White-Box Cryptography

[LaBBa]

"White-box cryptography is a technique to hide a secret key into a cryptographic software implementation in a white-box model. In such a model, an adversary has full control over the execution environment.

A white-box DES encryption binary with embedded secret key. If you like, try to extract the secret key, using all information you can find from this implementation (input-ouput attacks, so called black box attacks, are not allowed). "

here is there demo link (cygwin1.dll is needed):
https://www.cosic.esat.kuleuven.be/sopro/wbc/wbDES.exe

here is there website:
https://www.cosic.esat.kuleuven.be/sopro/
https://www.cosic.esat.kuleuven.be/sopro/wbc/

i'm currently working on this and look at the attached file that i have made :P

come and join to reverse this protection ... sooner the better

regards,
LaBBa.

[rendari]

Cloakware brags about this white boy stuffs on their website, and yet I wonder how effective it is in the real world... to prevent your softie from getting cracked.

[bwyseur]

The implementations you refer to, are implemented according to the description of white-box DES implementations by Chow et al. [http://crypto.stanford.edu/DRM2002/whitebox.pdf], with some minor improvements [`clarifying obfuscation'-paper, by Link et al.; and some personal improvements].

I implemented this binary, to test my cryptanalysis results mid 2006. The details of the cryptanalysis are published at SAC'07, and is mentioned on the website you refer to. See also http://eprint.iacr.org/2007/104 for technical details. A different cryptanalysis result has been published at SA'07 too (See http://eprint.iacr.org/2007/035).

More information on white-box cryptography will appear on a to-be-launched website: http://www.whiteboxcrypto.com. This will include a PhD thesis on this topic, which I'm writing at this very moment.


You mention that input-output (black-box) attacks are not allowed. I disagree. Every cryptographic implementation should at least be able to withstand black-box attacks. Given full control over the implementation and its execution environment, an adversary could just execute the implementation as black-box as many times as he wishes to. Why would an adversary bother to deploy an more sophisticated white-box attack when simple is possible?

Protection against implementation attacks (side-channel and white-box) is an extra level of defense, on top of black-box protection. Hence, it only makes sense to implement "secure" ciphers (DES, AES, ...).

Best regards,
Brecht Wyseur