Page 1 of 1
MD5 latest news (not good)
Posted: Tue Mar 21, 2006 11:18 pm
by Roja
http://cryptography.hyperlink.cz/2006/tunnels.pdf
In a nutshell, if this theory holds up to scrutiny (and it looks fairly well done, and reputable), MD5 will be utterly broken.
Now we wait for the crypto community to analyze it.
Posted: Tue Mar 21, 2006 11:31 pm
by feyd
well.. that came a tiny bit faster than I anticipated. If correct, a lot of applications need to bone up their hashes soon.

Posted: Wed Mar 22, 2006 7:45 am
by John Cartwright
good thing I have been using sha256 since feyd realeased his snipplet

Posted: Wed Mar 22, 2006 7:58 am
by Roja
feyd wrote:well.. that came a tiny bit faster than I anticipated. If correct, a lot of applications need to bone up their hashes soon.

The extremely bad news is that *if* this concept holds up to scrutiny (and that is NOT a sure thing - this is the first paper on the concept, to my knowledge), then SHA will be vulnerable to it as well - including the SHA-2 family.
It could be a very big break in hashing algo's. This could be a
very bad thing.
I've contacted Bruce Schneier, asking him to take a look and comment on it if he chooses to on his blog or newsletter. His insight is usually some of the very best in the crypto community, so it will be interesting to see his take on it.
Posted: Wed Mar 22, 2006 9:47 am
by pickle
Maybe this'll provide the kick in the butt the crypto community needs to come up with better methods of cryptography.
Posted: Wed Mar 22, 2006 11:34 am
by Chris Corbyn
Hmmm that doesn't look good. I'll have to wait for more news on this. Thanks Roja for the link!

Posted: Wed Mar 22, 2006 12:08 pm
by neophyte
Looks like md5 is "hashed".
Does this spell doom for the larger hash algo's like mmmm sha256?
Posted: Wed Mar 22, 2006 1:04 pm
by Chris Corbyn
neophyte wrote:Looks like md5 is "hashed".
Does this spell doom for the larger hash algo's like mmmm sha256?
Looking at that paper it technically applies to the whole concept of hashing. It would obviously be less predictable and take longer on higher bit hashes though. I haven't had a chance to read all 11 pages yet since I'm uber-busy but I'll have a read when I get chance

Posted: Wed Mar 22, 2006 1:11 pm
by Roja
pickle wrote:Maybe this'll provide the kick in the butt the crypto community needs to come up with better methods of cryptography.
I personally deeply doubt it.
The crypto community is smart. Very smart. Its not a lack of work/effort/intelligence on their part for not coming up with "better" methods. It is substantially an arms race between brains and processing power - and the brains don't
stand a chance.
I suspect that if it turns out to be accurate, it will cause widespread change - but "Better" is subjective.
I want to note that my view is *extremely* abnormal in the Crypto community. Many feel they already have solutions to the problems, or will soon.
Posted: Wed Mar 22, 2006 2:20 pm
by irisblaze
I agree with Roja, it'll not cause a big change, and this is not new, we were warned long time ago that md5 will be broken