UPDATE: Slides
Dan Kaminsky was wondering most of the weekend what I would post about Toorcon. If there’s one thing I learned it is this: Dan Kaminsky is nuts. The future projects and other theory that comes out of his mouth is awesome. I had a great time hanging out with him. His talk was similar to the one from this year’s Blackhat (slides here).
Dan started by discussing the breaking of MD5. In ’96 MD5 was theoretically broken and in ’04 two example “vectors” were released with the same MD5. Many denied that this was a problem since it was just a “toy” case so Dan set out to implement it. Once there is a collision anything appended to the vectors will also collide. Browsers are really good about attempting to render anything they’re given no matter how full of crap it is, this is the Geocities feature. Dan’s demo takes two web pages, appends a bunch of crap to both and uses Javascript to maintain the look of the original. The end result is two webpages with the same MD5.
He also covered fragmentation attacks to bypass IDS, his massive scanning project and visualizing the resulting data.
…so what’s his shirt say?
I (heart) 2.4 GHZ
Does that mean he has two DIFFERENT web pages that produce the same MD5 signature? Do the web pages render to the same thing, or do they contain visually identifiable differences. I’d really like some more info.
Here ya go…
http://www.doxpara.com/t1.html
http://www.doxpara.com/t2.html
Same MD5 Hash.