WEBVTT

00:00:00.040 --> 00:00:02.060
We've done that, and now I'm going to refer back

00:00:02.060 --> 00:00:04.110
to, back on script stuff, right?

00:00:04.110 --> 00:00:05.590
So I actually have these commands here.

00:00:06.250 --> 00:00:08.510
The second one that we're going to look at it is going to look at this SSL log.

00:00:08.740 --> 00:00:12.603
Okay. So we kind of got these kind of three ports that we know we can use a

00:00:12.603 --> 00:00:15.350
search. We can probably open up the pcap in Wireshark, and we'll get into that

00:00:15.360 --> 00:00:17.270
in a little bit and kind of identify what's going on.

00:00:17.280 --> 00:00:19.196
But especially, if this is encrypted information, we need

00:00:19.196 --> 00:00:21.020
to know that. It's going to be SSL, right?

00:00:21.360 --> 00:00:24.890
And this is where that JA3 certificate, or hash, is

00:00:24.890 --> 00:00:26.810
going to come into play as well.

00:00:26.820 --> 00:00:30.110
So what is interesting about this one is that there's

00:00:30.110 --> 00:00:31.710
not really a lot of hashes generating.

00:00:32.810 --> 00:00:36.070
Yeah, that is kind of interesting, especially on the, kind of the odd ports.

00:00:36.080 --> 00:00:37.140
Yeah, absolutely.

00:00:37.390 --> 00:00:41.580
And this is, I'm not, if you'll notice, I'm not sorting by port, right?

00:00:41.580 --> 00:00:44.140
I'm not I'm not filtering by port I guess is what I'm saying, right?

00:00:44.150 --> 00:00:44.490
Okay.

00:00:44.500 --> 00:00:48.790
So it makes sense SSL should mainly be over 443, but here we have that SSL or a

00:00:48.790 --> 00:00:53.500
TLS running over 447 and 449. So then we have another set of things lining up

00:00:53.510 --> 00:00:55.770
with that weird traffic that's associated TrickBot.

00:00:55.780 --> 00:00:56.520
Right. Right.

00:00:57.370 --> 00:01:03.340
So let me ask what with this, one of the things with analysis is understanding

00:01:03.340 --> 00:01:08.920
what looks normal versus what is unusual. If you were to just go onto a normal

00:01:08.930 --> 00:01:11.250
system that was doing all this network traffic,

00:01:11.340 --> 00:01:15.450
how would you expect the JA3 tags or hashes to look in that case?

00:01:15.440 --> 00:01:16.330
Yeah, absolutely.

00:01:16.330 --> 00:01:19.550
So we've got into it in some other courses and research that I've done as

00:01:19.550 --> 00:01:23.660
well, but we did traffic analysis on a medium‑sized network for about 3

00:01:23.660 --> 00:01:27.140
hours, and we had 16 different client JA3 hashes.

00:01:27.150 --> 00:01:27.460
Okay,

00:01:27.470 --> 00:01:30.160
so you would expect to see more than just what we're seeing here. There's mores,

00:01:30.390 --> 00:01:32.340
but it's still not like an exorbitant amount.

00:01:32.830 --> 00:01:33.680
And again,

00:01:33.680 --> 00:01:37.550
the least frequent case, we often saw like 10,000, 3,000, 4,000, whatever, and

00:01:37.550 --> 00:01:38.789
then there was like 5, and we're like, well, what is that?

00:01:39.210 --> 00:01:40.350
That's weird, right?

00:01:40.440 --> 00:01:45.462
Right. And then even here what points out that's more weird is SSL not

00:01:45.462 --> 00:01:50.930
on 443. Yeah, and would you normally expect to see the same JA3 hash

00:01:50.940 --> 00:01:54.015
on two different ports? You would not. That would normally be, well,

00:01:54.015 --> 00:01:56.920
it depends on the application, right? So ts is going to be application specific,

00:01:56.930 --> 00:02:00.690
which is why it's one of my favorite tools ever is because it is a

00:02:00.690 --> 00:02:03.360
network tool, but we're also, it's associated directly with the

00:02:03.370 --> 00:02:07.360
executable running on the device. And so this executable has only

00:02:07.360 --> 00:02:10.949
certain libraries built in for that cipher suite, so it can only offer

00:02:10.960 --> 00:02:13.280
those libraries in the TLS handshake.

00:02:13.290 --> 00:02:13.550
Okay, that makes sense.

00:02:13.560 --> 00:02:18.390
And so those libraries are what make the MD5 hash, and so it becomes kind of a

00:02:18.910 --> 00:02:22.560
potential unique identifier for that executable. And especially when we talk

00:02:22.560 --> 00:02:26.290
about malware and malware authors is they're not Google and they're not

00:02:26.290 --> 00:02:29.870
Microsoft, and so they're not including every possible cipher suite, so it

00:02:29.870 --> 00:02:34.070
doesn't end up being reasonably unique. And the other side of that, we can also

00:02:34.070 --> 00:02:36.150
get the server side of that,

00:02:36.160 --> 00:02:39.340
which is we can do that by just adding the s, right? So there's the

00:02:39.340 --> 00:02:43.217
server side as well. And so here we have the same application reach out

00:02:43.217 --> 00:02:47.040
on two different servers on 443. I can't verify that that is or is not

00:02:47.040 --> 00:02:52.400
TrickBot. I just don't know, mainly based on this. I can look at these

00:02:52.400 --> 00:02:56.010
hashes, and we can go back, and remember I said that JA3 on the

00:02:56.020 --> 00:02:58.100
Salesforce GitHub, they actually looked at  TrickBot,

00:02:58.390 --> 00:02:58.690
right?

00:02:59.540 --> 00:03:00.440
So if we come here,

00:03:00.780 --> 00:03:02.980
here's an Emotet sample, and we can go back and look and

00:03:02.980 --> 00:03:06.610
see if that's it, and then here's a TrickBot client sample.

00:03:06.650 --> 00:03:10.550
And if we look back at that, that 673 there, that's none of them.

00:03:10.870 --> 00:03:13.827
Right. So it's kind of interesting in the first place because this is, we're

00:03:13.827 --> 00:03:18.584
reasonably positive that 447 and 449 is TrickBot, but neither the server nor

00:03:18.584 --> 00:03:21.800
the client JA3 hash. So is this something new?

00:03:21.810 --> 00:03:24.430
Maybe. So let's keep scrolling down .And I was looking at

00:03:24.440 --> 00:03:27.740
this in preparation for this Play by Play, and I also saw

00:03:27.740 --> 00:03:30.290
that they have the C2 servers, right?

00:03:30.290 --> 00:03:33.100
So here's the fingerprint of TrickBot on the client, and then we

00:03:33.100 --> 00:03:35.425
have the fingerprint of the command and control server response.

00:03:35.425 --> 00:03:41.300
So there's the 673 for the client and the 623de for the server,

00:03:41.310 --> 00:03:42.150
and then look at that.

00:03:42.490 --> 00:03:44.500
So that is the exact same command and control.

00:03:44.510 --> 00:03:48.740
Okay. We can reasonably say that this is the same version of

00:03:48.740 --> 00:03:52.540
software for command and control running on that server that

00:03:53.170 --> 00:03:54.590
they saw when they looked at TrickBot.

00:03:54.670 --> 00:03:55.570
Okay, cool.

00:03:55.580 --> 00:03:58.340
Which is kind of neat because it's built into that binary

00:03:58.340 --> 00:04:00.110
that's running that command and control server.

00:04:00.190 --> 00:04:03.784
And it's just another way that we can kind of confirm that the 447 and

00:04:03.785 --> 00:04:06.969
the 449 traffic is really using TrickBot. Absolutely.

00:04:07.140 --> 00:04:09.550
And what's even more interesting is we now kind of discovered an

00:04:09.560 --> 00:04:13.460
indicator for this new version on the client side, is we have this JA3

00:04:13.460 --> 00:04:19.000
hash here, and they have an associated project called JA3 here where

00:04:19.000 --> 00:04:23.010
there's 1605 different JA3 hashes that have been submitted so that you

00:04:23.010 --> 00:04:24.500
can identify different applications.

00:04:24.730 --> 00:04:27.870
The idea here is remember I talked about there's only 16 JA3

00:04:27.870 --> 00:04:30.410
hashes identified in a normal network, is that you can

00:04:30.410 --> 00:04:32.690
whitelist them. Because why not, right?.

00:04:32.700 --> 00:04:34.990
Make it easy to spot the outliers.

00:04:34.990 --> 00:04:35.750
Yeah, exactly.

00:04:36.040 --> 00:04:41.090
So I'm going to search for this hash and see if we come up

00:04:41.100 --> 00:04:42.720
with that or if anyone's ever seen it before,

00:04:42.760 --> 00:04:45.410
because that's kind of cool thing to do.

00:04:45.820 --> 00:04:47.150
Maybe we just found something new,

00:04:48.340 --> 00:04:51.650
and maybe a copy paste will work out of my ‑‑‑ yes, it did.

00:04:52.150 --> 00:04:55.070
All right, so let's search for this, and no one's actually seen it.

00:04:55.070 --> 00:04:58.820
Okay. So the sample that you've analyzed is a newer version of software,

00:04:58.830 --> 00:05:01.410
including new cipher suites that wasn't included in the

00:05:01.410 --> 00:05:03.850
previous software while they still simultaneously did not

00:05:03.860 --> 00:05:05.100
update their command and control servers.

00:05:05.110 --> 00:05:05.860
Right, okay.

00:05:06.030 --> 00:05:08.544
Which is kind of a pretty interesting thing. So I'm going to go

00:05:08.544 --> 00:05:13.250
ahead and add a comment, a Trickbot Client 2019.

00:05:13.720 --> 00:05:15.440
Okay. And there they go.

00:05:15.550 --> 00:05:17.140
Now we've contributed to the community.

00:05:17.140 --> 00:05:17.470
Awesome.

00:05:17.480 --> 00:05:19.480
Great. So we're giving back already.

00:05:19.480 --> 00:05:20.220
Yeah, exactly.

00:05:20.580 --> 00:05:21.000
Yeah, cool.

00:05:21.000 --> 00:05:23.080
So that's done. And then you and I,

00:05:23.080 --> 00:05:26.010
we were kind of messing around earlier looking at the, so we're still

00:05:26.010 --> 00:05:31.600
looking at this SSL log here, but we started looking at the certificate

00:05:31.730 --> 00:05:33.860
itself. So this is now we're talking about certificate,

00:05:33.860 --> 00:05:34.950
not JA3 hash,

00:05:34.950 --> 00:05:36.620
right? So cipher suite is available versus the

00:05:36.620 --> 00:05:38.200
certificate used for that exchange.

00:05:38.730 --> 00:05:41.990
So when we looked at this and identified that same 449 port,

00:05:41.990 --> 00:05:44.630
we also see the Internet Wdgets Pty thing,

00:05:44.710 --> 00:05:47.670
which we've learned is self‑signed certificates.

00:05:47.670 --> 00:05:48.220
Right. Right.

00:05:48.230 --> 00:05:50.010
But there's also some other stuff here.

00:05:50.010 --> 00:05:52.010
It was like they didn't really go to a ton of effort, right?

00:05:52.280 --> 00:05:57.550
Even self‑signed certificates you can change some state to actually be a state.

00:05:57.860 --> 00:06:03.920
Some state in the country. Here it's pointing to Australia, AU,

00:06:04.060 --> 00:06:04.960
which I think is Australia.

00:06:05.210 --> 00:06:07.935
I think so . I hope so. It's embarrassing if it's not. Yeah, that's why I

00:06:07.935 --> 00:06:13.540
agree. But that doesn't necessarily mean that these IP addresses or the

00:06:13.540 --> 00:06:15.060
certificates are actually on a server in Australia,

00:06:15.380 --> 00:06:15.800
right?

00:06:15.810 --> 00:06:16.120
Right.

00:06:16.450 --> 00:06:18.930
You could just go in and set it to be whatever you want.

00:06:18.940 --> 00:06:19.550
Absolutely.

00:06:19.600 --> 00:06:22.820
That's the kind of indicators that I look at for more behavioral stuff

00:06:22.820 --> 00:06:25.730
that doesn't change as often. And we can use this.

00:06:25.730 --> 00:06:28.568
You can use Bro to look at‑‑‑you can capture pcaps at

00:06:28.568 --> 00:06:29.730
different places in your network.

00:06:30.000 --> 00:06:33.530
Yes, you need to capture it external to internal egress,

00:06:33.530 --> 00:06:36.830
right? That's a good point to look because this stuff is facing out.

00:06:36.840 --> 00:06:37.950
But you also want to look internal.

00:06:37.950 --> 00:06:39.140
There's other things we talked about.

00:06:39.140 --> 00:06:40.855
Like you said, it was using EternalBlue, EternalRomance,

00:06:40.855 --> 00:06:44.330
EternalChampion, but that's more lateral spread stuff, right?

00:06:44.340 --> 00:06:44.940
Correct.
