WEBVTT

00:00:00.785 --> 00:00:03.781
So requirement 5, anti-malware/anti-virus,

00:00:03.781 --> 00:00:09.519
and I know this as an assessor, and actually someone who runs networks now,

00:00:09.519 --> 00:00:11.781
I know this is a hard requirement because,

00:00:11.781 --> 00:00:14.785
you know, we have basically people having service,

00:00:14.785 --> 00:00:18.131
and I don't, I nearly sad a bad word there.

00:00:18.131 --> 00:00:21.133
I do not want this stuff running on my servers,

00:00:21.133 --> 00:00:26.383
and then we have people managing fleets of point-of-sale devices saying,

00:00:26.383 --> 00:00:30.378
I definitely don't want this stuff running on my point-of-sale device,

00:00:30.378 --> 00:00:34.572
it slows it down, it might not make it not work,

00:00:34.572 --> 00:00:35.450
you know?

00:00:35.450 --> 00:00:39.120
So this is a tiny requirement with two big problems.

00:00:39.120 --> 00:00:42.153
The first big problem is I just don't want to do this requirement.

00:00:42.153 --> 00:00:43.483
What's your answer?

00:00:43.483 --> 00:00:49.791
Well, if it is a system that is commonly affected by malicious software,

00:00:49.791 --> 00:00:51.258
which for a long time, you know,

00:00:51.258 --> 00:00:54.555
was a particular sort of set of operating systems.

00:00:54.555 --> 00:00:55.873
Okay, but that was always a euphemism.

00:00:55.873 --> 00:00:56.922
It was a euphemism.

00:00:56.922 --> 00:01:00.223
Because the Council would not say if you're running Windows you must run AV.

00:01:00.223 --> 00:01:01.423
Right, right.

00:01:01.423 --> 00:01:07.291
And so it was a euphemism, and for those systems the answer is kind of tough,

00:01:07.291 --> 00:01:07.938
right?

00:01:07.938 --> 00:01:09.879
Like, do this.

00:01:09.879 --> 00:01:12.179
Even things like that point-of-sale,

00:01:12.179 --> 00:01:15.108
the embedded Windows that people want to sell run.

00:01:15.108 --> 00:01:18.135
Are they still susceptible to viruses?

00:01:18.135 --> 00:01:20.027
Yes, they are; they are.

00:01:20.027 --> 00:01:23.164
One of the sort of unfortunate things,

00:01:23.164 --> 00:01:26.037
if somebody talks about running a stripped-down version of a

00:01:26.037 --> 00:01:28.122
particular off-the-shelf operating system,

00:01:28.122 --> 00:01:30.953
what they have stripped down is not the attack

00:01:30.953 --> 00:01:32.851
surface or the potential vulnerability,

00:01:32.851 --> 00:01:36.821
but usually the means to do something about it.

00:01:36.821 --> 00:01:40.025
So saying it's stripped-down Windows or it's stripped-down Linux

00:01:40.025 --> 00:01:45.995
doesn't usually mean it is less susceptible to attack.

00:01:45.995 --> 00:01:46.544
I mean,

00:01:46.544 --> 00:01:49.000
unless it's very specifically sort of stripped down to

00:01:49.000 --> 00:01:50.966
minimize those kinds of attack surfaces,

00:01:50.966 --> 00:01:52.888
that that is relatively rare.

00:01:52.888 --> 00:01:58.044
Right, and do I need to go and buy AV or anti-malware software,

00:01:58.044 --> 00:02:00.905
or can I use the stuff in the operating system?

00:02:00.905 --> 00:02:03.311
Well, it's tough.

00:02:03.311 --> 00:02:07.896
You know, the folks that do the most sort of talking,

00:02:07.896 --> 00:02:08.333
thinking,

00:02:08.333 --> 00:02:12.545
and writing about anti-malware have for a long time

00:02:12.545 --> 00:02:18.379
espoused use a free utility and a commercial utility and

00:02:18.379 --> 00:02:20.723
when you look at malware trends,

00:02:20.723 --> 00:02:24.125
when you look at particular attacks using motion software,

00:02:24.125 --> 00:02:28.554
there's always that question, you know, these vendors and these products got it,

00:02:28.554 --> 00:02:30.013
these products didn't, you know,

00:02:30.013 --> 00:02:32.430
or at least a certain point in time before they all got caught up.

00:02:32.430 --> 00:02:36.946
So having more than one gets you a better cross-section of coverage.

00:02:36.946 --> 00:02:39.119
The standard still says anti-virus.

00:02:39.119 --> 00:02:40.542
It does.

00:02:40.542 --> 00:02:43.558
And I know from friends that there was discussion about

00:02:43.558 --> 00:02:45.946
changing that at one stage to say anti-malware.

00:02:45.946 --> 00:02:46.988
Right.

00:02:46.988 --> 00:02:50.447
But if I use an anti-malware tool, say some of the more modern,

00:02:50.447 --> 00:02:52.683
I hate to use the word machine learning,

00:02:52.683 --> 00:02:55.954
artificial intelligence, endpoint protection agents.

00:02:55.954 --> 00:02:58.001
Things that are not merely signature based.

00:02:58.001 --> 00:03:00.143
Things that are not signature based like the traditional AV.

00:03:00.143 --> 00:03:03.434
That meets the quality of the requirements, doesn't it?

00:03:03.434 --> 00:03:04.920
It can, yeah, I mean you evaluate it on its merits,

00:03:04.920 --> 00:03:07.765
right, but does it stop malware?

00:03:07.765 --> 00:03:12.910
Now if you look at any of the major cyber-security advisories,

00:03:12.910 --> 00:03:16.697
if you look at NIST, if you look at the U.K.

00:03:16.697 --> 00:03:17.351
Cyber Essentials,

00:03:17.351 --> 00:03:20.569
if you look at the Australian National Cyber-Security

00:03:20.569 --> 00:03:22.850
Center Essential Eight or Top Four,

00:03:22.850 --> 00:03:25.405
they all talk about application white-listing as being

00:03:25.405 --> 00:03:28.589
one of the most important controls.

00:03:28.589 --> 00:03:31.914
Is application white-listing good enough rather than anti-virus?

00:03:31.914 --> 00:03:38.082
It can be useful, right?

00:03:38.082 --> 00:03:41.516
I would say again, it comes down to what are you white-listing,

00:03:41.516 --> 00:03:45.915
right, and what kind of happens at the detail level?

00:03:45.915 --> 00:03:48.956
I've seen organizations do application kind of white-listing

00:03:48.956 --> 00:03:51.525
things where one of the things that they white-list is like

00:03:51.525 --> 00:03:55.187
the Java runtime on a workstation, right?

00:03:55.187 --> 00:03:59.316
Well, are you limiting what Java files can be executed?

00:03:59.316 --> 00:03:59.710
Well, no,

00:03:59.710 --> 00:04:02.491
it doesn't really support that; it just prevents the or

00:04:02.491 --> 00:04:04.424
allows the execution of this thing.

00:04:04.424 --> 00:04:07.104
Well, so now I can run any hostile Java code on there, right?

00:04:07.104 --> 00:04:08.372
That doesn't do you any good.

00:04:08.372 --> 00:04:11.484
So application white-listing on its own wouldn't be sufficient?

00:04:11.484 --> 00:04:13.960
Application white-listing on its own,

00:04:13.960 --> 00:04:17.416
I would say you need to evaluate what is being white-listed,

00:04:17.416 --> 00:04:17.909
right?

00:04:17.909 --> 00:04:19.442
In certain cases, right,

00:04:19.442 --> 00:04:22.123
if there are really only a limited set of things

00:04:22.123 --> 00:04:25.269
that can execute on that system,

00:04:25.269 --> 00:04:28.812
and they don't offer the opportunity for the attacker to

00:04:28.812 --> 00:04:32.512
specify some parameter that's hostile, maybe so.

00:04:32.512 --> 00:04:33.566
You mean, like PowerShell.

00:04:33.566 --> 00:04:36.913
Yeah, if you white-list that, right, then yeah,

00:04:36.913 --> 00:04:40.304
you've basically permitted, you know, hostile execution,

00:04:40.304 --> 00:04:40.792
right?

00:04:40.792 --> 00:04:44.462
And that's one thing; the other thing I think is

00:04:44.462 --> 00:04:47.598
standards has got this thing about anti-malware removing

00:04:47.598 --> 00:04:50.373
the malware after it's disabled it,

00:04:50.373 --> 00:04:53.516
and typically application white-listing doesn’t have that capability.

00:04:53.516 --> 00:04:55.504
So you sort of need the two things running.

00:04:55.504 --> 00:04:59.394
Often, often, or at least the ability to sort of quarantine something harmful.

00:04:59.394 --> 00:04:59.828
Yeah.

00:04:59.828 --> 00:05:03.308
So can we talk about, there's probably two elephants,

00:05:03.308 --> 00:05:04.745
there's two elephants in requirement 5.

00:05:04.745 --> 00:05:05.773
Yes.

00:05:05.773 --> 00:05:06.725
I'm going to talk about one because it's really funny.

00:05:06.725 --> 00:05:12.476
As you might've noticed, requirement 5 at one stage changed to say and operating.

00:05:12.476 --> 00:05:19.219
So I think in version 3 or version 2 of PCI DSS it said put anti-virus on there,

00:05:19.219 --> 00:05:22.371
and then in version 3 it said put anti-virus on there and make it work.

00:05:22.371 --> 00:05:23.416
And make it actually do something.

00:05:23.416 --> 00:05:26.299
Because there was a very famous compromise where

00:05:26.299 --> 00:05:28.788
someone had installed the anti-virus, they'd done everything,

00:05:28.788 --> 00:05:29.485
but they disabled it,

00:05:29.485 --> 00:05:32.566
and their argument was the standard didn't say it had to run,

00:05:32.566 --> 00:05:36.216
which is one of the reasons why DSS is so prescriptive.

00:05:36.216 --> 00:05:39.593
Right, it has to contend with this essentially bottomless well of bad faith.

00:05:39.593 --> 00:05:43.995
Yeah, absolutely, and it's really hard to write bad faith interpretive standards.

00:05:43.995 --> 00:05:45.532
It is tough.

00:05:45.532 --> 00:05:50.361
But the real elephant we should talk about on requirement 5 is

00:05:50.361 --> 00:05:53.594
this used to be a euphemism for Windows,

00:05:53.594 --> 00:05:55.960
commonly affected, so as I still have clients saying,

00:05:55.960 --> 00:05:58.183
well, I'm running Linux, I don't need to run anti-virus.

00:05:58.183 --> 00:05:59.133
Right.

00:05:59.133 --> 00:06:00.171
That's the most common one.

00:06:00.171 --> 00:06:02.084
I'm running the macOS, I don't need to run anti-virus.

00:06:02.084 --> 00:06:03.503
What's your take on that as an assessor?

00:06:03.503 --> 00:06:03.732
Well,

00:06:03.732 --> 00:06:07.663
so the requirement that showed up around version 3 in 5.1 .2 says for

00:06:07.663 --> 00:06:10.798
systems not commonly affected by malicious software,

00:06:10.798 --> 00:06:15.016
essentially do some threat evaluation to say,

00:06:15.016 --> 00:06:18.324
are we actually affected by malicious software,

00:06:18.324 --> 00:06:19.295
and if so,

00:06:19.295 --> 00:06:23.881
should we employ some sort of anti-malware utility to protect these systems?

00:06:23.881 --> 00:06:30.412
Now I read that as show me as your assessor some sort,

00:06:30.412 --> 00:06:32.607
and it is an interview requirement, right?

00:06:32.607 --> 00:06:36.865
But show me some capacity that makes a meaningful

00:06:36.865 --> 00:06:39.432
kind of threat analysis decision, right,

00:06:39.432 --> 00:06:44.080
to say what are the threats to these systems that are,

00:06:44.080 --> 00:06:48.668
you know, Linux or macOS or mobile devices or what have you,

00:06:48.668 --> 00:06:48.948
right?

00:06:48.948 --> 00:06:52.842
And then are you susceptible to malicious software, right?

00:06:52.842 --> 00:06:55.711
If you've got a Linux system, you know,

00:06:55.711 --> 00:06:59.278
that's like a SAMBA file server that has a bunch of Windows clients attached,

00:06:59.278 --> 00:07:03.636
yeah, I think there's a really good case for running anti-virus software there.

00:07:03.636 --> 00:07:09.365
Me just making a statement to you saying, well, when it's Linux, bone head.

00:07:09.365 --> 00:07:09.962
That doesn't count.

00:07:09.962 --> 00:07:11.142
That doesn't cut it, right?

00:07:11.142 --> 00:07:17.154
Saying, we have evaluated our threats, we can substantiate our reasoning,

00:07:17.154 --> 00:07:19.507
we don't think this particular case merits it,

00:07:19.507 --> 00:07:21.820
well, you know, then we can evaluate that,

00:07:21.820 --> 00:07:26.526
right, and realistically, I don't think you can necessarily just say,

00:07:26.526 --> 00:07:28.035
Linux systems, no, right?

00:07:28.035 --> 00:07:31.916
Maybe our Linux web servers get one sort of treatment and

00:07:31.916 --> 00:07:33.785
our application servers get another treatment and our

00:07:33.785 --> 00:07:36.293
file servers get yet another, right?

00:07:36.293 --> 00:07:37.776
And that seems reasonable.

00:07:37.776 --> 00:07:39.247
Okay, anything else we should think about requirement 5,

00:07:39.247 --> 00:07:43.077
because it's quite a short requirement, it's pretty self-explanatory.

00:07:43.077 --> 00:07:44.216
It is, it is.

00:07:44.216 --> 00:07:47.357
Just a slight sort of thing about conventional anti-virus,

00:07:47.357 --> 00:07:50.627
it does talk about running periodic scans,

00:07:50.627 --> 00:07:53.981
and so that actually means periodic scans and not

00:07:53.981 --> 00:07:57.182
merely just on-access scanning.

00:07:57.182 --> 00:07:59.672
So do actually configure it to run weekly or something

00:07:59.672 --> 00:08:02.079
like that and not just on file access,

00:08:02.079 --> 00:08:05.471
because that will catch things that the on-access won't.

00:08:05.471 --> 00:08:08.296
I always try and ask in each requirement,

00:08:08.296 --> 00:08:11.675
has a failure of this ever led to a compromise you're aware of?

00:08:11.675 --> 00:08:16.233
Sure, or at least again has exacerbated an existing failure.

00:08:16.233 --> 00:08:19.501
I don't do PCI forensic investigation anymore,

00:08:19.501 --> 00:08:21.224
but I have,

00:08:21.224 --> 00:08:24.738
and we get forensic images of systems that we've

00:08:24.738 --> 00:08:26.185
acquired that have been attacked, right,

00:08:26.185 --> 00:08:28.353
where there have been payment card compromises and we find

00:08:28.353 --> 00:08:31.115
some piece of malware and we extract it out,

00:08:31.115 --> 00:08:35.042
and then the anti-virus software running on our lab machines

00:08:35.042 --> 00:08:36.377
then immediately lights up and it's like,

00:08:36.377 --> 00:08:39.112
oh, this thing is bad, I'm going to quarantine it!

00:08:39.112 --> 00:08:42.880
So despite maybe the arguments that conventional

00:08:42.880 --> 00:08:46.289
signature-based anti-virus software is a relic,

00:08:46.289 --> 00:08:48.966
which you know is not entirely without merit,

00:08:48.966 --> 00:08:52.797
it does still catch plenty of bad stuff that you know,

00:08:52.797 --> 00:08:53.684
captures payment cards.

00:08:53.684 --> 00:08:57.218
But does more modern things that look for malware-type behavior,

00:08:57.218 --> 00:09:00.860
i.e., oh look, it's trying to scrape memory.

00:09:00.860 --> 00:09:03.883
That's going to give you an indication that there's a bad guy that's got in,

00:09:03.883 --> 00:09:06.243
and it's part of your defense in depth.

00:09:06.243 --> 00:09:06.953
This is a pretty important control.

00:09:06.953 --> 00:09:08.025
It is.

00:09:08.025 --> 00:09:11.039
I want to finish requirement 5 by telling a story.

00:09:11.039 --> 00:09:13.697
Great!

00:09:13.697 --> 00:09:16.256
Because you know about Struts and that Struts has vulnerabilities,

00:09:16.256 --> 00:09:20.284
and I had a client who was running Struts on their web server,

00:09:20.284 --> 00:09:22.309
and the Struts vulnerability,

00:09:22.309 --> 00:09:25.148
this isn't the one that hit the big company last

00:09:25.148 --> 00:09:28.521
year; this is about three years ago.

00:09:28.521 --> 00:09:31.313
And it was a 0 date that was renounced, there's Struts,

00:09:31.313 --> 00:09:33.739
there's a vulnerability, here's the patch,

00:09:33.739 --> 00:09:36.869
and you know DSS says patch within 30 days.

00:09:36.869 --> 00:09:39.546
They were hit within a day and a half of that

00:09:39.546 --> 00:09:45.904
vulnerability being publicly published, and they're running Linux servers,

00:09:45.904 --> 00:09:49.223
but they had decided to put anti-virus software on there.

00:09:49.223 --> 00:09:53.713
And the anti-virus software spotted the root kit that the attacker

00:09:53.713 --> 00:09:57.196
installed once they got access to the server.

00:09:57.196 --> 00:09:59.413
And so the fact that he was looking for root kits

00:09:59.413 --> 00:10:04.087
saved them a major data breach, because they got an alert when,

00:10:04.087 --> 00:10:08.176
like within about a couple of hours, and they fixed it straightaway,

00:10:08.176 --> 00:10:09.919
and without that running on their Linux server,

00:10:09.919 --> 00:10:11.775
they'd have been had.

00:10:11.775 --> 00:10:16.806
And that's actually why I would always put AV on Linux servers,

00:10:16.806 --> 00:10:21.136
because at least it will give me root detection of those types of things,

00:10:21.136 --> 00:10:22.452
known bad software.

00:10:22.452 --> 00:10:23.452
Yes.
