I thought I'd mention, the Reddit user in question has an updated post. Seems his findings have changed, Emsisoft has failed his latest experiment and now only Bitdefender blocked whatever he built.
That Eset is heavily reliant on User Mode components, he didn’t need to waste time writing tools and testing, he could ask me and I would tell him
Eset doesn’t keep it a secret that the behavioural monitoring is HIPS with heuristics. They further disclose that it operates in user mode.
So he is testing whatever Eset clearly wrote in a whitepaper…
Very useful!
Anyway, again there are few faults:
First of all, the author is mentioning “these are the same techniques”. Yes, they are. But in home AVs, often single techniques are not blocked as vigorously as they are in business products.
“They should block in kernel mode”….
They would, but the sample is not malicious enough.
Also, additional protection layers like reputation and so on haven’t been accounted for.
This doesn’t come down to the “true infection source”, even executables on the desktop have their reputation inspected.
It comes down to configuration. As well as I am getting the sense that his sample may have been excluded and some products may have been tweaked a little bit to allow the sample to run.
The author is not biased but is expecting very aggressive response just from one isolated module (behavioural blocking). But this is not how things work.
I dare him to sit down and try to build 20 malware detection behavioural profiles. Then let him test on 20 trusted executables (perhaps base the profiles on sandbox reports) and let him see how many false positives he will get. His profiles will flag at least 10/20.
Then he will change his mind.
Bottom line is we know very well what’s malicious but it often intersects with poorly written. Hence, single techniques cannot be an accurate point to block samples.
In the case of “stealing your discord token”, there will be more indicators, the discord tokens will be accessed (handles will be open to files), probably archive will be created in the temp folder, domains will be contacted, eventually persistence hooks will be creates… these are the high confidence indicators that can produce a block (though these can be iffy as well).
It’s ok, he’s a newbie and will learn.
@KnownStormChaser you can send him this post.