Latest / Expert Insights Podcast / #62 – Phishing Attacks in the Age of AI and Voice Cloning (Charles Henderson, Coalfire)
Transcript
- 0:03SPEAKER_00: Hello and welcome to the Expert Insights Podcast.
- 0:06I'm Caitlin Harris,
- 0:07Deputy Head of Content at Expert Insights and your host for
- 0:10this episode.
- 0:14SPEAKER_00: I'm joined today by Charles Henderson,
- 0:16SPEAKER_00: EVP of Cybersecurity at Coalfire,
- 0:18SPEAKER_00: to discuss the recent surge in vision attacks,
- 0:21SPEAKER_00: the risks of Gen AI powered voice cloning,
- 0:23SPEAKER_00: and how businesses can improve their resilience
- 0:28SPEAKER_00: Thanks so much for joining us on the podcast today.
- 0:30SPEAKER_00: One of the themes we want to discuss is the rise of vishing attacks.
- 0:34SPEAKER_00: They were up 442% in the second half of 2024.
- 0:37SPEAKER_00: And what is behind that spike?
- 0:40Are attacks getting more sophisticated or are organizations
- 0:43falling behind in their
- 0:44SPEAKER_00: detection and prevention?
- 0:45SPEAKER_01: You know,
- 0:46SPEAKER_01: when you think about security,
- 0:47I like to think of security as sort of like a giant water
- 0:50balloon that we're all
- 0:50SPEAKER_01: holding up.
- 0:51SPEAKER_01: And as we get better in certain areas,
- 0:54it just shifts the water around and other areas kind of fall a
- 0:57little bit.
- 0:58It isn't necessarily that defenses in certain areas are getting
- 1:01worse so much as a
- 1:02SPEAKER_01: sign that defenses in other areas may be getting better.
- 1:05SPEAKER_01: And when...
- 1:07SPEAKER_01: threat actors,
- 1:08SPEAKER_01: when criminals out there figure out something that works,
- 1:11SPEAKER_01: they're going to really sort of keep hitting it until it stops working,
- 1:16SPEAKER_01: until we get better at holding up that part of the water balloon,
- 1:20SPEAKER_01: so to speak.
- 1:21And I think one of the things that you're seeing is criminals
- 1:24and threat actors
- 1:25SPEAKER_01: have found a new way to really target the human element.
- 1:31SPEAKER_01: And especially with
- 1:34SPEAKER_01: the rise of AI and other technologies, it's becoming a lot easier.
- 1:40SPEAKER_01: And that means higher return on investment for those that are focused on
- 1:45SPEAKER_01: profiteering and for threat actors that are more nation state oriented.
- 1:49SPEAKER_01: You know, it just means that they're able to do more with less.
- 1:52SPEAKER_00: Absolutely.
- 1:53SPEAKER_00: And you mentioned there sort of the rise of AI.
- 1:55Coalfire recently conducted an investigation with Albert Invent
- 1:59into the risks of
- 2:00SPEAKER_00: Gen AI-powered voice cloning.
- 2:01Could you walk us through how threat actors are leveraging
- 2:04voice cloning tools to
- 2:06SPEAKER_00: impersonate executives or technical leaders?
- 2:08SPEAKER_01: So,
- 2:09SPEAKER_01: you know,
- 2:10SPEAKER_01: it's interesting because anytime that you have a new technology,
- 2:13SPEAKER_01: you know,
- 2:14you think back to the dawn of email and the initial phishing
- 2:19attempts there.
- 2:21uh it's sort of ripe for abuse in terms of really taking that
- 2:27human trust and
- 2:29exploiting it and what you're seeing with ai is that um you
- 2:34know the impersonation
- 2:35capabilities of ai betray what we think of as the art of the
- 2:39possible so when
- 2:40SPEAKER_01: When you see someone's face and hear their voice and it's them talking,
- 2:45you don't have that skepticism yet because you haven't been
- 2:49exposed to it for very
- 2:50SPEAKER_01: long.
- 2:50And what that means is you're far more likely to fall for these
- 2:54types of attacks
- 2:56SPEAKER_01: when everything in the back of your mind,
- 2:58SPEAKER_01: so to speak,
- 2:59SPEAKER_01: is telling you that it's real.
- 3:01And one of the things that we do as an organization is really
- 3:05try and stay...
- 3:06SPEAKER_01: head of the threat actors.
- 3:07We try and look for new and interesting attack vectors and find
- 3:12them and help
- 3:13people defend against them before the threat actors start
- 3:16exploiting them.
- 3:16SPEAKER_01: And so in this case,
- 3:18SPEAKER_01: you know,
- 3:19SPEAKER_01: we did some work with Albert and Vint really looking at the feasibility
- 3:24SPEAKER_01: and the effectiveness of these attacks.
- 3:27And what we found is that not only is it feasible,
- 3:30it's very effective as well.
- 3:32And I think it's good that we're bringing attention to
- 3:37something like this before
- 3:38SPEAKER_01: it becomes a problem from threat actors.
- 3:40And you're already seeing cases where threat actors are using
- 3:43this in the wild.
- 3:44SPEAKER_01: It's not
- 3:45you know it's not some sort of weird hollywood plot you're
- 3:49seeing this leverage and
- 3:50so it's important i think from a reader of this white paper to
- 3:55really understand
- 3:56the art of the possible so that you can start to prepare your
- 3:59defenses prepare your
- 4:00user base and start to build that skepticism that you know that
- 4:04that voice in the
- 4:05back of your head again that says hey just because i'm seeing
- 4:08somebody's face and
- 4:09hearing their voice doesn't necessarily mean that it's it's
- 4:12it's them delivering
- 4:13SPEAKER_01: the message
- 4:14SPEAKER_00: Absolutely.
- 4:14SPEAKER_00: And when you're seeing these cases out in the wild,
- 4:17are the threat actors using this method standalone or is it
- 4:20more part of a wider
- 4:23SPEAKER_00: spearfishing campaign?
- 4:25SPEAKER_01: You see it both ways,
- 4:26SPEAKER_01: but very often when you see a new technology adopted by threat actors,
- 4:30SPEAKER_01: you'll see it
- 4:30sort of used in the lone gunman scenario first because the
- 4:34established threat
- 4:34actors that really have wide expanse don't necessarily want to
- 4:38put at risk their
- 4:39SPEAKER_01: operations by trying something new.
- 4:41They're going to use what works and they're not going to
- 4:43overcomplicate it necessarily.
- 4:45But I think what you're seeing is those early adopters in the
- 4:49threat actor space
- 4:50SPEAKER_01: using it,
- 4:50SPEAKER_01: demonstrating that it works.
- 4:52SPEAKER_01: And now you're starting to see the more established threat actors say,
- 4:55SPEAKER_01: hey,
- 4:55SPEAKER_01: if this works,
- 4:56SPEAKER_01: let's bring this attack factor up to the big leagues.
- 4:59SPEAKER_00: Absolutely.
- 4:59And you've already mentioned that these attacks are really
- 5:02successful because
- 5:03obviously they sound exactly like the person you're expecting
- 5:06to be on the other
- 5:07SPEAKER_00: end of the phone.
- 5:08So how can organizations actually assess and improve their
- 5:11resilience to phishing
- 5:12SPEAKER_00: and in particular,
- 5:13SPEAKER_00: AI enhanced phishing?
- 5:15SPEAKER_00: Is there a way to maybe simulate these kinds of attacks safely?
- 5:18Look,
- 5:19you know,
- 5:19hey,
- 5:20simulating these attacks safely,
- 5:21that's something that we do.
- 5:23SPEAKER_01: And we're happy to help our clients in that regard.
- 5:25SPEAKER_01: But that's not the end all be all.
- 5:27It's also important and something we work with our clients on
- 5:29is setting up
- 5:30SPEAKER_01: processes and procedures such that even if these are successful attacks,
- 5:35SPEAKER_01: there's something on the back end that prevents them from being.
- 5:39usable so to speak meaning there's some sort of safeguard in
- 5:42your systems and
- 5:43processes and you're not relying on a lack of human error and
- 5:48the reason we do both
- 5:49of those techniques is first of all defense in depth is is the
- 5:53basic uh basis for
- 5:55any good security strategy but additionally um if if you were
- 5:59to get your user base
- 6:00SPEAKER_01: from say
- 6:02SPEAKER_01: 50% vulnerable to 20% vulnerable,
- 6:05SPEAKER_01: meaning you reduce your vulnerable rate,
- 6:08SPEAKER_01: you're still 20% vulnerable.
- 6:09SPEAKER_01: And, you know, all it takes is one.
- 6:13SPEAKER_01: So you really want both ends of the spectrum to be working together.
- 6:21SPEAKER_01: And, you know, that's why
- 6:23SPEAKER_01: Really,
- 6:23SPEAKER_01: when we're talking to our customers,
- 6:25SPEAKER_01: we're partnering to develop a strategy,
- 6:29SPEAKER_01: more of an approach rather than just a point solution.
- 6:32SPEAKER_01: And we talk about, you know, a security program, not a security solution.
- 6:38SPEAKER_01: I'm really wary of,
- 6:40SPEAKER_01: you know,
- 6:40a lot of the folks out there that you hear at sort of the early
- 6:44points of threat
- 6:45SPEAKER_01: actors using a new technology that say,
- 6:47SPEAKER_01: hey,
- 6:47SPEAKER_01: we've solved this problem.
- 6:48SPEAKER_01: Here's our solution to this problem.
- 6:51SPEAKER_01: And it's a single solution.
- 6:52SPEAKER_01: They may have a portion of the solution,
- 6:54but really it's about bringing a strategy together and really
- 6:58executing on a
- 6:59SPEAKER_01: programmatic level.
- 7:00SPEAKER_00: Absolutely.
- 7:00And as the SWAT actors continue to evolve the ways that they're
- 7:03doing this,
- 7:04SPEAKER_00: then we need to be able to evolve our strategies as well.
- 7:07So can you talk me through a little bit the sort of layered
- 7:11approach and the
- 7:11strategy approach that you have at Coal Fire to helping prevent
- 7:15these attacks?
- 7:16SPEAKER_01: Yes, so some of it is just as simple as looking at the workflow.
- 7:22SPEAKER_01: Say you were dealing with somebody in a call center,
- 7:24SPEAKER_01: looking through the workflow and saying,
- 7:26SPEAKER_01: hey,
- 7:26where can we insert safeguards in the employees' daily
- 7:33activities in terms of their
- 7:34SPEAKER_01: access,
- 7:35SPEAKER_01: in terms of everything?
- 7:36SPEAKER_01: so that they don't have the ability to make this mistake.
- 7:39SPEAKER_01: But it's also an element of,
- 7:41SPEAKER_01: and that sort of consultative approach is key,
- 7:43SPEAKER_01: but it's also an element of then testing,
- 7:45SPEAKER_01: saying,
- 7:46SPEAKER_01: hey,
- 7:47SPEAKER_01: can we get around these safeguards?
- 7:50If we set up some sort of AI payload where we're having AI
- 7:55impersonate a customer,
- 7:57SPEAKER_01: impersonate a corporate officer,
- 7:58SPEAKER_01: impersonate any number of things,
- 8:00SPEAKER_01: is there a way for this person to bypass that control?
- 8:04And that's important,
- 8:04not because we're saying,
- 8:06okay,
- 8:06this person did bad.
- 8:07SPEAKER_01: No, this person was set up for failure.
- 8:09SPEAKER_01: So there's an element of testing.
- 8:11Then there's also elements of looking at applications and
- 8:16things of that nature to
- 8:17see if there's any kind of information leakage that can be
- 8:20leveraged to bypass
- 8:22SPEAKER_01: those controls or to build a more convincing payload from that
- 8:28SPEAKER_01: that AI attack.
- 8:30SPEAKER_01: And then finally,
- 8:30you're looking at sort of defensive strategies like threat
- 8:34hunting and whatnot to
- 8:35SPEAKER_01: see if,
- 8:36SPEAKER_01: hey,
- 8:37SPEAKER_01: has something already occurred?
- 8:38SPEAKER_01: Like, you know, it's one thing to close the barn door.
- 8:42SPEAKER_01: It's another thing to close the barn door when the horse is already out.
- 8:45SPEAKER_01: So, you know, looking around the
- 8:50occurred and you know do we see indicators of compromise do we
- 8:54see evidence of data
- 8:56leakage and then there's a final thing that we do for a lot of
- 8:59our customers um
- 9:02SPEAKER_01: And it's basically dark web monitoring.
- 9:03And you can think about in the US,
- 9:05there's this common post identity leakage.
- 9:09SPEAKER_01: So if,
- 9:09SPEAKER_01: you know,
- 9:10SPEAKER_01: your social security numbers leak,
- 9:11SPEAKER_01: you get free credit monitoring for like six months or a year or whatever.
- 9:14I don't know how many free credit monitoring this I'm signed up
- 9:17for now because
- 9:18SPEAKER_01: it's,
- 9:19SPEAKER_01: you know,
- 9:19SPEAKER_01: it seems like everybody's gifted me free credit monitoring.
- 9:22SPEAKER_01: Well,
- 9:23there's a concept of this same kind of monitoring that exists
- 9:26for corporations
- 9:27SPEAKER_01: where,
- 9:28SPEAKER_01: you know,
- 9:28SPEAKER_01: you can look for corporate information linking in the dark web.
- 9:32SPEAKER_01: And, you know, is this corporate information being traded, whatnot?
- 9:37SPEAKER_01: And that can be a sign that something bad has already occurred.
- 9:40So we'll set up these kinds of monitoring services on a
- 9:43subscription basis for clients.
- 9:45And that can be really helpful in sort of that early warning
- 9:48of,
- 9:49hey,
- 9:49something's happened.
- 9:50SPEAKER_01: Maybe I can mitigate whatever it is.
- 9:52SPEAKER_00: Absolutely.
- 9:53So it's a really in-depth solution covering from the prevention
- 9:56stage all the way
- 9:57SPEAKER_00: through to the detection and response stage.
- 9:59Is it something that you find more organisations are wanting to
- 10:03set up for just
- 10:04their executive team or is this something that you would
- 10:07recommend that they apply
- 10:08SPEAKER_00: to all of their users?
- 10:10you know you you take a risk-based approach so you know you you
- 10:14start at the the
- 10:16areas that make the most sense but you know when you're talking
- 10:20about a strategy
- 10:22you know against human error you're really talking about
- 10:25everyone because you know
- 10:26there's this there's this weird misconception at the executive
- 10:30level and i say this
- 10:31is an executive that somehow as you increase someone's pay they
- 10:35are have more
- 10:36SPEAKER_01: access into the confidential information within the company
- 10:40SPEAKER_01: And that's simply not true.
- 10:41Some entry-level positions have a huge amount of access and are
- 10:47integral into your
- 10:48SPEAKER_01: company's success or failure.
- 10:50And it's really important that you start thinking about the
- 10:54fact that this can
- 10:55SPEAKER_01: affect anyone in your company.
- 10:58SPEAKER_01: You know,
- 10:58SPEAKER_01: regardless of how intelligent or amazing you feel an employee is,
- 11:04SPEAKER_01: they can still make mistakes.
- 11:05SPEAKER_01: And setting them up so that if they make a mistake,
- 11:08SPEAKER_01: it's not catastrophic for the company is really important.
- 11:12SPEAKER_00: Absolutely.
- 11:12So we've spoken a little bit about the sort of current threat
- 11:16landscape and what
- 11:17SPEAKER_00: you guys are currently doing at Coalfire to address that.
- 11:20SPEAKER_00: Look into the future.
- 11:21SPEAKER_00: How popular do you think this attack method could become?
- 11:24Will organizations need to validate phone calls with the same
- 11:27level of security
- 11:28SPEAKER_00: that we're currently using for email?
- 11:30You know,
- 11:30I think,
- 11:31honestly,
- 11:32you need to validate phone calls already at the same level.
- 11:36SPEAKER_01: And I think if we do that well,
- 11:38SPEAKER_01: you have the potential to,
- 11:40SPEAKER_01: I think,
- 11:41SPEAKER_01: stymie these attacks before they gain a lot of popularity.
- 11:44SPEAKER_01: That said,
- 11:45SPEAKER_01: you know,
- 11:45SPEAKER_01: if history is any indication,
- 11:47I think we're going to have some firms that are very successful
- 11:50in defending
- 11:51SPEAKER_01: against these kinds of attacks now.
- 11:53SPEAKER_01: And we will have other firms that will fail miserably.
- 11:56SPEAKER_01: And,
- 11:57SPEAKER_01: you know,
- 11:57so you'll have one half of the group serve as the shining
- 12:00example on the the the
- 12:02SPEAKER_01: other half,
- 12:02SPEAKER_01: you know,
- 12:03SPEAKER_01: serve as a brilliant funeral pyre,
- 12:05SPEAKER_01: I guess,
- 12:06SPEAKER_01: for security.
- 12:07But that's what I mean to say is that,
- 12:09you know,
- 12:09not everyone's going to get it right.
- 12:11The the in those that don't will serve as a warning to to
- 12:18future organizations on
- 12:19SPEAKER_01: what not to do,
- 12:20SPEAKER_01: I think.
- 12:22SPEAKER_01: And that's a sad fact, because the truth is that threat actors are really
- 12:26SPEAKER_01: good at what they do.
- 12:29SPEAKER_01: They're not not-for-profit.
- 12:31They are very much in something that they've tried,
- 12:35they've tested,
- 12:36and they've perfected.
- 12:37SPEAKER_01: And as a result of that,
- 12:39I think the more that you do proactively now is going to make
- 12:44your experience
- 12:45SPEAKER_01: better in the future.
- 12:46SPEAKER_00: Definitely.
- 12:47And just on that point of threat actors trying and testing
- 12:50these different methods,
- 12:52how do you expect them to evolve their use of Gen AI and social
- 12:55engineering looking
- 12:56SPEAKER_00: forward?
- 12:57And how can organizations stay ahead of that curve without
- 13:00burning out their teams
- 13:01SPEAKER_00: or their budgets?
- 13:02SPEAKER_01: You know,
- 13:02I think that one of the things that we've had a real history
- 13:06with in security
- 13:07that's going to catch up with us is there's just an amazing
- 13:11treasure trove of
- 13:12SPEAKER_01: information that's been compromised through all these breaches.
- 13:14So you think about all this information that threat actors have
- 13:17collected.
- 13:18SPEAKER_01: And at some point,
- 13:19I think they're going to start to use AI to cull this
- 13:22information to look for items
- 13:24SPEAKER_01: that can be better used.
- 13:26SPEAKER_01: You know,
- 13:26SPEAKER_01: historically,
- 13:27SPEAKER_01: we've seen like,
- 13:28SPEAKER_01: say,
- 13:28credential stuffing and other attacks come out of many of these
- 13:33information
- 13:34SPEAKER_01: compromises.
- 13:35I think you'll start to see a lot of the PII and other
- 13:39information that we have that it's
- 13:41SPEAKER_01: the threat actors have that has been compromised in bulk,
- 13:45I think you'll see AI start to make use of that information
- 13:48better because it's
- 13:49almost a situation where we have the threat actors have
- 13:53information overload at
- 13:54SPEAKER_01: this point.
- 13:55And AI will be really good at distilling that information and
- 13:59creating profiles for attack.
- 14:02And I think that as you can get AI to start calling that
- 14:06information and as you're
- 14:08SPEAKER_01: developing social engineering attacks,
- 14:10SPEAKER_01: they're going to be very much more successful.
- 14:12SPEAKER_01: So,
- 14:13SPEAKER_01: you know,
- 14:13you think about all the information we have on social media
- 14:15with regard to family
- 14:16SPEAKER_01: members and everything,
- 14:18SPEAKER_01: you can start to validate these vishing attacks with really high fidelity
- 14:23SPEAKER_01: information about the targets.
- 14:25SPEAKER_01: That really kind of scares me that,
- 14:27SPEAKER_01: you know,
- 14:27SPEAKER_01: if you talk about things that keep you up at night,
- 14:29SPEAKER_01: that's probably chief among them.
- 14:31SPEAKER_00: Definitely.
- 14:32And in terms of what organizations can do now to protect
- 14:36themselves against that,
- 14:37are there any proactive steps that they should be taking today
- 14:40to try and get ahead
- 14:41SPEAKER_00: of that?
- 14:42SPEAKER_01: You know,
- 14:42SPEAKER_01: I think that,
- 14:43SPEAKER_01: you know,
- 14:44SPEAKER_01: the biggest thing is,
- 14:45SPEAKER_01: again,
- 14:46setting yourself up so that your processes do not allow human
- 14:50failure to undermine
- 14:51SPEAKER_01: your security policy.
- 14:52SPEAKER_01: That's that's key.
- 14:53SPEAKER_01: You know,
- 14:54SPEAKER_01: just as if,
- 14:54SPEAKER_01: you know,
- 14:55SPEAKER_01: you think about simple things like that are analogous,
- 14:58SPEAKER_01: you know,
- 14:58SPEAKER_01: if you think about
- 15:00SPEAKER_01: the failures in password authentication.
- 15:02The reason we have two-factor and multi-factor authentication
- 15:05is we realized we had
- 15:06SPEAKER_01: a problem,
- 15:07SPEAKER_01: so we put in a safeguard.
- 15:08SPEAKER_01: And so that multi-factor authentication is your safeguard.
- 15:11Well,
- 15:12authentication is not the only thing that has that sort of
- 15:17second-tier safeguard.
- 15:18SPEAKER_01: So you can look at all the processes you have in place and say,
- 15:22do we have a secondary control in place that's going to protect
- 15:26us in the event of
- 15:27SPEAKER_01: human error?
- 15:27SPEAKER_01: Yeah.
- 15:27SPEAKER_00: Absolutely.
- 15:28SPEAKER_00: So it's not all doom and gloom.
- 15:30SPEAKER_00: There are things that we can do.
- 15:32You know,
- 15:32our job is in many ways to be the doom and gloom for our
- 15:37clients.
- 15:37SPEAKER_01: It's to spot those problems before they are major problems.
- 15:41SPEAKER_01: And,
- 15:41SPEAKER_01: you know,
- 15:41I've run multinational incident response firms that answer the
- 15:45phone when on a
- 15:46SPEAKER_01: client's worst day.
- 15:47SPEAKER_01: And I'll tell you,
- 15:48SPEAKER_01: there's a lot of ways you can find you have a security gap,
- 15:52SPEAKER_01: but one of the worst is to find it in the course of an incident response
- 15:55SPEAKER_01: investigation.
- 15:56SPEAKER_01: So
- 15:57Anytime that we can help a client identify problems before it's
- 16:00an incident
- 16:01SPEAKER_01: response investigation,
- 16:02SPEAKER_01: that's a great day.
- 16:03So,
- 16:04you know,
- 16:04I would say that one person's doom and gloom is another
- 16:07person's bed of roses.
- 16:08SPEAKER_00: Definitely, definitely.
- 16:10SPEAKER_00: I really like that outlook.
- 16:11SPEAKER_00: So I've just got one last question for you today.
- 16:14What would be your final words of advice to any security teams
- 16:17listening who are
- 16:18looking to protect their users against phishing and voice
- 16:20cloning threats?
- 16:21You know,
- 16:22I think I would say find out what you don't know before it
- 16:25becomes a problem.
- 16:26SPEAKER_01: You know,
- 16:26I have a large number of longtime clients who talk about the
- 16:31fact that they're
- 16:32SPEAKER_01: really excited that they found a certain thing,
- 16:35SPEAKER_01: again,
- 16:35SPEAKER_01: the bed of roses.
- 16:36And I've never had somebody say,
- 16:38hey,
- 16:39I really wish I didn't find out about this vulnerability.
- 16:42And,
- 16:42you know,
- 16:43you're sitting there listening to this and say,
- 16:45am I vulnerable?
- 16:46SPEAKER_01: Isn't it time to find out?
- 16:47SPEAKER_00: That's excellent, excellent words of advice there.
- 16:50I think it just,
- 16:51it all comes back to you can't protect something that you don't
- 16:53know about.
- 16:54SPEAKER_00: Really good advice there.
- 16:56SPEAKER_00: Thank you so much, Charles, for joining me on the podcast today.
- 16:58SPEAKER_00: It's been really, really great speaking to you.
- 17:00SPEAKER_01: Hey, thanks for having me.
- 17:01SPEAKER_00: Thank you.