Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

CYBR.SEC.CAST

CYBR.SEC.CAST artwork

It’s not magic – it’s just AI with Daniel Miessler

Sep 3, 202538 minAudio
  • Apple Podcasts
ShareXLinkedInFacebookWhatsAppEmail

In this episode, Michael and Sam sit down with Daniel Miessler, founder of Unsupervised Learning, for a thought-provoking conversation that spans the future of cybersecurity and Daniel’s unique personal journey. They dive into the escalating arms race between attackers and defenders, explore how Daniel’s path shifted from pre-med to cybersecurity thought leader, and preview his upcoming HOU.SEC.CON. talk, “Killer Context: How AI Will Eat Security and Software.” Things Mentioned: Unsupervised Learning - https://newsletter.danielmiessler.com Google says its AI-based bug hunter found 20 security vulnerabilities - https://techcrunch.com/2025/08/04/google-says-its-ai-based-bug-hunter-found-20-security-vulnerabilities/ Register for HOU.SEC.CON. - https://web.cvent.com/event/9ba9c5ea-9502-44a2-922e-d026c047c9f3/register?rp=7fc495f3-0ae2-4b86-a115-0ce784642a9f Do you have a question for the hosts? Reach out to us at podcast@houstonseccon.com Keep up with HOU.SEC.CON.: LinkedIn Twitter Facebook Instagram YouTube Bluesky Keep up with CYBR.SEC.Media: LinkedIn Twitter Facebook Instagram Check out our Conferences and Events: HOU.SEC.CON. OT.SEC.CON. EXEC.SEC.CON. HSC User Group Support or apply to our Scholarship Program: TAB Cyber Foundation Subscribe to the podcast: Apple Spotify In this episode: Host: Michael Farnum Host: Sam Van Ryder Guest: Daniel Miessler Production and editing: Lauren Andrus Music by: August Honey

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

SPEAKER_03And welcome everybody to HughSet Cast, the HusekCon podcast. This is episode 4-9 or 4-9er, as we would say, in the Army, because our guest was in the army as well. So, you know, cool stuff like that. How are you doing, Sam? Good. How about you? I'm good. I'm good. Always excited to have a friend come on and talk to us. Daniel Meessler is a good dude. He's uh does stuff over at Ofer Unsupervised Learning. He does really cool content over at Unsupervised.

SPEAKER_03Yeah. All his podcasts and everything. He's big on AI. Really glad to have him. And we'll I think we're gonna share that same picture that we shared back when we had Jason Haddock's because he was also in that picture when we were all at HusekCon back like 10 years ago when I started working with him and Jason Haddock's and David Nestor and all that stuff.

SPEAKER_02So I wasn't in that picture because I wasn't cool enough.

SPEAKER_03But you didn't work at Fortify with us at the time. I tried to work with you and then you left HP right when I moved over there. I was taking it personally. Well, anyway, enough personal history. Let's go talk to Daniel, man. Let's go. All right, Daniel. I think this is the first time you've been on the USEC cast, not the first time, obviously, we've met, we've worked together, been friends for a really long time. Um, really glad to have you on, man. How are you doing?

SPEAKER_01Yeah, thanks for having me. Doing well, just back from Vegas.

SPEAKER_03Yeah, same here. Same here. Sam is back from Europe not too long ago, but so I think he had more fun than we did in Vegas, probably.

SPEAKER_02Pretty sure. And unlike you guys, I didn't want to come back.

SPEAKER_03So probably yeah, we were definitely ready, I think.

SPEAKER_01So yeah, I did 10 days.

SPEAKER_03Oh, no, thank you. You did longer than I did. Well, glad to have you on. We will get into all things unsupervised learning. Well, I should say that Daniel is the founder at Unsupervised Learning. Daniel's done a lot of stuff over the years. Daniel and I worked together way back in the day at Fortify, known each other through the community for a really long time, been friends for a really long time. So, really happy to have you on. You've done some great stuff over supervised learning. So, we're gonna we're gonna get into that and talk about that in your Daniel segment here in just a little bit.

SPEAKER_03But before we do that, we're gonna jump into the HueSecCon segment. So, Sam, let's let's talk HUSECCON.

SPEAKER_02Yeah, so user group come. There's not gonna be one in September, but we do have one in November. And I think you know who's coming to speak at that one.

SPEAKER_03Yeah, Tommy P from from Layer X, he is a strategic solutions director over there. I've known him for a long time, worked with him way back in the day. So he's coming on. I'm excited to have him. If if you don't know, the reason we're not having one in September, everybody, is because we got HuseCon going on on September 30th and October 1st. What? What? Yeah, something's something's happening there. That's kind of crazy. Other announcement we've got cybersec media. So if you are not familiar with it, we've got all of our content on cybersec media.

SPEAKER_03This podcast goes up there, all of our different videos from all of our conferences go up there. We've got a lot of news that goes up there, we've got blogs that are going up there, we've got a lot of good stuff. So go check that out. It's at cybersecmedia.com, cyber without the e because we're cool like that. Sam's tired of hearing me say that, but I say it anyway. And if if you go out there and log in or go out there and connect up, you can get on our newsletter and you'll get a weekly roll-up feed of all of our stuff. And you can control your how often those come through on your feed.

SPEAKER_03And if you want to become an advertiser or a contributor out there, reach out to media at CSC group llc.com. We'll put that in the show notes as well. But we are looking for anybody who wants to advertise there or who wants to contribute. If you want to do do some blog posts, stuff like that, let us know. All right. So we talked about HughSightCon. What we got going on?

SPEAKER_02We got we got good stuff good there. The tickets are flying. That's awesome. Great to see everybody that took advantage of the lower price point before the the price went up here. They're$125 now. It's still a great price for as much content as we have for y'all over two days. So um, and just a reminder that the time, the starting time is going to be different, the format's gonna be a little bit different. We're gonna do the keynote a little bit later in the morning, and we start at 9 30. So you got to get there early, make sure you get in, get your badge and bag and everything. And then the talks in all the 10 tracks will start at 9:30 in the morning.

SPEAKER_02The first 100 people in each of those tracks is gonna get a free limited edition gift, surprise cool thing that Lauren and and Mary brought out for us. So we're excited about that. We only have a thousand of those, so it's the first hundred in each in each of those tracks, and we are expecting almost around 3,000. I mean, right now, the way ticket sales are going, we're definitely gonna hit way beyond that mark. So, but the kickoff officially, the official kickoff with keynote will be after lunch at 1:30. You're not gonna want to miss it. We're gonna be in the General Assembly Hall, and we're gonna have a ton of updates for you.

SPEAKER_02All these things that we've been building up and talking secretively about and giving the you know, the little hints to, we're excited to finally give that out and and get everybody excited about the future of what we're doing. And then the opening keynote, Dmitry Opilovich, he'll be at 145. We're super excited about him. And if you've not read his book, A World on the Brink, he'll be there for the book signing as well. And then from a sponsor perspective on that, so as of this recording, we have three exhibit sponsors, 17 anchors, one add-on left available.

SPEAKER_02So you can check out Houstonsetcon.com slash prospectus or reach out to sponsor Wrangler at HoustonsetCon.com. Exhibitors will get a first dibs on all 2026 events and other marketing opportunities as well. And so, last thing here, we need we need mentors. We're still looking for the mentor program. So expanding that to a year-round program. We're gonna be this is a big opportunity for you to give back to the community. And uh, you know, it's so if you've got some time, it's not a huge amount of time to invest, but it's super important to help us grow the community. Go to Houston HoustonSetcon.com slash mentorship to get more information.

SPEAKER_03That's it. Go get your tickets. We got a great show. Like I said, we've got great keynotes. Uh the timing's a little different, so just make sure you pay attention to that and get out there and get those special gifts and those 9 30 tracks. Really excited about that. I think everybody will get a kick out of those. All right, now that everybody is up to date on HughSecCon, we are gonna talk about our article of the day. So I was listening to the Risky Biz podcast, Patrick Gray, Adam Boileau over there the other day.

SPEAKER_03They had a really good, they brought one of their producers on as well to talk about this, I think. So Google says it found AI-based bug bugs or 20 security vulnerabilities using their AI-based bug hunter. So this is, and and Daniel, I you know, chose this one on purpose because I know you're big into AI and that's kind of a lot of your focus these days, but really thought this was one was interesting. They've kind of got their a couple of groups combined to to try to start finding flaws right now in a bunch of open source software because that's the easiest one to go grab.

SPEAKER_03I think they went after FFmpeg and something else, I can't remember. So they're basically using that to go run through, find a bunch of bugs, see what they can find. What was interesting about the conversation that they were having on Risky Biz was the kind of the the arms race that looks like it's gonna go back and forth. Because obviously the bad guys are gonna use AI to try to find this stuff and try to take advantage of it, you know, maybe automating the process of finding the bug and then creating a an attack for it.

SPEAKER_03The good guys are trying to find it and patch it. And I thought it was kind of an interesting discussion back and forth on the arms race of how this might flesh out over the years. And before you, Daniel, before you commented on it, I also wanted to ask you because I had a kind of a thought on this and see what and you talk to however you want to about it, but AI, so we've got AI coding more these days. AI learns from crappy code, so it gives you crappy code a lot of times.

SPEAKER_03But the more bugs we find, and I know this is not just a straight line logic here, but because there's a lot of nuance to this, but if over time, if we're fixing bugs, then AI learns from good code and should produce better code in the long run. To me, that means the good guys ultimately win, of course, unless the bad guys are messing with AI to get it to still produce bad code, that kind of stuff. So that was my thought on kind of the the timeline of this. Thought it was really interesting how it could play out.

SPEAKER_03So I wanted to get your thoughts on it.

SPEAKER_01Yeah, I think it's a huge problem that AI is building bad code right now. I think the biggest reason for this is that kind of the current version of AI is built on kind of stack exchange, stack overflow, you know, the old forums of code. And you would have like the green check mark for how to do input validation or something, and it would be like the worst code snippet ever. And like millions of developers have copied and pasted that and put it into their code, and that's of course most of the samples that the AI saw, right?

SPEAKER_01So there's a whole bunch of examples of that. The other thing is AI is just super like unevolved and unadvanced right now. So it's like it's just early days. So it doesn't really understand how to build software. It's kind of just getting lucky when it assembles something and get it could do it very fast and it works and it looks good. But it's not, it's not engineered software. It's assembled.

SPEAKER_04Right. Good point. Yeah.

SPEAKER_01And um I think the the move to being engineered software, I don't think it's gonna take too long. I think it's just gonna be gradual over the next, you know, one to whatever three years or something, where most software being built with AI, I think is gonna get pretty good. As far as the offensive defense, I I think, and this is largely what I'm gonna talk about in the talk as well, is like it it turns into this attacker-defender where you have you basically have attacker automation going against your defender automation.

SPEAKER_01And that's the fight. And the question is like, well, who's going to naturally have more advantage there? In the beginning, I think the attacker has the advantage because they can hear about a new AI thing on Sunday night and they could try it on Monday morning. Whereas a defender team has to have 20 meetings and propose a different change to their defensive system. Oh, we're gonna do this with AI. Oh, cool, let's have nine meetings about it. Let's put it into our planning cycle, right? So it's like anything that a defender, yeah.

SPEAKER_01I mean, you have to sell the idea, then you have to sell a plan, then you have to actually build it and find the people to build it. So the attacker just moves so much faster. So I I think for a very long time the attacker's gonna have the advantage. Now, ideally, defender is gonna have more information about their systems and it's gonna be more real time. So ideally, the defender wins the context war. And so if the defender gets up to its highest level and the attacker is at its highest level, I think the defender has an advantage.

SPEAKER_01For example, like Google, I think Google is probably going to be extremely good with agents, as good as their attackers, but most companies not.

SPEAKER_03Well, I mean, Google being a company that values innovation and and speed on from a business perspective, you think they may be quicker to adopt stuff and not have to go through the the rounds of approval, or how are you?

SPEAKER_01No, no, it's it's more so that they're just on top of their game security-wise. Like it when they get hit with something, they're I would say the fastest company in the world in being able to say, okay, update that across all of our systems. Yeah, now if any of our systems in the world get hit with that, the defense is in place. Kind of like the Borg. I don't know if you remember the Borg. Absolutely. I don't know why.

SPEAKER_03I couldn't be on this podcast without knowing what the Borg through the board.

SPEAKER_01Yeah, so you remember like you would hit hit the Borg with a phaser and they'd fall over. And the very next one, it doesn't matter where it was, it would block that one. So all of the Borg got updated with that shield frequency. And and I I definitely see Google as as kind of being the closest thing to that.

SPEAKER_03Google's a hive mind, is what you're saying.

SPEAKER_01It is, yeah, yeah. Really, really efficient.

unknownYeah.

SPEAKER_03Well, I mean, that's that's a lot like EDR, anything else. When they any of them that say we we see one attack in one place, we update across the board, could cause mass disruption, as we've seen in the past, but generally that's the whole the idea. Anybody, any customer of that's using it. So, but Google being able to do it themselves in their own infrastructures is strong.

SPEAKER_02So yeah.

unknownYeah.

SPEAKER_02Well, back to that article, I want to say that the good the if I'll take put a positive thing on it, like it's good to see it being used for something good. Like we talk as we are as security professionals, right? It's all doom and gloom a lot of times when we talk about this stuff, but leveraging AI to do something good, basically, and and work to find these things that we can fix them. And that being said, I know I know you're you're you're not wrong at all, Daniel, about having 20 meetings before you can fix stuff, especially in the larger organizations. But I feel like it it seems to me like some of these larger attack organizations would have similar type of reporting structures, et cetera.

SPEAKER_02And probably while they may be able to act faster, I'm sure they have a few meetings too, wouldn't you think?

SPEAKER_01I'm sure there are some that are like that, that that do have some meetings and have some sort of process that they have to follow. I'm more so thinking, and and this is just intuition, I'm I'm not 100% sure about this, but I imagine most attacker groups being fairly small. Yeah, being being 10, 20, 50 people, right? If not five people. And so if if that's just the majority, I think I think they can move very fast.

SPEAKER_04Yeah.

SPEAKER_01The other cool thing to say about this whole tech thing, so you there's Expo, which was mentioned in the article. So they they are a company that's doing automated bug bounty and they've they've reached the top of Hacker One now, both US based and globally. Then there was the AI XCC competition in uh DEF CON. And this is literally a government uh competition which ran for multiple years for who could who could build software that just scours the internet looking for open source projects and fixes them.

SPEAKER_01So it doesn't just find vulnerabilities, it finds vulnerabilities, writes fixes that actually solve the problem. Yep. And the government, this is DARPA, they put this on. So it's to me, it's like it's like old Bell Labs or old DARPA stuff where it's like really benefiting society. And and the coolest thing is it was all released as open source.

SPEAKER_03Well, that was part of the requirement for it, right? Yeah, you had to yeah, if you were gonna be part of that competition, then you had to release it as open source, which yeah, I thought was fantastic. Like nobody is gonna take this and just keep it for themselves, it had to be released for the good of the community. I thought that was fantastic.

SPEAKER_01So yeah, yeah. And so I I think we're definitely seeing this automated agent stuff. It's real, right? It's real, not theoretical.

SPEAKER_03No, and that was that was a big thing in risky biz. They kept talking about it. It's like, you know, two years ago, even a year ago, you're kind of, you know, you kind of go, Yeah, yeah, well, I'm not sure. I, you know, maybe that may be a marketing thing. And then I see more, I see more curmudgeony stuff still these days from certain people on like LinkedIn and Twitter and everything else, X, whatever, and there's still curmudgeony about it. And like this is not, it's like, guys, y'all need I I get you get aggravated when some somebody, you know, on the on the marketing side, on the vendor side or something goes, we've got AI and we're doing all this cool stuff, but that this stuff is not it's not fake anymore, or it's not just marketing, man, it's real.

SPEAKER_03So well, Daniel, thanks for your comments on that. Um, I just super interesting to me and how that's moving. And again, I know this is a big space for you um in the AI world. So um appreciate your expertise. So we are gonna take a quick break, and on the other side of that break, we're gonna do our usual uh interrogation of our guests to find out all of their super secrets from their background and why they got into cybersecurity. Daniel, I think, produces good opsex, so we'll he'll be okay.

SPEAKER_03And uh then we'll talk about your talk that you're gonna be presenting. Uh, very excited to have you come out to HughesetCon this year. Gonna be good to good to see you. Uh so let's take a quick break, and on the other side, we'll get into it. And we are back. Hope everybody enjoyed the break. So, Daniel, give us a rundown of why you do what you do. What got you into it? You know, as far back as you want to go, you know, what what influenced you to get inside?

SPEAKER_03I know you got a very inquisitive mind and you're you're a smart dude, but what made you turn all of that to what you're doing today?

SPEAKER_01Yeah, so out of high school I joined the army. So I was military for a while. And uh when I got out of the army, yeah, yeah, 101st airborne. So when I left there, I went to college where my girl was going to college, and I was gonna do pre-med actually. And I got a job at the computer lab and started building websites for money. And uh everyone was being attacked. There was lots of worms propagating the internet at the time.

SPEAKER_01And university networks, at least in Georgia, were there flat IPs, they're all public IPs, every single desktop. So there was someone there named Jason Orm. And he was like, Oh yeah, we're being attacked, blah, blah, blah. He's like, Yeah, check the server. And I'm like, what's a server? And he tells me what a server is. And so that was that was total, total gateway drug. And the moment I asked what a server was, and he answered, I was no longer pre-med.

SPEAKER_01Um, and just went directly into computers and not and just kind of skipped computers and went right into security. Because he's like, hey, so-and-so, you know, Windows boxes down. And I had just learned from Ken Swain, I started taking a computer class right after that, and he taught me about Linux. So I started installing Linux all over the campus to defend against all these NT4 boxes and Windows 98 or whatever it was at the time being taken over.

SPEAKER_01So I ended up like shoring up the network. I installed a new DNS server with the DJB DNS, and like within a couple of months, I was the head of security for the university, got my own office. Whoa. Yeah, I got I still never had an office after that. It's like this kind of this happened immediately. So I was just completely into it. And of course, I switched my major and started studying that. And then after that, just a bunch of different security roles.

SPEAKER_01I went right into basically firewall engineering right after that, and then directly into pen testing. And then that that was kind of that was kind of the launch.

SPEAKER_03Wow. And when did you start that? When round about what year were you starting that?

SPEAKER_0198.

SPEAKER_03Okay. So I'm trying to remember when we connected in the blogosphere back in the days. That'd have been like mid-2000s or something like that. Yeah. You were already in it for getting going on close to a decade, maybe.

SPEAKER_01Yeah, that might have been like 2000. Yeah, maybe like six or seven. I don't know.

SPEAKER_03Yeah.

unknownYeah.

SPEAKER_03I was uh security manager at the time of that psychiatric clinic. I think that's when we started connecting and met you at RSA and stuff like that, if I remember correctly. So did the bloggers meet up, all of that stuff way back in the day. So well, yeah, I mean, there's a lot of people. It it's I find it interesting how we talked to some folks who did that whole, you know, war games type of thing where they were doing bad stuff when they were in high school, you know, and their dad got them a computer or anything like that. And then there's people who really didn't have any exposure to that stuff, like to your point, until college.

SPEAKER_03And it's it's either way. I mean, I you know, it's no gatekeeping involved, right? Who however you got to where you are, you've obviously been extremely effective at what you're doing. So love the story. I similar story when I I mean, I got exposed when I was a kid to computers and stuff, but I was way out in Mississippi and couldn't, you know, didn't have any freaking background. So Couldn't uh hack in to get on BBS's from any in the mid 80s. So it wasn't until college that did the same thing. I got exposed at at Tomball College up here and started getting into it, taking computer classes and all that.

SPEAKER_03So that's that's a really cool, cool background. And to go from it, you know, up in the airs to whether you'd have been more effective as a doctor for saving lives than security. What do you think about that?

SPEAKER_01No, I think probably I think tech, because doctors usually don't get online and and try to help other people. Yeah, usually. Nothing against them, it's just that they have a practice, they're dealing with physical people in a physical location. I mean, maybe if I had written some book about cholesterol or something like that, maybe I don't know.

SPEAKER_02Yeah, well, we're all lucky for it that that happened, but I think the the the interesting thing there, the the the the nugget is you saw something, you went to fix it, and that's how it starts, right? So showing initiative is my point, you know, show initiative, get in there and start working on it, and people notice, and that's really cool.

SPEAKER_01Yeah, you know, one thing to mention is so one of my mentors there was Ken Ken Swain, who he was the one who showed me Linux and Google. I was using Northern Lights as my browser or my search engine.

SPEAKER_03I don't even remember northern lights. I remember like web crawler and stuff, but not northern.

SPEAKER_01Yeah, yeah. But the one who showed me what a server is, he left the job like uh one or two months after that. So he like literally taught me how to Google because I would just ask like kids why, why, why, why. Like, how do I do that? How do I do that? How do I do that? And one day he sits me down, he's like, Hey, watch this. This thing that you just asked me, I'm gonna go search, I'm gonna find the answer, and then I'll know the answer. And he was a nice guy, so he was just like giving me a hint, but he left right after that to go work at Fermi Labs.

SPEAKER_01Oh, wow, he still works there.

SPEAKER_03Whoa, wow, yeah, yeah, that's a long time.

SPEAKER_01Yeah, yeah. And he's been, you know, promoted multiple times, but um yeah, so this guy teaches me what a server was and bails and hands me the whole lab, which is why I had to accelerate so quickly. He's like, Yeah, this is all yours now, and make sure things don't go down. But yeah, you still got the same job.

SPEAKER_03Well, I was going two different ways with him showing you Google. Either he was sick of your questions and he was like, Send this kid over to Google so he'll quit asking me stuff, or I know I'm about to leave, so he needs a resource because I ain't gonna be here to do it for him. Maybe a combo, probably both, yeah. No, I mean that's what I said about your inquisitive mind because you you've always asked a lot of questions, and then I've learned a lot from you over the years as well. So it's good to have that inquisitive mind in this. You got to in this field, you can't not be inquisitive. If you're not curious, then what are you doing here?

SPEAKER_03Yeah, totally. Yeah, all right. Well, thank you for that background, and let's jump into your presentation. So, obviously, without giving away everything that you're gonna be doing, let's talk about what you're gonna be doing at HughSecCon. So, the title is Killer Context: How AI Will Eat Security and Software. So, I think that's a little play on the software is eating the world, I'm assuming. That's right. Yep. So get get into it, man. What are what are you gonna be doing there?

SPEAKER_01Really, it it's highlighting this whole concept of context, which I feel like when I wrote the title of the talk, it was not as well known as it is now. So maybe I don't know, maybe a title change or something. But I feel like the more context that you have about something, the more the easier it is to do. So I think what we end up having is is essentially world models for things. And we have to keep those world models updated as possible.

SPEAKER_01So attackers want a clear world model of the target they're looking at, and defenders want a clear world model of obviously themselves. And the more up-to-date that it is, and the more thorough it is, the better it is. And my favorite example from this is you have imagine a military leader. Way back in the day, there's no satellites, there's no advanced technical intel, and it's like a general or something. And the general is like, where should we attack?

SPEAKER_01And you know, what do we know? How big are the troops, the size of the troops? Oh, we don't know. Well, what about our troops? Oh, we're not sure. We we're cut off from the lines. How much food do we have? We don't know. So a whole bunch of unknowns. You essentially need a genius to navigate in that much uncertainty. Okay, so this is the rough argument. Now, what happens if you have real-time data of everything? So you could see every troop of your own, every troop of theirs, you could see all the weapons are highlighted.

SPEAKER_01You know exactly what direction they're going at all times. And I imagine having this general call over a private who's like mopping the floor. And it's like, hey, private, what do you think we should do here? And he walks over chewing gum and he's like, I don't know, shouldn't we blow that up? Because it's got a big red thing around it that says, blow this up. Because it has the most important weapons inside of it. It's like the the most important thing to the enemy. And he's like, Shouldn't we just blow that up?

SPEAKER_01Isn't that like all their weapons? Wouldn't we just win? And he's like, You're an absolute genius, Private. You're an absolute genius. The context makes the required expertise come down.

SPEAKER_04Yeah.

SPEAKER_01So what what if we knew the state of all the storage that we have? All the different configs, we could see which ones had open S3 buckets or whatever. If we knew the state and what was wrong with the state, and we could clearly see it on a giant board, then you don't need security wizards anymore. It's just kind of obvious. So I think the future of tech in general is creating these deeply rich and real-time world models where you just you can see everything and you understand it.

SPEAKER_01And the whole AI play here is like AI is not special, it's not magic sauce, it's just a whole bunch of eyeballs and a whole bunch of voices and a whole bunch of hands. So imagine sending millions or billions of agents out into your environment, and all they do is report back. They just report up through a chain to update this dashboard to tell you what the exact current state is.

SPEAKER_03Yeah. Well, I think we're already seeing that with what some people are were showing off on the whether it's effective or not at this point. I don't know, but we're already seeing that with some of the, I don't want to name them necessarily, but talk to two or three on the black app floor.

SPEAKER_04Yeah.

SPEAKER_03That were specifically that. It was all agentic. It was now they're the level of where they were getting to right now is level one and level two analyst type of work where they had a lot of agents pulling data from a lot of different sources that you would typically have and just replacing SOAR at this point in a little way in a bit where it's automating that stuff and giving those reactions or those, I guess, reactions, those actions that you would take normally in doing all of the lookups to see if something was real, but they they weren't taking actions necessarily at that point to start.

SPEAKER_03I mean, they could, but they weren't touting that. It was more the the gathering that a typical analyst would do using agentic AI. So we're already seeing some of that going down that path. So is your contention not to get too crazy or theatric or dramatic about it, but is your contention like this once we get this that it's just gonna take care of the security problem and we're not gonna need security people, or how do you see that?

SPEAKER_01No, first of all, this is hard. So it's it's gonna take it's gonna take forever to do this. And the bigger the org, the harder it is. And the other thing is what does it mean to get the state of an AWS environment? You've got nine accounts, you know, millions or billions, and these are containers, Kubernetes are coming up and down constantly. Yeah, so even polling the stuff and having a real-time view of it, that is so much data that just overwhelms AI.

SPEAKER_01So you need this combination of regular tech and AI. And this is why I think AI is going to go absolutely crazy straight up into the air. It actually has nothing to do with the intelligence of the models. The models are already smart. I don't think they need to be that much smarter. The problem is the orchestration of getting terabytes of data, but I want to update it every minute, every second. That is just massive orchestration.

SPEAKER_01You still need security people, though. You first of all, you have to design the system. You you have to know what questions you want to ask. Right? Once you have the data, if you're not a security person, you don't necessarily know what to look for, even if you're staring at it on the wall.

SPEAKER_03So but doesn't AI also figure that out after a while?

SPEAKER_01Yeah, I mean, eventually. Yeah, very long term, I think you need fewer people, but what are you thinking about from a timeline? I mean, if you if you had to guess, if you if you dare to guess to where yeah, I mean, I think today we are already need fewer people. We will need fewer people. I just think it ramps up pretty slowly. It also depends on how good we get at doing this AGI thing, which which I define as just the orchestration, not the super intelligent AI.

SPEAKER_01It's really well coordinated AI, is what I consider to be AGI, if it's good enough to replace an average knowledge worker. If that happens, my my year for that is 2027. If that happens in 27, roughly, we'll need a lot fewer people from that. I'm actually super concerned about this. Wrote a post about that yesterday. But yeah, I'm really worried about the impact of that on so many different jobs. But but ultimately, yeah.

SPEAKER_02The challenge I was gonna say though, on all this, though, is we talk constantly about how we have a lack of, you know, we we have a deficit in people joining cyber, right? And getting those people into the workforce. And yet you gotta you gotta figure out some way to fix the to fill the gap, right? I mean, and that drives technology, and that's why we're at these kinds of things, is because they there's a need, there's a need for it. It's hard, but it happens with every industry at some point or another. The the key here is just to be prepared for it and you know, and and start working towards basically understanding it and being part of the part of it, part you know, understanding AI, understanding how to leverage those things, right?

SPEAKER_01Yeah, uh I've always wondered about this security person shortage. It doesn't really make any sense to me because I I've been a hiring manager multiple times. If if you if you have money to hire people and you need people, you just hire people. So if if they're saying there's a security shortage, there's not enough people, well, it's simply not true because there are you know hundreds of thousands or millions of people who want to do security work, they're sitting there, they're not getting calls, they're sending in resumes, they're not getting callbacks.

SPEAKER_01So I don't know. I there's some kind of real disconnect here about those numbers and about the claim.

SPEAKER_02Yeah, you surf LinkedIn any given day and you see all that, right? There's a lot of people looking.

SPEAKER_03And well, there's there's a lot of people looking. I think there's still a lot of people hiring. I mean, I know this is a whole nother discussion point, but I get really really concerned and like just critical of the numbers because for one, it tends to go up every year. It started with one million and it went to two, and I think now we're seeing five and six million. And those numbers are generally all published by organizations that I will not name on here for sake of fear of pissing anybody off, but they're published by people who have a very uh big concern about getting more people into the industry, and that's how they make their living.

SPEAKER_03Uh, so they create those studies in some ways out of whole cloth or or at least swizzle in numbers. So, but again, that's a whole nother discussion. Maybe we could do another podcast on that, Daniel, sometime.

SPEAKER_01Yeah, yeah. This is like the most common thing that's thrown around, probably in security. It's like oh, by next month we will need 75 trillion new security analysts.

SPEAKER_03It's like well, how's how's the number keep multiplying by that that hockey stick definitely takes effect really quick because the people keep it's like when they start seeing their numbers get hurt because not as many people are going after certifications to get into security, and AI is absolutely going to have an effect on it. It's like, oh, once you create an organization, that organization's goal from that point on is to remain in business. So it's like, yeah, that's just a self-propet perpetuating thing.

SPEAKER_03Yep. Well, man, thanks for joining us. Always great to talk to you. Great discussion. I mean, I now too put too fine a point onto it, not trying to blow smoke or kiss your butt here. I mean, you do great work, dude. I've always appreciated and respected for what you do. So thanks for unsupervised learning. It's just straight. You've done uh you've done a great job with it. So thanks for sharing your knowledge with us here and coming to Houston.

SPEAKER_02Go sign up for that newsletter. I'll put it in the show.

SPEAKER_03Yeah, we'll we'll do a link. We'll do a link. I mean, it's it's worth your time. Go sign up.

SPEAKER_01So yeah, I appreciate that. Thanks for having me.

SPEAKER_03Yeah, man. Great to have you, Sam. Any parting thoughts, sir?

SPEAKER_02No, this is good. Big it definitely jogs a few brain cells about other topics, right? And that we could have taken this way off the rails on some of that conversation, but oh yeah, no time today.

SPEAKER_03Oh, we can go have a political discussion after this. That that's that's always fun. Daniel and I have been involved in a few of those over the years.

SPEAKER_01Yeah, we have.

SPEAKER_03Too fun. All right. Well, everybody, thanks for joining us on episode 49 of the HUSEC cast with Daniel Meessler, and we will see you soon, if you can believe it, on episode 50. So, all right, everybody, have a good one.

SPEAKER_00This has been a Cyberset Community Production. HUSECT Cast is hosted by Michael Farnum and Sam Van Ryder, with production and editing by Lauren Andres. Our music is by August Hunt. The views and opinions expressed in this show are those of the speakers and do not necessarily reflect the views or position of any entities they represent. This show is for informational purposes only and does not render or offer to render personalized advice. Subscribe now so you never miss an episode. You can find all of our podcasts, articles, blogs, and conference talks on cybersecmedia.com.

SPEAKER_00That's cyberwithout the e. And follow cybersec Media on LinkedIn, X, Instagram, and Facebook at Cyberstech Media. You can keep up with our conferences by following us on LinkedIn, X, Instagram, and Facebook at HughSetCon. And you can learn more about our events or by tickets at HoustonSecCon.com.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Asking too Many Questions with Michael Santarcangelo

    Nov 11, 202633 minAudio

    In this episode of CYBR.SEC.CAST, Michael Farnum and Sam Van Ryder sit down with Michael Santarcangelo, founder of ValueSmith, to discuss delivering value that others can actually recognize. With three decades in the...

    • Apple Podcasts

    Credit the people on this episode

  2. Keeping up with the Times with Jody Brazil

    Nov 2, 202630 minAudio

    In this episode of CYBR.SEC.CAST, sponsored by FireMon, Michael Farnum and Sam Van Ryder sit down with Jody Brazil, founder and CEO of FireMon, to explore how AI is reshaping network security and why traditional...

    • Apple Podcasts

    Credit the people on this episode

  3. AI, Risk, and Ewoks with Peyton Kennedy

    Oct 28, 202635 minAudio

    In this episode of CYBR.SEC.CAST, sponsored by Endor Labs, hosts Michael Farnum and Sam Van Ryder sit down with Peyton Kennedy, Senior Security Researcher at Endor Labs, to explore the hidden risks of AI adoption,...

    • Apple Podcasts

    Credit the people on this episode

  4. Not a Finance Guy: Building Companies with Bob Ackerman

    Oct 14, 202634 minAudio

    About this episode: On this week’s episode of CYBR.SEC.CAST, Michael Farnum and Sam Van Ryder sit down with Bob Ackerman, Managing Director at DataTribe, to share lessons from decades of building and investing in...

    • Apple Podcasts

    Credit the people on this episode

  5. The Wild West of Deploying Agents with Nadav Cornberg

    Oct 6, 202634 minAudio

    About this episode: In this episode of CYBR.SEC.CAST, Michael Farnum and Sam Van Ryder sit down with Eve Security founder and CEO Nadav Cornberg to explore the challenges of securing agentic AI. They discuss AI...

    • Apple Podcasts

    Credit the people on this episode