3 ms·
Ask HN: Can we still trust SHA-1 and SHA-2 created by NSA?
- EthanHeilman 13y agoWhy not use Keccak/SHA-3 instead, it was developed in an open competition run by NIST with some NSA involvement. http://en.wikipedia.org/wiki/SHA-3 http://en.wikipedia.org/wiki/SHA-3
- lmm 13y agoYes (at least, as much as we could before; SHA1 in particular has been showing weaknesses). They are still the hash functions that have had the most attention from the academic community, and so far no workable attacks have been found.
- venomsnake 13y agoAs much as you could trust them 5 days ago. While SHA-1 should not be trusted too much because it has shown possible theoretical attacks SHA-2 still holds. Also these kind of things are IP - there are a lot of eyeballs and scrutiny going on. There is much bigger chance of fraked up implementation that will make it insecure than the theory - there are a lot of independent researchers that have scrutinized them quite a bit. And while I am sure NSA employs a lot of very capable people they do not hold monopoly on world class cryptographers.
- ig1 13y agoSHA-1 should be assumed to be broken in any case. The Flame malware was distributed using a fake certificate that was generated via a brand new (publicly unknown) chosen prefix collision technique against SHA-1.
- tptacek 13y agoFlame used an MD5 collision.
- ig1 13y agoThe Microsoft certificate revokation list for Flame: Microsoft is providing an update for all supported releases of Microsoft Windows. The update revokes the trust of the following intermediate CA certificates: Microsoft Enforced Licensing Intermediate PCA (2 certificates) Microsoft Enforced Licensing Registration Authority CA (SHA1) (from http://technet.microsoft.com/en-us/security/advisory/2718704 http://technet.microsoft.com/en-us/security/advisory/2718704) Two of the certs were MD5, the third was SHA1.
- tptacek 13y agoThe hash collision exploited by Flame was in MD5, not SHA1.
- jayfuerstenberg 13y agoBCrypt is superior to the SHA family of hash algorithms. That should be reason enough not to use SHA.
- tptacek 13y agoThis is a non sequitur. BCrypt isn't an alternative to SHA hashes; it's a password storage construction.
- tptacek 13y agoYes, but you shouldn't anyways; SHA-2 is inferior to SHA-3. SHA-3 is the product of a peer-reviewed cryptographic contest.
- isaacb 13y agoThese aren't closed algorithms. They are well understood and explored in-depth by the academic community.