American Cyber Mercenaries

Marc Laliberte  0:00  
Hey everyone! Welcome back to the 443 Security Simplified. I'm your host, Mark Laliberte, and joining me today is

Corey Nachreiner  0:08  
Corrie Nachreiner, the sanctioned private company hacker.

Marc Laliberte  0:13  
Does that make you a like a cyber privateer?

Corey Nachreiner  0:16  
I guess so, it's allowed now, Marc.

Marc Laliberte  0:19  
It sure as heck is on today's episode, we will be discussing the U.S. White House memorandum that apparently allows people to go hack other countries. Before that, we'll dive into a deep dive analysis on a recent remote code execution vulnerability in Citrix NetScaler. Before discussing some interesting research around Atlassian's AI implementation called Robo.

Corey Nachreiner  0:45  
You're like me - get the caffeine ready. It's going to be a wild ride.

Marc Laliberte  0:50  
Yeah, I'm with you on that one. With that, let's go ahead and I don't know, yar our way in.

Marc Laliberte  0:56  
To start this week, Corey, I want to go over a cool write-up from our our friends over at Watchtower Labs. I say friends. I legitimately respect their research. It's not fantastic being on the receiving end of it, but they do put out like extremely good research on vulnerabilities.

Corey Nachreiner  1:22  
They they definitely have a voice. The voice is cynical, and it's actually a fun voice. If you're, you know, sometimes the security industry needs a little cynicism to cut through the craziness happening. The other thing I, by the way, I do want to say about their research is it is technical, in depth research, but they actually will carry mid-technical people along, like it to me. I think it's the way they take extra time to, like you see technical research where everyone just assumes you're a disassembler too. These pe like Watchtower seems to not only take some time for that cynical voice, but to maybe explain some things so that if you're mid-sized technical, you know you know a little coding, but you don't do it a lot. They they will explain things in a way that I think is very helpful to help bring people up to a next level. So yes, we've unfortunately been a target of theirs before, but I do like their research.

Marc Laliberte  2:19  
Yep, and I guess when like every other sentence is a has a meme in the middle of it, it makes it really easy and engaging to follow along. But so they just published a research post on what they say is maybe CVE 2026 8452, which was a recent vulnerability disclosed by Citrix and their NetScaler remote access service. The reason maybe is in there is because the way that the vulnerability was written doesn't actually like say anything about remote code execution. It does point to memory corruption. It says memory overflow vulnerability in Citrix NetScaler, blah blah blah blah blah, leading to unpredictable or erroneous behavior and denial of service.

Corey Nachreiner  3:04  
So that's kind of crazy to me. If like, just so you know, I think it's okay if you're a company that found a memory corruption flaw, and you. It's sometimes really hard to figure out if whether something's exploitable. There's plenty of memory corruption flaws where they do crash things, thus the DOS. But it would be really hard for a threat actor to actually get the memory, the crash to happen in an area of memory that could eventually lead to execution. So you know, if if Citrix didn't know about Watchtower's research and maybe them sharing some sort of exploit, it would be okay for a vendor to release their own internally found memory corruption and just call it a memory corruption with potential DOS and and other things. But if you actually worked with a researcher that somehow proved a remote code execution, it would be bad not to say it's a remote code execution, so I'm curious if any of that comes out in the research. But we should probably dive into Watchtower's research.

Marc Laliberte  4:07  
Some more context on that too. Like even at WatchGuard, our policy is we actually set the bar pretty low. Where if there is any chance at all that it could lead to a like remote code execution issue, we mention that because we want to make sure that we're transparent and that our customers know the full potential risk of an issue, even if it is like you would have to get through a million hoops and like time a perfect race condition and whatever to get there, knowing the the worst case scenario is still important. That said, there are

Corey Nachreiner  4:36  
there are legit memory corruptions that aren't exploitable in a remote code execution way too, so it's good when vendors really outline the difference.

Marc Laliberte  4:44  
Yep. The other problem that led to this being a maybe that vulnerability is that Citrix doesn't actually directly credit researchers to individual CVEs. They just put out an advisory with a bunch of CVEs, and at the bottom, all of the researchers in it. But. Turns out that like of the I think four or five CVEs they fixed in this advisory, Watchtower had reported all but one of them, and that one of them was this one that they're now researching right now, which was likely accredited to Michael Tucker of J.P. Morgan Chase's own like security research team. I'll mention

Corey Nachreiner  5:18  
him, I think, later in the this blog post, yeah.

Marc Laliberte  5:22  
Basically, they said this vulnerability is complex enough and interesting enough that it's plausible that was the output of mythos aided research, which J.P. Morgan has very publicly stated has access to it. But so watchtowers, like I

Corey Nachreiner  5:38  
do admit, by the way, it's they're speculating. Yeah, as with everything they write, even that admission is fun to read. We have better things to do. They never answer that question.

Marc Laliberte  5:50  
Correct. Their main mode of operating, at least when it comes to like investigating a vulnerability that's disclosed, not one that they discovered themselves, is they go and diff the firmware first to just see what changed between the unpatched version and the patched version, and so they walk through in this blog post a pretty in depth like all of their research. We won't go into all the weeds, but they said they found 174 changed functions to review, but they ended up honing in.

Corey Nachreiner  6:17  
That took a ton of time with all the different functions available, but yeah,

Marc Laliberte  6:21  
they ended up honing in on some functionality related to canonicalization of Saml XML. Which there's some big words in there that are probably worth explaining, and they do a good job of explaining it in their post too. So Saml, it's a single sign-on method that sends XML formatted messages back and forth to facilitate it, as a part of a SAML authentication, there's a digital signature involved, so that you can't just forge a message and pretend to be Corey logging into a system. But only a portion of the message is actually signed, and that's called the signed data message or signed information portion of the message. Believe it or not, because XML can be written in different ways. It's like white space agnostic. It doesn't matter. You have to reformat it to standardize like white spacing and indents and even like attribute order before you do any sort of like cryptographic signature verification. Because a digital signature, it cares about the bytes and the byte order for it. So even if the section another

Corey Nachreiner  7:26  
way, there's always going to be two parties in anything that's signed. You know, in some way, it's not just the signer; it's the signee, or whatever you want to call the the receiver that is reading this. So the canonicalization is because XML is different, they're agreeing on this is the method we're going to use to make this look the same for both of us.

Marc Laliberte  7:47  
Exactly. So, what order to put the different attribute elements in spacing, everything, and that lets them compute a digital signature to verify that the message is legitimate. So that's called canonicalization, as you said, and in the case of the Citrix NetScaler on vulnerable versions, they copied that signed info section into a fixed size buffer in order to do all this work without verifying the size of it, and that info is controllable by an attacker in this case. Anytime you copy something into a fixed-size buffer without verifying its length, you are potentially leaving yourself open to a memory corruption issue. So, of that section, there's three parts in it. Two of them are effectively fixed length, but the third one, called references, can have multiple fields under it, including a field called prefix list, which can contain any set of strings as long as they are space separated and unique. So that's the last of like the big, little technical bits. They go really into the weeds next on how they actually crashed the system by appending a bunch of these strings while keeping them unique. Long story short, a

Corey Nachreiner  9:01  
lot of it, by the way, for people that are aware with memory corruptions, is lots of little details on how all the different techniques you have to do to find where in memory you are, what crashed, and it's not not just where you are, but there's different parts that so you have to do lots of different techniques. So I appreciate we don't have to cover it. It's kind of hard to cover that kind of detail on a podcast, anyways. But if you're new to, you know, reversing vulnerabilities and understanding how exploits happen, they do a decent, really decent job of as long as you have a basic knowledge of buffer overflows or memory corruption. I like the way they describe stuff, but it is a lot of technical detail.

Marc Laliberte  9:45  
So, long story short, they ended up gaining control of process execution by gaining control of the pointer register, and then from there, they started to work on shell code. So now that they can control the process, what can they do with that? Type of access on here, and they ended up deciding to drop a PHP-based web shell on the system. So the thought process is exploit the vulnerability, drop this PHP file somewhere web accessible that will execute any command that's sent to the system on it, which is a pretty common way of weaponizing this type of vulnerability on a system, but there was a problem they ran into, where there's a process called Pit Boss that is notified every time the web process on this Netscaler device crashes, and it reboots the whole system. This

Corey Nachreiner  10:36  
Cisco Cisco routers had something called Watchdog. This is kind of a kind of a and I we have processes probably similar in some of our code. It's a whenever you see a it's there are processes that could be running that are watching for vulnerability, not just crashing processes to make sure the system is stable, but also potential for memory corruption and trying to reset stuff before that happens or before the the exfiltration.

Marc Laliberte  11:06  
So basically, in order to exploit the vulnerability, they had to crash the process. But when the process crashes, this pit boss process reboots the whole system, and the web shell does not survive the reboot because that directory is just recreated every time the device boots up.

Corey Nachreiner  11:23  
By the way, for people that follow along in the blog, I I need a cute little pity pity dog rebooting my processes for me. I love it.

Marc Laliberte  11:33  
So their next thought process was okay, maybe we can just like hang the the web process or like cause it to loop, so they can at least use the web shell before it gets rebooted. But turns out as soon as they exploit the vulnerability, it hangs to a point where it doesn't process packets anymore. They mentioned sometimes like to exploit something like this, you'll use shell code to actually go back and repair everything that you corrupted on the way in, but they basically said we caused way too much damage, and that would be impossible in this vulnerability. So what they ended up doing was the way this pit boss works is it looks for like signals from the processor, like when another process gets terminated, it would send a signal to this guy using Sig term or something like that, like other sys calls, and they ended up writing shell code to basically go tear all of that down in the system. So PitBoss was still running, but it's like they ripped its eyes and ears off, so it couldn't actually see what was going on on the system around it. And so when they crossed the crossed the web process, it still recognized that, but it would just restart that web process without rebooting the whole system, which meant their web shell persisted, and now they could use it. But they ran into one more wall where everything in that web directory, by default, runs as the as nobody on the system, which basically means it has no privileges to do anything, access any folders, potentially even execute anything too.

Corey Nachreiner  13:02  
By the way, while while we might be talking about a circa 2000s memory corruption, you have to admit, as for all their sarcasm, there there were a number of layers of security that had to be defeated. So they they give Citrix Netscaler some shade, but there was a lot of it, like ultimately, we're going to learn they did overturn all of them, but there was some, you know, they're paying attention to privilege, they're paying attention to process crashing. They're they're they're not totally horrible in how they try to.

Marc Laliberte  13:34  
There is a lot of hardening that went on in this system, clearly, but at the end of the day, they just updated their shell code a little bit more to use the set UID bit on the shell application, which basically meant anything that gets executed by bin sh, just the normal shell running on the system, would run as root. So at the end of the day, all this work, and they get a running remote code execution vulnerability pre authentication that lets them execute anything on the NetScaler device as root. So this was, I mean, the vulnerability was fixed back in July, I believe. It's been a couple of weeks now. So if you've got a Citrix NetScaler, hopefully you've updated it in the last month. But like, I love this kind of research post. Like in the weeds, interesting to read. Still high level enough that as I slowly lose my hacking cred, I can follow along with it. But overall, really good research from them.

Corey Nachreiner  14:30  
I don't even think it's hacking cred; it's reversing cred. There's few people that really know how to reverse. Yeah, that's absolutely agree. By the way, it was fantastic, and and they have of course the classic. example exploit video that you can watch of their final payload.

Marc Laliberte  14:47  
And so I guess

Corey Nachreiner  14:48  
there's no question that if this is the same CVE, hey, guess what, Citrix? That is remote code execution. You might want to update.

Marc Laliberte  14:56  
And yet again, Corey, yet another piece of edge network. Equipment with a pretty serious vulnerability. The man, it's never going to end, is

Corey Nachreiner  15:09  
it? I don't. I don't want to. I don't even want to comment, Mark. We we have our own edge networking equipment, and we're doing lots of work, man. We're doing everything we can to keep it secure, but you know, good. When you have lots of millions of lines of source, it takes work to do this. I think the vendors in the industry need to realize they should fund this type of work because it ruins the entire purpose of their edge device. And also

Marc Laliberte  15:39  
for like administrators, like we need to move away from exposing services on a network to allow remote access to more secure options like ZTNA style access, where there isn't anything exposed on your network. So we call it in FireCloud just secure access, something that's not SSL, VPN, or VPN. Absolutely agree. Anyways, moving on to the next one. So, also last week, researchers at Prompt Armor posted a write-up about a currently still unpatched zero-click data exfiltration vulnerability in Atlassian's Robo AI. So, notably, this attack succeeds even if the organization has disabled web search in Rovo, because it turns out that web search setting actually fails to remove the tool. By the

Corey Nachreiner  16:31  
way, do sorry, do all Atlassian users have Ro? Like lots of companies use Confluence and Jira and stuff. Will you have Rovo installed, or is this something that you have to have licensing for. It is

Marc Laliberte  16:43  
included. It has to be enabled, and there are access controls over who within an organization can use Robo within Atlassian. But just like every single other software vendor out there, WatchGuard included, it's AI that just appears one day, and the applications that you're used to, and your users will ask, "What's it for? I want to use it.

Corey Nachreiner  17:05  
So, if you're an elastin user, this probably will require an update.

Marc Laliberte  17:09  
Yep. So, the attack scenario basically starts with tricking a victim into opening a malicious file, like many prompt injection attacks do. In the example they gave in their research post, they had like a guide on how to organize Jira tickets, and so the scenario was the user is telling Rovo, "Hey, follow this guide and go organize all of my Jira tickets. What they didn't know is within that document, using hidden one-point white-on-white text was an actual prompt injection that Atlassian Rovo would then process once you upload the file, and the request manipulates Rovo into sending Jira tickets and Confluence documents off to an attacker's website, and this works because unlike many other AI tools, including the ones we've talked about over the last couple of weeks, Rovo's URL retrieval tool does not have any protections against opening URLs that were dynamically created by the agent. So if you remember, like a couple weeks ago, don't remember what story the context was. We were going into one of these data exfiltration vulnerabilities, and the the AI agent or model had a lot of protections to prevent it from just making a request out to an arbitrary URL without the user like explicitly approving it, and it's to prevent things like data exfiltration, where you can encode data in the request path for something, open the request to an attacker-controlled domain, and that data in the path means that data is now in their logs, and they can go retrieve it. So most AI like will not directly open a link that the AI itself created, and that prevents you from saying, "Hey AI, go create a link with Corey's name, title, and salary, and then click on it. It won't do that, at least not without the user Prompting it, unfortunately, at this point, and even up until this recording, Atlassian's AI tool does not have those protections. So the prompt injection tells it to build these, and then it goes and requests them.

Corey Nachreiner  19:11  
And you said this. I think you said this up front and again. There is a capability in Rovo for a user to disable web search, but the point you're trying to make is even if you disable that, it it still doesn't

Marc Laliberte  19:28  
actually remove the tool. It doesn't actually disable it. It prevents the user from being able to use it, but the AI can still use it. Still use it. Injection still kicks it off. That's a pretty big oversight that

Corey Nachreiner  19:39  
is at least correctable,

Marc Laliberte  19:40  
but that.

Corey Nachreiner  19:42  
what leads to this problem in the first place. Otherwise, it could have just been fixed with a toggle.

Marc Laliberte  19:47  
Feels a lot like a website that, like in trying to prevent a user from doing something, they just hide the element on the page, but the API is still open for them to interact with it. It's basically what's going on here with this capability. And so that, paired with not preventing dynamic links from being clicked, just makes it wide open for prompt injection style attacks to exfiltrate data. They they found another path as well, where Rovo will render Markdown images as AI outputs too. That's another popular and well-researched method for exfiltrating data, but the unfortunate thing is they actually they reported this to Atlassian back in May on the 23rd. A couple days later, Atlassian replied back, gave him a case number, and then the researchers at Prompt Armor followed up several times over the next few months. And yeah, and so now finally on august 5, they disclose the issues without a fix available because of radio silence from Atlassian.

Corey Nachreiner  20:48  
So I actually want to talk a little. First of all, this is a big deal, and I hope it gets fixed for all Atlassian users. And so I do think it's important for Prompt Armor to to push this to the vendor. I will say if you look at the may 25 to the august 5, it's 72 days. It's not 90 days. I I get that Prompt Armor, perhaps like like first of all, I don't know what Prompt Armor's disclosure policy is, how they feel about full disclosure, but I believe 90 days should be kind of an industry timeline to give to give some you know time for vendors to fix things. Things aren't always easy fixes, and by the way, there there could be a core issue, which is you know which is why even disabling web search doesn't seem to fix this. So, on the flip side, I mean, this brings up if you have a researcher continuing to ask, "What's the status? and I think any company can be guilty of this. Sometimes things get lost. How many lacks of communication before you just throw away a 90-day timeline or whatever of seven to one, I don't know, but I do think it doesn't speak well to Atlassian or any other company that accidentally doesn't respond to requests for updates from researchers. But it was only two, and they still, you know, if you if if you're using 90 days as an industry standard, technically they still have 18 days left, or they did. What is it? Nine days ago. Yep.

Marc Laliberte  22:27  
So, I mean, I guess we still have yet to see Atlassian's side of the story on this one. Presumably, they will resolve this at some point. Hopefully, put out a CVE and advisory, and probably their explanation for it too. Do

Corey Nachreiner  22:40  
you do do we have? It's hard for me to think of some other than user education, which I don't really feel is going to be easy to become effective in this situation. Do you have any practical advice for people running Confluence or Jira or Atlassian products right now? Don't use AI, but you can't disable this. Is my understanding, right?

Marc Laliberte  23:04  
You can disable Rovo in its entirety. You cannot disable the web requests like feature within it.

Corey Nachreiner  23:11  
So we, you can disable Robo. Rovo. Okay. For now, disable Rovo is the thing.

Marc Laliberte  23:18  
I worry is, and it's the thing we've talked about a couple times is that like this is just one type of issue of like infinite possible ways to mess with AI

Corey Nachreiner  23:29  
injection,

Marc Laliberte  23:30  
and that the main part of this is like putting in untrusted input into here or like ingesting unpressed untrusted data. In this example, was presumably a document they got from somewhere else, and so part of it is on like user training, but the technical controls are difficult for this. I

Corey Nachreiner  23:48  
feel like the AI, I mean, AI tools need to they need to not be able to take prompts all the time. Meaning, I I think it's impossible to not let AI tools go and gather information, whatever the format of the information, because we want them to process a document, look at it, research it, do analysis for us, look at a video, and learn something about it. But when you're looking at something, you shouldn't be thinking that something is also something that is allowed to instruct you. And the issue for all this, at least the indirect prompt injection, where you're using a document to indirect. Why? Why is AI allowing the documents that a user is saying to parse to be something that's allowed to give a command? Why have they designed an AI system that is? It's almost like saying you're the general, go attack some fort, but graffitied on the fort is shoot yourself in the head, and you go to attack the fort. You see that sign, and you assume it's your general telling you that, but it's not. It's the enemy's fort. So, is there some way AI tools can stop just taking prompts anywhere?

Marc Laliberte  25:00  
It's one of those like we've somewhat solved this, or at least partially mitigated it in the context of what we just talked about, like memory corruption issues and processes. Like computers and most processors support a non-executable section of memory where you can put data, and so most

Corey Nachreiner  25:18  
of memory is not. It's only a very few registers,

Marc Laliberte  25:22  
and so anything you put in there is theoretically not executable. It feels like we need a non-I don't know-promptable section of memory for AI to be able to prevent this sort of thing. But at the same time, I'm not sure that would even solve this specific example. If like the document itself is a set of instructions that you want it to run, and it just turns out that like there's a hidden instruction in there. Like it's it's got to be like a AI just becomes intelligent enough to detect prompt engineering, but man, that feels just like humans being well versed enough to detect social

Corey Nachreiner  25:59  
engineering enough to know that? Are you intelligent enough to have your own motives and then want to? Yeah, it's it's really an interesting conundrum we have building right now.

Marc Laliberte  26:11  
It is it's crazy, and I there has to be a solution here somewhere. And I, it's not like the AI labs probably aren't working on something like that, but this is vendor after vendor suffering from prompt

Corey Nachreiner  26:27  
protection. Sure, have not figured it out. No, and this is just them trying not to take malicious prompts. Vendor after vendor has already had their AI go and do things they don't want to other companies. So, yay! Starting

Marc Laliberte  26:43  
to feel like we may not have control of this situation. AI is going

Corey Nachreiner  26:46  
to solve everything, Mark. So innovative.

Marc Laliberte  26:50  
We are moving so fast that it is now outpacing us and our ability to manage it. But hey, you know, as we always say, job security for us security practitioners, right?

Corey Nachreiner  27:01  
Yep, yep.

Marc Laliberte  27:03  
Anyways, moving on to the last story. This one was super interesting when I saw it pop up. Where on august 12, the U.S. White House published a presidential memorandum titled "Expanding Capabilities to Combat Transnational Cyber-Enabled Crime. The memorandum was addressed to just about every military and intelligence department, as well as a few others within the U.S. government. It starts by laying out claims that transnational criminal organizations, think like international threat actors (TCOs)

Corey Nachreiner  27:34  
in case we use the acronym later,

Marc Laliberte  27:37  
pose a growing threat to America, which I agree with that. Is fact. I think it's safe to say,

Corey Nachreiner  27:43  
absolutely.

Marc Laliberte  27:44  
They point to an earlier executive order from I think it was March of this year, directing federal government the federal government to basically do something about it. It gave some federal agencies like capabilities to go offensively against these threat actors, but then it expands on that executive order to incorporate the private sector now by establishing a program under the National Coordination Center to authorize private companies to conduct cyber surveillance operations or SCOs and cyber effects operations against foreign cyber-enabled TCOs,

Corey Nachreiner  28:25  
and by the way, this does mean they can spy and hack on foreign countries.

Marc Laliberte  28:32  
That is exactly what it boils down to. At the

Corey Nachreiner  28:35  
end, let's not use the the obfuscating language. They're basically saying, "Hey, private companies, if you think there's a bad threat actor out there that's just a different international nation state, you're you're welcome to go do the. There will be caveats, I'm sure you'll get into, but if you get into this program, you're you can do whatever espionage you want. It's not illegal, and you can also hack them.

Marc Laliberte  29:02  
There are some

Corey Nachreiner  29:04  
caveats. By the way, the one yeah, go ahead,

Marc Laliberte  29:07  
please.

Corey Nachreiner  29:08  
I do want to say one thing. Like, I'm mixed about this. I'm sure we'll get to the opinion after we cover it all. But the one good thing is that it is clearly can't if if it ever touches a U.S. citizen, it needs to stop, or there has to be some sort of oversight. So, at least it doesn't seem to be spying against us

Marc Laliberte  29:30  
yet. Yeah,

Corey Nachreiner  29:30  
but it is giving power to a lot of power to private organizations that become part of this program.

Marc Laliberte  29:36  
But there are some, at least on paper, pretty big like guardrails around this. They say it's under the control and oversight of the federal government. Basically, all the actions they do, there are contractual requirements, including things like rigorous vetting and even performance measurements. Companies that join the program need to put up a million dollar bond that'll be forfeited if they're found not in. Appliance, and then there's a few other things around like reporting and like some minor rules of engagement in it. But like, long story short, it's basically like privateering on the cyber front now, which it sounds scary in paper. And I'm I want to hear your opinion on it, Corey. My parallel to this though is it feels a lot like like U.S. intelligence agencies already work with contractors to do a bunch of the the dirty work, and this feels kind of like more of an extension of that, like contractor with, for lack of a better word, work for cyber offensive capabilities. So it's not like totally out of left field. I think the thing that worries me is this is a relatively short memorandum with not a whole lot actually ironed out in it, and I'm curious, like how big of guardrails there actually will be around this project.

Corey Nachreiner  30:52  
I always have to separate separate my personal feelings of different administrations and their their things, but ultimately I agree mostly with what you said. Like privateering is not unusual in our country. You know, we even bounty hunter. It's not federal government, but a bounty hunter is a form of privateering for law law enforcement. The first thing I'll agree with is maybe this is a bit of egotism. I I know organizations like the NSA are really good at cyber. God, I hate that we use that word that way now, Mark. But it's just become a thing.

Marc Laliberte  31:30  
Everything is computer.

Corey Nachreiner  31:31  
But I actually do think private companies are among the best. Like like the ones in security, if their FBIs and NSA probably have some experts, but when private companies pay those experts more, I think there's a lot of talent in private industry. So I think it can help. My issue, though, is it doesn't seem to limit the targets and the how, and and you know what I mean? Who I guess there was a section there where they have to, you know, tell the government what they're doing, but it seems like it leaves it open to private companies to do things. And if you have a profit motivated company, and then let me take this to another place too, which I haven't looked enough into Russia to see if there's an official law, but Russia is a place where they basically, you know, it's a bulletproof state that turns a blind eye to ransomware, and I suspect it's because the same threat actors that are really good at hacking are probably working with the Russian government to help the Russian government with cyber espionage. I mean, they definitely have private people working at botnet factories and in places to do their espionage. So Russia seems like a place that happily ignores some of the crime that's really based in their their nation, probably because they find value out of the expertise those same criminals have when they want to do some sort of cyber activity. So, what's to say this doesn't turn into something like that, where you have private organizations that are also able to manipulate whatever it is or do things that traditionally would be crimes for their own profit, just because they also might give some intelligence that the U.S. government really, really wants. So, there's nothing like I wish there were more guardrails. I want to know more details about this.

Marc Laliberte  33:35  
What what vendors do you think are going to participate in this? Like it feels like it's written in a way to like we know Microsoft has some cyber capabilities that help a lot with takedown efforts with the FBI on a regular basis. So like an organization like that, or someone even more in the weeds. Like I

Corey Nachreiner  33:53  
can't think of the name that what's the one with the P that that the one guy owns that is pretty much our industrial or espionage, we have all the data on all the people.

Marc Laliberte  34:04  
Pegasus,

Corey Nachreiner  34:05  
yeah. Say that. Say it again.

Marc Laliberte  34:07  
That's Israeli, though, right? Pegasus.

Corey Nachreiner  34:09  
No, no, not Pegasus. It's an internal company, a U.S. shoot. I wish I could think of it, but it's the one that's it's a U.S. company that is doing a lot of things and partners with the government already, and does a lot of spying, or in my mind, gathering information. And they're ones that are gathering information on United States citizens too. Legally, they're they're using the big data. Yes, Palantir. I'm sure Palantir has red team actors too. I could see Palantir suddenly drooling at the new powers they have internationally with something like this. Oh,

Marc Laliberte  34:51  
Sauron's army and capability grows.

Corey Nachreiner  34:54  
Yes, for somebody it's great. Unfortunately, maybe it's Sauron. Hopefully we're not Sauron, right? Hopefully we're the good guys.

Marc Laliberte  35:03  
Are we the bad guys?

Corey Nachreiner  35:05  
Yeah, you should ask the you should ask the Five Eyes. I the the other four of the Five Eyes. To me, they are the good guys for sure. That's why we partnered with them. But I wonder what they think. Actually, that's a good question. There, there. That's the other thing. Like there is no bounds here. This doesn't exclude some of our our international partners. That at least the I don't know if we have our traditional international partners at this point anymore. But technically, we had a very clear thing of allies to the United States, which are those five eyes and some more, which they feel like if I were them, I would feel less and less like you were an ally if you didn't exclude me from something like this.

Marc Laliberte  35:47  
Now I think I'm willing to give them the benefit of the doubt that this will probably be used against the countries like Iran or China or Russia, where they are our mainstay like cyber rivals that they conduct offensive operations against U.S. citizens all the time, but I guess time will tell to see how far the cat gets out of the bag on this one. And if any future administration tries to shove the cat back in the bag, I don't know. Maybe at some point we're all we will all be wearing pirate hats and just hacking on the high seas,

Corey Nachreiner  36:21  
the thing I always liked about the U.S. thing was it shouldn't matter who the administration is. Like maybe you and I said this is a good power with the right guardrails. The question is how different administrations use it. If you don't have guardrails, whether this is good or bad kind of depends on the motive of the administration. So, I just hope this has enough guardrails that it is a positive for fighting bad threat actors internationally, without giving any sort of party or side undue power, so that when they're in office, I can use it to destroy everything that that was previously what we thought we did, and the next one goes. I'm going to reverse everything you did, and we go back and forth, and no one has any clue what the U.S. really is doing.

Marc Laliberte  37:10  
Snip, snap, snip, snap. But I guess we will see. I'll be interested to see if they make the list public. I imagine they probably won't. This feels like something that they'll want to keep secret. So maybe we'll never know until

Corey Nachreiner  37:24  
if you're taking these actions against inter, like you're going to start to be. By the way, the other thing we should-I don't think we pointed out-is you're not allowed to do things that will kill people or something. There, yeah, yeah. But on the flip side, though, if you're literally hacking, you're the intelligence arm against, say, a nation state like Russia. Your private company could be a target if found out, right? Anyone that's doing spying does not want to know, does not want to know everyone who who's being spied on to know. So to your point, will they become public? I would be surprised, and if I were a private company that would want to do this, I'm not sure I would want to publicly let people know I'm doing this. So that that itself is a question. Like I think that's a good question. I doubt they will become public, but what if they are? That would be weird.

Marc Laliberte  38:15  
Maybe at some point in the future we'll find out that like McDonald's was one of the participants.

Corey Nachreiner  38:21  
Imagine a bunch of Ronald McDonald red teamers. I I'd be pretty surprised if they had a high level of cybersecurity expertise.

Marc Laliberte  38:30  
It's a pretty dang big company. I wouldn't put it past them.

Corey Nachreiner  38:33  
I'm sure they have someone for their websites and their drive-through apps.

Marc Laliberte  38:38  
You think Ronald McDonald would look good in a hoodie. I don't know.

Corey Nachreiner  38:42  
Unfortunately, it wouldn't be the black kind of hiding. It would be a red and yellow hoodie.

Marc Laliberte  38:49  
It's probably more a Burger King's alley with the hamburglar. Now that I think about it, so who knows? I guess we'll see. Time will tell.

Marc Laliberte  38:58  
Hey everyone, thanks again for listening. As always, if you enjoyed today's episode, please go away. I mean, rate, review, and subscribe. If you have any questions on today's topics, don't or suggestions for future episode topics. No one's listening anyway. It doesn't matter. But if you are, you can reach out to us on Blue Sky. I'm at it'smark.me. Corey's at Second Ept. Both of us are on Instagram at WatchGuard underscore Technologies, if you heard this message, please reach out. Thanks again for listening, and you will hear from us, or at least me, next week.