bdamm on Tuesday February 11, 2014 @02:16AM (#7216314)
This combo really is something. I saw your other comment about private-key being the interesting part (and not high-speed symmetric operations). This really is a breakthrough but I am not sure today where it will go.
Re:In Cryptography, Advances in Program Obfuscation by DerpDerpDerp on Thursday January 30, 2014 @06:56PM (#7154262)
How can you tell it's not swapping text to include "Fire all the missiles!" (or other message of choice) to whatever you're encrypting?
Re:In Cryptography, Advances in Program Obfuscation by jerf on Thursday January 30, 2014 @09:10PM (#7154818)
If you are running an obfuscated blob, it's probably not because you intend to trust it, it's probably because you are offering the ability to run some computation to a another person to run in your context in a way that they can trust. (It wasn't mentioned, but I imagine that trying to falsely run a program like this would also be very difficult; the slightest deviation probably shreds the output; my intuition says this would be a necessary condition for any such obfuscation scheme to stand. Combine this with the obfuscated code signing the result and you're probably looking at something extremely hard to forge.) Your trusted environment and input, their trusted code, a result that either they trust or perhaps you mutually trust (with some protocol stuff added in).