MD5 and SHA cracked/broken...

Kirk Job-Sluder kirk at eyegor.jobsluder.net
Mon Sep 13 06:32:08 CEST 2004


On 2004-09-13, Paul Rubin <> wrote:
> Sam Holden <sholden at flexal.cs.usyd.edu.au> writes:
>> Creating a collision between a "useful" file which people can
>> examine and use and a "trojan" file which does "bad things" is
>> significantly more difficult than creating two files whose
>> MD5 sums collide but whose contents are essentially "random".
>
> Of course it's not.  Just have a block of random-looking data
> somewhere in the file, like in a bitmap image or something.  Who's
> going to notice, if the bitmap doesn't actually get displayed?


Mr. Sam Holden addressed the technical problems.  

But again, why bother when there are easier ways to get your malware out
into the wild?  You could play the odds and trust that a fair number of
people will install your software without checking MD5 sums or
signatures.  If you can plant the trojan, you can also plant your own
fake MD5 sums or signatures.  You can use phishing techniques to attract
people to your fake download site.  If anything the fair number of
recent viruses show that social engineering methods are still the best
way of making mischief than attacks on software or hardware.  

One of the more interesting aspects of the recent discoveries is that so
much attention has been placed on the least critical, least probable,
and least critical way to exploit these discoveries.  If you are going
to be paranoid, I'd be much less worried about malware using faked MD5
sums, and more worried about groups with the money, computing power, and
technical expertise to make practical use of these discoveries.

-- 
Kirk Job-Sluder
"The square-jawed homunculi of Tommy Hilfinger ads make every day an
existential holocaust."  --Scary Go Round



More information about the Python-list mailing list