Cybercriminals Use Malicious Memes That Communicate with Malware | Hacker News Reader

Cybercriminals Use Malicious Memes That Communicate with Malware

blog.trendmicro.com

66 pointssus_00718 comments

If I were in a band, I'd seriously consider naming it the "Malicious Memes"

"It should be noted that the malware was not downloaded from Twitter and that we did not observe what specific mechanism was used to deliver the malware to its victims."

Is this not a waste of resources? the malware creators could open a twitter account and post commands directly using established and apparently inocuous phrases or words. tweeting “hi” is not a breach of ToS and is less complicated.

I think it's because a meme-repeating Twitter account looks more real than one which is randomly saying 'hi' (or whatever) over and over and over.

It's pretty clever, because the malware needn't even necessarily be checking the same account over and over: a DGA-style[0] approach could be used, in which the malware checks a different account each day (or hour, or whatever).

Twitter has some options to find this sort of thing, of course, but malware authors have some options to more-cleverly hide data, too. Encoding raw text in an image is easy, and easily-found; encrypted (and hence random-looking) data is harder to detect. Re-encoding images can destroy some sorts of hidden data, but not others (e.g. lossy encoding will likely destroy data encoded on low bits, but information encoded as large swaths of light or dark is less likely to be lost).

Error coding could even be used to make the data more resistant to loss. I don't know if there's a good way to make data all of resistant to loss, encrypted and hidden though: seems like the structure of error coding would make it detectable, while a high-apparent-entropy code would be more susceptible to unrecoverable errors. But I'm not an expert in the field.

Those are just my off-the-cuff thoughts; someone is an expert no doubt has better thoughts to share.

0: https://en.wikipedia.org/wiki/Domain_generation_algorithm

Using Twitter and one way broadcast, plus stegonography seems like the malware equivalent of a numbers station.

Can anyone elaborate how this code is actually executed?

If I download one of these images, which appears to embed CSharp Winform code, how does it actually run within the system?

I suppose it couldn't unless taking advantage of some other exploit along the way. And if that is the case, this article is perhaps a little incomplete.

From the article:

The memes contain an embedded command that is parsed by the malware after it’s downloaded from the malicious Twitter account onto the victim’s machine, acting as a C&C service for the already-placed malware. It should be noted that the malware was not downloaded from Twitter and that we did not observe what specific mechanism was used to deliver the malware to its victims.

It's not the images that are executed. The images are used to pass commands to malware that's already installed on a machine.

TLDR: malware uses meme images containing hidden instructions posted on Twitter as a way to receive commands. It's one-way communication though - the result of the operation is still uploaded to a C&C server.

you could ex-filtrate data by the same mechanism though (e.g posting images to twitter)

That might attract some attention, for example, in a corporate environment where everything is locked down.

Some things will seem very innocent though. Having a video call with a family member while sending out data under the noise floor.

In reality the amount of alerts a SOC has to triage is so big on avg I doubt they would really be looking into someone accessing twitter.

So why would the malware not just check the C&C server for orders? Why hope that your victim checks Twitter and sees your memes? This seems like a really roundabout way of accomplishing something trivial (I mean, trivial as compared to infecting a machine with your malware in the first place...)

If your C&C server has been "compromised" or blocked by, say, law enforcement, the tweets could instruct malware to use another?

Because that would reveal the location (IP address) of the C&C server, in addition to making it a point of failure.

But the output of all the commands still sends to the C&C server, so I don't see how this protects that. Unless, as the other user suggested, the memes could actually also contain the IP to use for the C&C server. So the "order" is akin to "upload username to ###.##.###.#"

Perhaps the malware will sit idle like a "sleeper" agent until activated, and you want to reduce the chance of preemptive detection.

Having its background polling look like a typical Twitter user agent would help it remain hidden and ensure that its command channel is not blocked. (At least for a large class of victims who do not have severe network restrictions in effect.)

You could also send the destination address via Twitter, so that the malware does not leak information about the controller ahead of its activation or so that the controller can make last minute arrangements of temporary resources...

Reply on news.ycombinator.com