Global Survey on Email Security – Brianna Leddy, Darktrace

This is Techstrong tv. Hey everyone, welcome back to techstrong tv. Um, we've got another great guest here for you, and another would promises to be great.

Uh, discussion. I want to introduce you to Brianna Leddy, and I hope I pronounced that right, Brianna, is that the right pronunciation? Yes, you did.

Very cool. Brianna is director of Analysis at Darktrace. And, um, first of all, Brianna, welcome and thanks for coming on our show today.

Uh, let's start off with a little bit about who's Brianna Leddy and kind of your journey. Yeah, and thank you all. I'm really, really happy to be joining today.

So director of analysis at Darktrace. I started with the company a little over seven years ago as an analyst. So working directly in our customer environments, helping them detect and respond to all sorts of attacks, I think ransomware was, was kind of the big thing we were seeing back then.

But everything from banking, Trojans, insider threat, and now in my role I do a lot of work with our customers on how they're using Darktrace, helping them to operationalize our self-learning ai and talking about some of their own use cases for detection and response. Excellent. Um, you know what I realize, Brianna, some people, I mean we're talking about Dog Trace almost as a noun, right?

But a lot of people may not be familiar or may think they're familiar, but why don't you give folks a little darktrace kind of primer here. Yeah. So Darktrace is about 10 years old now, and the core of our approach has always been in self-learning, ai, understanding the business, understanding what is normal for your unique organization, and using that understanding to best protect the digital environment.

So using that sense of self to find what is not self, what might be a threat that is present in the environment, and then going on to take proportionate action through Darktrace respond to better contain those threats. So the core of Darktrace has has and always been an ai. Really, Really, that's one way of looking at it.

Ai before it was cool to be ai and of course Darktrace headquartered in New York, right? Yes. We actually found it out of Cambridge in the uk.

So we are a global company now. Um, but we've got our, our research center in Cambridge as well. Really Very cool.

I always see the building with the Darktrace name on it in downtown New York. I, I usually stay when I'm in New York City. I stay, uh, near Barrick Street down there and I always see Doc Tracy and I'm always tempted to go knock on the door and say hi, but you should, You need more than all.

They don't need me knocking on their door. But, uh, anyway, good, good with all of that. So we were talking off camera, man, you know, we started Techstrong do AI now about two months ago, right?

Because frankly, the AI stories were overwhelming. com and cloud native now and digital cxo because it is, it, it it's not just our, it's a overwhelming a lot of things, right? We have, I was telling you, we, I prior interview I did, 60% of developers claim they're using generative ai, not necessarily to write code, but as part of checking code, writing code is part of their work.

Um, so, you know, there's all these positives about all the great things AI can do, but like everything else on the internet, bad people are gonna be bad and they're gonna take good tools and use them for bad purposes and, and AI's no different. So interested in, in, you know, you guys must be seeing this already. Yeah, it certainly, and, and interesting about that, that 60% stat because like you said, there's a lot of benefits around these AI tools in productivity gains, even just in better phrasing of things.

I've seen people using them to, to beef up their resumes. Um, at Darktrace we've actually found 74% of our active customer deployments are utilizing generative AI chat G P t, at least in some form. And as you said, you know, there's risks that are both outside of the business of attackers using that, but there's also the risks inside of the business of the unintentional data leakage or exposure of intellectual property, even by well-meaning employees trying to leverage these tools, but accidentally uploading the wrong thing into some of these large language models.

Agreed. Agreed. Um, and, and so that is, let's call it unintentional, an unforced error, right?

That's like a tennis term, an unforced error. Um, and that that's gonna happen. And um, you know, I guess you guys are seeing that, but then there's just the malicious people where it's intentional what they're doing too, and, and there's a, there's a lot of potential for intentional kind of maliciousness there.

Um, you guys recently have done some research on this though, and you, you know, with the global survey, why, if you don't mind, Brianna, share with us some of the findings. Of course. So, so our survey found that from January to February of this year, kind coinciding with the widespread adoption of chat G P T, there was a 135% increase in novel social engineering attacks.

And, and what I mean by novel is these were using more advanced linguistic capabilities, you know, longer text, more punctuation. They were also using less links and attachments. So the obvious red flags that many of us look for to spot a phishing email, you know, hey, click on this link, or hey, you've won a giveaway chat, G P T and other tools like that are making it easier for attackers to, to circumvent that and to be a lot more sophisticated and targeted, but at machine speed and scale.

Agreed. Um, you know, the other thing, and I've heard this from other people too, is look, they're using chat G P T to write their emails. So a lot of phishing scam emails were easily identifiable because, you know, generally pa generally PayPal doesn't send out stuff with grammar or spelling issues, right?

But now with chat G P t writing these emails, it's not as easy. I mean, they, they look native, right? They're not as, not as easy to just spot.

Agreed. You know, I, I like to think I'm a a pretty suspicious person, pretty keen eyed when it comes to email. I'm, I'm sure you are too, Alan, you know, anyone in the, the tech space knows what to look out for, but I can, this is making it easier to impersonate trusted senders, making it easier to avoid those obvious red flags.

And humans alone aren't enough to be the last line of defense anymore. We can't put it all on our employees to spot 100% of these attacks. I don't disagree.

So how so what are we to do here then? Yeah, and, and this has always been, you know, dark traces thought about attackers using ai defenders are going to need to use AI to keep up. While a human can't spot those nuanced differences between a legitimate email and, you know, an impersonated email.

AI can and Darktrace email takes that approach of learning what is normal, both for your internal and external senders, so that when there is a deviation, say a trusted supplier had their account compromised, the AI can pick up on those subtle differentiators, uh, slight change in language, a slight change in intent over what the sender is, is trying to get you to do with that email. And so in doing so, that can help our humans help by protecting us with that AI itself. Absolutely.

Um, let's talk more about the survey. Give us some other findings that you think might be interesting. Yeah, I mean, I think the overall decrease in the links and attachments was, was another thing that was interesting to me because an attacker's goal is, is often to, you know, compromise your account, steal money, get into the network somehow.

And, you know, we always think they will need some payload to do that, but by exploiting that trust relationship, they can easily kind of take that conversation maybe offline to somewhere that's not monitored by security and use that as their initial intrusion factor. Or they could, you know, just establish that trust relationship and convince you to hand over some confidential information without even needing to compromise your account in the first place. Yeah.

So I'm, I'm starting to see more mails like that myself. So I get to catch all mails here, right? I've done it from day one at my company, and, um, so this way I get a good, you'd be surprised all the good scam stuff I see.

And, um, I'm seeing a lot of them lately. It'll be something like, oh, Brianna, I, I know we spoke earlier, just I lost your cell phone number. Yep, yep.

It's sliding that Previous, just, I'll give you a call or, you know, Brianna, Alan needed something, I know he was trying to reach you. Mm-hmm. You know, and there was something like that, or Alan called me, but he got cut off and, you know, what's, how can we reach him?

Or something like that. Or do you have his, I mean, just like you, right? No, no links, no, no, nothing to click.

They're just, they're trying to get you to actually go give them the information Mm-hmm. Yeah. To, to pick up the phone and call them.

I know there's, there's a lot of the gift card type scams. Oh, hey, you know, Brianna's Well between that and the, and the messages about your account being limited and here's your invoice for that, you know, McAfee, uh, endpoint security for $5,000 that you bought and stuff like that. I mean, it's just, you know, the typical sort of, but there I definitely see a, an improvement in the quality and I'm assuming it, it it's AI that, that's driving that.

Um, look, we, we can't go over the whole report obviously here and research, but for people who want to get more data and more information on this, is it on the Darktrace website? Yes. com.

We've got a lot of information around that email research and, and how Darktrace email can help protect you as well as some, some really interesting research by our team that's looking into AI itself. Yep. You know, you mentioned, um, uh, I just threw a blank there for a second.

A ransomware Mm-hmm. Right? Was very big when you first joined.

It's still, it didn't go away here. Definitely. Are we seeing, are we seeing AI being used in ransomware scams as well now or No, I, I think it's mostly in that initial intrusion that's the lowest barrier to entry.

Uh, you know, I think when you look at some of the other aspects of either developing or deploying ransomware, that still takes some technical know-how, you know, if, if you're using chat g p T to help develop malware, you still have to understand how the malware works and, and be able to test it. So I don't think we're seeing it in as many of the other stages of attack so far. But that's not to say that it, it couldn't come up again in the future to make attacks faster, uh, harder to detect, you know, make it so that humans alone can't keep up with the pace and the scale of what we're seeing from all types of threats.

Absolutely. You know, I mean, unfortunately the, just as we're all experimenting with generative AI and ways it could help us in our business models, the bad guys are as well, and, and you know, you don't, and they're not, they're not dumb people. They may be evil, but they're not dumb.

Certainly. And, uh, you know, it'll be interesting to see, I bet you if we discussed next year's global survey, you know, new vectors, new techniques mm-hmm. That these, these people are harnessing AI for to help, you know, help them within nefarious objectives.

Yeah. And I think as we think about all the innovation that, that we are doing on the good side with ai, you know, leveraging AI to help us investigate leveraging ai, AI to help us with threat intelligence, all of those different use cases, as exciting as that innovation is, the attackers are also doing that innovation on their ends too, and, and figuring out new ways that they can leverage things. Agreed.

Agreed. Anyway, Hey, Liz, we're about outta time. I, Liz, excuse me, Brianna, we're about outta time, Liz, on my brain.

We're about out time. I want to make sure we hit again. com is the main website for people wanting to get more information.

They can get information on this survey right from there as well as about all the great Darktrace products, right? Of course, yes. Fantastic.

Thank you for coming on text on TV today. We appreciate your time. Thank you very much, Alan, for having me.

All righty. Brianna Leddy here, uh, from Doc Trace on Textron tv. We're gonna take a break.

We'll be right back.