Hacker News Reader

Comment by beefhash | Hacker News Reader

Are the people over there sure that it's a good idea to rely on the broken[1] MD5 and the close-to-be-broken[2] SHA-1 for verifying checksums in the context of malicious actors? Though I guess the hashes and file sizes differ, so I guess this is just being pedantic.

[1] https://en.wikipedia.org/wiki/MD5#Collision_vulnerabilities

[2] https://sites.google.com/site/itstheshappening/

View on HN

MD5 is broken in a sense that you can craft two different files with the same hash, but it's still hard to create a file with specific MD5 hash.

So it's still good for identifying files that aren't specifically crafted to have a malicious counterpart.

From this use case, where one person makes a file, and then an attacker forges another one with the same hash, MD5 was just recently broken and SHA-1 is far from it. Also, it's not immediately clear if a pair of broken hashes is also broken or not (depends on many things, and I probably don't even know half of them).

So, I'd say this is safe, for now. Yet, I also get bad feelings when somebody either comes with an MD5 hash or uses a SHA-1 as the most secure option.

It seems more reliable to check the developer signature in Properties on the EXE - the correct developer is "Ivaylo Beltchev". The infected download is unsigned and requires you to click through a warning.

Though annoyingly it's signed using sha1, not sha256.

So if they were gonna put effort into making an sha1 collision, they'd probably target the signed payload, not the overall exe.

Though it doesn't look like sha1 is that broken yet, for the budget of this grade of attacker.

Reply on news.ycombinator.com