d3ad social login

[d3ad]

tags labelled 'hash' e

please login to post
@barray on Mon Jan 02 12:34:43 UTC 2023 said: &e
I *really* want to see #rss get into the #mainstream https://www.niemanlab.org/2022/12/this-i.. I still believe it needs to handle two problems: 1. #redistribution checking (content #hash ) and 2. #comments . #comments could be handled via #email but ideally #rss itself woudl handle this. Anyway, I still want to see #deadsocial based on #rss and completely #distributed .
@barray on Sat Oct 22 11:48:34 UTC 2022 said: &e
Wow, somebody found a #sha 3 #hash #buffer #overflow https://mouha.be/sha-3-buffer-overflow/ Pretty impressive! Worse still, this #code has been #vulnerable for over 10 years!
@barray on Tue Sep 20 14:07:06 UTC 2022 said: &e
Yeah you probably shouldn't be using MD5 #hashes anymore: https://github.com/corkami/collisions It seems now #trivial to #generate #collisions - very cool for #cracking old #technology . We can now make *some* SHA1 #collisions happen, I imagine in the #future we can make #hash #collisions happen trivially for something like #git or for #password checking.
@barray on Tue Aug 30 13:19:47 UTC 2022 said: &e
Very interesting #visualization for #maze #solving https://www.jwz.org/blog/2022/08/solving.. I wonder what other #problems could be looked at in this way? What would a #hash look like when #visualized in this way?
@barray on Mon Oct 11 22:06:27 UTC 2021 said: &e
#opensource work towards #video #hashing ( #perceptual video #hash ): https://pypi.org/project/videohash/ The idea is that you can check whether two different videos have matching #content instantly. It is still a way off and makes a bunch of #assumptions - but in theory it could lead to some interesting ideas, like the ability to #rebuild #torrents from #random #sources ...
@barray on Tue Aug 24 05:23:04 UTC 2021 said: &e
Interesting #rolling #hash #concept for #rabinkarp #string #search https://yurichev.com/news/20210205_rolli.. Always great to see something I hadn't even ever considered before. Very cool.
@barray on Thu Aug 19 18:02:03 UTC 2021 said: &e
#apple 's #csam is even worse than previous thought, people are finding #hash #collisions in the wild: https://blog.roboflow.com/nerualhash-col..
@barray on Thu Aug 19 02:26:49 UTC 2021 said: &e
@barray on Thu Aug 19 00:10:40 UTC 2021 said: &e
Scratch that, it looks as if people are actually generating #hash #collisions by modifying an existing image to match a #hash https://github.com/AsuharietYgvar/AppleN.. #apple have really fucked up on the response to this one... They should at the very least put #csam on ice until #researchers have been given more time to figure this out. Also, claims that it's "not the final #version " is not good enough, especially when you invited people to #attack it!
And now the #source #code is out there for generating your own #csam #hash #collision that can fool #apple 's #detection #algorithm https://github.com/anishathalye/neural-h.. Apple claim it will be impossible to figure out what #vector they are using to trigger detection, but it's only a matter of time before somebody figures them out with a #database of #childporn (something the #uk #police for example store). So essentially #stateactors have the ability to #reverseengineer their vectors.
@barray on Thu Aug 19 00:10:40 UTC 2021 said: &e
@barray on Thu Aug 19 00:00:50 UTC 2021 said: &e
#apple are claiming there is no issue with their #csam #detection #algorithm because #researchers are not using the "final version", despite being invited to #test it: https://www.vice.com/en/article/wx5yzq/a.. The lack of awareness on this is actually shocking... Even if researchers can't produce #hash #collisions on the #private version, #independent #nations can!
Scratch that, it looks as if people are actually generating #hash #collisions by modifying an existing image to match a #hash https://github.com/AsuharietYgvar/AppleN.. #apple have really fucked up on the response to this one... They should at the very least put #csam on ice until #researchers have been given more time to figure this out. Also, claims that it's "not the final #version " is not good enough, especially when you invited people to #attack it!
@barray on Thu Aug 19 00:00:50 UTC 2021 said: &e
@barray on Wed Aug 18 10:48:11 UTC 2021 said: &e
Just a few days ago somebody replicated the #apple #csam #detection #algorithm https://github.com/AsuharietYgvar/AppleN.. Just now, somebody has successfully produced a #hash #collision - *exactly* what people were concerned about: https://github.com/AsuharietYgvar/AppleN.. Apparently it's even worse that this - different #hardware generates a different hash because of #float #precision ...
#apple are claiming there is no issue with their #csam #detection #algorithm because #researchers are not using the "final version", despite being invited to #test it: https://www.vice.com/en/article/wx5yzq/a.. The lack of awareness on this is actually shocking... Even if researchers can't produce #hash #collisions on the #private version, #independent #nations can!
@barray on Wed Aug 18 10:48:11 UTC 2021 said: &e
Just a few days ago somebody replicated the #apple #csam #detection #algorithm https://github.com/AsuharietYgvar/AppleN.. Just now, somebody has successfully produced a #hash #collision - *exactly* what people were concerned about: https://github.com/AsuharietYgvar/AppleN.. Apparently it's even worse that this - different #hardware generates a different hash because of #float #precision ...