# Debating the CISA Secure by Design Pledge

URL: https://securitytable.ai/debating-the-cisa-secure-by-design-pledge/
Published: 2024-05-31
Duration: 2381 seconds
Season: 2
Episode: 16
Hosts: Chris Romeo, Izar Tarandach, Matt Coles

## Show notes and chapters

In this episode of 'The Security Table,' hosts Chris Romeo, Matt Coles, and Izar Tarandach discuss the CISA Secure by Design Pledge, a recent initiative where various companies commit to improving software security practices. The hosts critique the pledge, arguing that many of the signatory companies have long been focused on software security, making the pledge redundant for them. They dissect specific goals of the pledge, such as increasing multi-factor authentication (MFA) and reducing default passwords, and express concerns about their actual impact.   
  
Despite their skepticism of the pledge’s effectiveness and measurability, they do acknowledge CISA's intention behind the pledge is to move the industry forward.  
  
Secure by Design pledge:  https://www.cisa.gov/securebydesign/pledge

Follow AI Security Table:

➜ Home: [https://securitytable.ai/](https://securitytable.ai/)

➜ X: [https://x.com/AISecurityTable](https://x.com/AISecurityTable)

➜ LinkedIn: [https://www.linkedin.com/company/ai-security-table/](https://www.linkedin.com/company/ai-security-table/)

➜ YouTube: [https://www.youtube.com/@AISecurityTable](https://www.youtube.com/@AISecurityTable)

## Transcript

**Chris Romeo (0:09)**: Hey folks, welcome to another episode of the Security Table. My name's Chris Romeo, joined by Matt Coles and Isar Terrandas, who are now doing the wave apparently. Thought only happened inside of large sports stadiums, but here we are. Apparently it happens in podcasts now too. It's a new thing that we're pushing the envelope on, just like everything else. So the topic of the day is the CISA Secure by Design Pledge. And so I have to set this up a little bit because I was out at RSA a couple of weeks ago and CISA had a little, little ceremony and I didn't get a chance to attend it, but I did hear about it where They had a number of big name companies come in and sign this Secure by Design pledge. And while that sounds oh so moving industry forward-y, I don't think it actually was. I think it was just a whole bunch of performance art and theater.

**Izar Tarandach (1:19)**: What?

**Chris Romeo (1:20)**: No. And didn't really move the needle forward because just to throw, to just to throw the grenade into the, into the room here, like the companies that signed this for the most part have been focusing on software security for 10 and 20 years. So my premise is they are going to do nothing different as a result of signing this pledge. There's going to be no, they're not going to go back and start a new subgroup or, you know, create a new team that's going to focus on this. They're just going to keep doing what they've been doing. It's, but that ends the intro because he's already, he's ready to jump in. All right. Go ahead, Izar.

**Izar Tarandach (2:06)**: I just want to say that after looking at the pledge, and we're going to go into detail on this, but take into consideration what you just said. And the items of the pledge. That's exactly what people doing up to now.

**Chris Romeo (2:20)**: Like, if this is new to anyone, uh, I saw a concept on the pledge. It was this MFA. It was MFA.

**Izar Tarandach (2:32)**: MFA.

**Chris Romeo (2:33)**: So what is that? Like, I wanna, I wanna try to extend my, oh, maybe it was the letters MFA. That's what it was.

**Izar Tarandach (2:40)**: Ooh, ooh, ooh.

**Chris Romeo (2:41)**: I need, I wanna have one. So I think I could, maybe I could, I could have some people extend our utilization of it as a gigantic company. Uh, which that ship is, it's already sailed. That, that, that ship is across the ocean. Like who doesn't have MFA capability right now in the things that they're building? Like if you don't have that, I got a question whether I would even use your solution. If you told me, if I said, hey, I just saw your new SaaS product and I want to sign up, but I don't see MFA. So, wait, wait, but— Okay, I'm out.

**Izar Tarandach (3:14)**: For people who are not familiar with the pledge itself, okay, let's just walk them through it. So these companies came and signed this thing in a ceremony. I am really bothered that there were no capes, goblets, invocations. You can't call it a real ceremony if you don't have one of those. No, no, no goals, no nothing. So yeah, terribly misnamed. But what they said was that we are going to strive for these aims in the next year or so. Now, drive you with a positive. One thing that's in the pledge that I really, really like is that they are asked to succeed in these tasks and then tell everybody what they did. But in case they don't succeed in those tasks, they are also asked to come and tell everybody what they did. And that, I think, is like sort of a first in a, in a official framework to come and say, hey, come and tell us how you failed because everybody can learn from that.

**Chris Romeo (4:13)**: There's no teeth in this thing though.

**Izar Tarandach (4:15)**: Oh, no, no, no, sure. It's everything completely voluntary. Everything is voluntary.

**Chris Romeo (4:19)**: And, you know, it, it, I, and I, I had a similar thought, but then I started to study what they actually wrote here. In the kind of subtext below the goal statement. And, and so let's start with the first one in the pledge here. Multi-factor authentication. I was kidding. I actually knew what it was.

**Matt Coles (4:36)**: MFA.

**Chris Romeo (4:37)**: With, here's what the goal says. Within 1 year of signing the pledge, demonstrate actions taken to measurably increase the use of multi-factor authentication across the manufacturer's products. That sounds really, really good. Like, that sounds like a positive move. What the heck does measurably increase mean? It doesn't mean anything. I can, I can measurably increase by 1. That's what I measured and I increased by 1. So I added MFA to 1 more thing. So my point is it doesn't, when you start, it, 1 is better than none, but when you start to unpack this, does this really move the industry forward? I don't see how this makes the world drastically a better place.

**Izar Tarandach (5:23)**: So wait, I, I would say that I agree with you, but I think that the, the fact that this exists, the fact that there is a first out there, the fact that there is something to start from by itself is a good thing. Yeah.

**Matt Coles (5:39)**: So let's, let's keep, so first off, I, I agree with you, Chris, in part that And I guess I agree, but there's our two, uh, that the, the goal here is to, is to reduce the use of passwords and, and move towards multi-factor authentication to re— to reduce the chances of phishing and credential stuffing and other attacks from being successful. If one, if one critical system or, or well-used product or com— or, you know, thing from a vendor gets that additional level of protection, That's a net positive. Does it move the needle a little bit?

**Izar Tarandach (6:17)**: It's just, just a little bit, right?

**Matt Coles (6:19)**: On the grand scheme of things. And it's a start. And if every manufacturer who is a signatory to this pledge moves one of their applications, just one of their applications for our products forward, then that's a, that's a good thing. Now, I, one might, an astute reader might take a look at the list of people who signed and some of those are not product vendors, their service providers, for instance, uh, and so what are they really going to implement from an MFA standpoint? Are they talking about the, the things that they deploy, or are they talking about the things that they don't build? So maybe the goal needs a little bit of work in that regard, but if you move the needle, if you move the needle a fraction of a degree, it's still a fraction of a degree in the right direction.

**Chris Romeo (7:09)**: Yeah, but it's, it's, people are gonna think that this is such a bigger thing. People are gonna interpret this that aren't cynics like me are going to—

**Matt Coles (7:20)**: if it was, if it was gonna be interpreted as a bigger thing, there would be more signatories to it.

**Izar Tarandach (7:26)**: Imagine if there were goats in the ceremony.

**Chris Romeo (7:30)**: That would've been something to see. But the point is, and someone else brought the, I'm, I'm gonna share another point that, that I was talking this was somebody else and this, this was, uh, the point that they brought into the conversation was these aren't the people that really need the pledge. The people that need the pledge are not on the list here. You got to get this pledge to them. They're the ones, there are people who don't use MFA as a base security feature for users inside of their platform. They're not on this list.

**Matt Coles (8:02)**: Well, they're not because no, but they don't know about it.

**Izar Tarandach (8:03)**: No, no, wait, wait, wait, wait. It's clear that the intention here is a monkey see monkey secure thing.

**Chris Romeo (8:11)**: Right. T-shirt idea. Monkey see, monkey secure.

**Matt Coles (8:17)**: Send, send your hate mail to— T-shirt.

**Izar Tarandach (8:21)**: No monkeys were harmed in the making of this podcast. Anyway, so fine. These are like big names that everybody recognizes. And I think that's the importance here is that SISA has had a lot of things coming in the last year. And this is sort of, I don't wanna say culmination, but this is sort of like a material step that they are taking into, hey, we are taking these people here and we are giving it some measure of accountability, right? So at, at the end of the day, I think that they're building a momentum towards some measure.

**Chris Romeo (8:53)**: I said, there's no, there's no accountability in this. There's no accountability.

**Izar Tarandach (8:58)**: There's self-accountability in here.

**Chris Romeo (9:00)**: Self-accountability is not accountability. Self-accountability. Like, there is no— they don't— the people that sign this never have to come back again. And there's no consequences.

**Izar Tarandach (9:10)**: They don't have to, but a year from now, if we can— if one of them got poked because of a default password, and we come back and say, bro, you signed the Caesar pledge! First, that would be extremely horrible to do, because everybody can get poked by all numbers of things, and sometimes not under their control. but for them it would be like, you know, it would be an egg in the face. So there, there is a small teeny measure of, sorry, man.

**Matt Coles (9:39)**: So, so let's, so let's just, sorry, let's, let's indicate that Izzard just called out the second of the pledge goals, which is around default passwords.

**Chris Romeo (9:49)**: Okay.

**Matt Coles (9:49)**: Within one year of signing the pledge, demonstrate measurable progress towards reducing default passwords across the manufacturer's products.

**Izar Tarandach (9:56)**: But you see, that goes back to what I said in the beginning. How come you're not doing this already? Why do you need a pledge to do this?

**Chris Romeo (10:03)**: What is measurable progress?

**Izar Tarandach (10:05)**: Yesterday I had two different passwords. Today I have one default password.

**Chris Romeo (10:09)**: I'm an old school requirements person. And most of the time I was on the other side of the equation trying to twist the requirement based on the words that were put into it to make it fit whatever I actually was doing already. But when I see something like this, demonstrate measurable progress, you have to define what that means.

**Matt Coles (10:29)**: So this item seeks to reduce the percent of exploitable default passwords in the wild in order to drive down attacks.

**Chris Romeo (10:36)**: Okay. So what is measurable progress? Is it 0.0001%? Is it 99%? What's measurable?

**Matt Coles (10:43)**: One is measure— one is better than zero.

**Chris Romeo (10:46)**: They're all measurable, but they're all just both Numbers are measurable progress.

**Izar Tarandach (10:50)**: They're just talking about the capability of measuring something. They're not specifying results or thresholds.

**Matt Coles (10:56)**: Right. And more importantly, actually, and more importantly, ideally it's a net, it's a net change to the positive.

**Chris Romeo (11:04)**: So here's, here's what I wrote about. Here's what I, here's what I want to see in this, in, in Pledge V2. Okay. So let's, let's use another, I liked it was reduced the next one, reducing entire class of vulnerabilities. I'll make my point here.

**Izar Tarandach (11:15)**: Yeah. This one. Oh my God.

**Chris Romeo (11:17)**: So here we go. So I'll read it just for those that are listening. On audio or at the gym or at the beach or wherever you are. And if you're in the gym, you've already fallen off the treadmill multiple times. Yeah.

**Izar Tarandach (11:27)**: Demonstrating.

**Chris Romeo (11:28)**: Here it is. Reducing entire classes of vulnerability. Within one year of signing the pledge, demonstrate actions taken towards enabling a significant, not just measurable, a significant measurable reduction in the prevalence of one or more vulnerability classes across the manufacturer's product. Here's my idea. You state the one that you're going to, that you're going to measurably drop. And you pay $10,000 to charity for every exploit, every vulnerability that hits a CVE that's, that's known, known if you can make a CVSS score above 6 or something, you pay 10 grand to charity for each one of the next year that comes out.

**Izar Tarandach (12:02)**: Now we got some teeth.

**Chris Romeo (12:03)**: Now we got some teeth on this thing. And now it's fun to watch. I'll make a dashboard and people can follow along and watch how the money's flowing. Cause that would be fun.

**Izar Tarandach (12:11)**: That would be awesome.

**Matt Coles (12:12)**: See, so it's good accountability. Reach out to Chris first. Yeah.

**Chris Romeo (12:16)**: I mean, if somebody build the dashboard, it'll be great.

**Matt Coles (12:18)**: So what you're saying is to, what you're saying is to gam— game, do gamification on this pledge as a means of—

**Chris Romeo (12:25)**: gamify the pledge. That's my new art piece.

**Izar Tarandach (12:27)**: Hold people accountable in a, in a material way, you know, in a way that we, somebody feels it.

**Chris Romeo (12:33)**: But hey, that's a great thing. No lawyer in any of these companies would sign off on that deal that I just described.

**Izar Tarandach (12:40)**: Let's for a second talk about the pledge of reducing entire classes of vulnerability.

**Chris Romeo (12:44)**: Okay.

**Izar Tarandach (12:46)**: Of all the 7 ones in here, I think that that for me is the one that most approaches what I understand as secure by design. Oh, absolutely. The others are security patches, vulnerability disclosure policy.

**Matt Coles (13:03)**: Yeah, before you jump ahead to the rest, let's— I mean, actually, no, now that you have, not all of these are secure by design. You're absolutely right. Enabling MFA, having MFA is a design issue. Default passwords is a design issue. Reducing classes of vulnerability, in part a design issue, although, although not exclusively. And then the rest, the rest are not really design problems, right? I mean, you're just— the design of a program. Well, the design of a program or the design of, uh, of your infrastructure or a design, you can relate it to design, but it is not strictly design of a system.

**Izar Tarandach (13:41)**: Yes. I would say that even the password and the MFA, I could relate to design, but are not design issues. Right.

**Matt Coles (13:48)**: Well, if you don't have MFA enabled, you have to at least design it into your product. Right. So it is, uh, it's potentially redesign.

**Chris Romeo (13:54)**: Yeah. We, we would not, we would not agree. If somebody put a product in front of us that only had username password authentication and said, is this secure by design? We would say no, because MFA is part of the corpus of just expected functionality that, that if you don't have it, you, you're not there.

**Matt Coles (14:18)**: And having hardcoded and having hardcoded credentials, the exact same thing, right?

**Izar Tarandach (14:22)**: That's not by design. That's just dumb. But anyway, uh, secure by 3 out of 7.

**Matt Coles (14:29)**: Right, because the rest are operational issues, right? So one of the, one of the ones towards the bottom there is, is issuing CVEs. When you do vulnerability reports, it's a great idea. If you're not issuing CVEs, you're really behind the times. You should have been doing that since, you know, 2008, at least.

**Chris Romeo (14:48)**: But But the CVE one, the CVE one has actually more measurability to it because it says demonstrate transparency in reporting by including accurate CWE and CPE fields in every CVE record for the manufacturer's products. This we can measure. You could literally use NVD if they ever refund it or get it moving again or whatever. You could, you could use that. To, you could actually measure that and say, okay, in 97% of your CVEs in 2025, you had accurate, you had CWE and CPE fields applied to it. Like that's, that's, that's actually more measurable.

**Izar Tarandach (15:30)**: But let me reel this in for a second. Just quick question. They have separate parts of the pledge, vulnerability disclosure policy and CVEs.

**Matt Coles (15:39)**: Yes.

**Izar Tarandach (15:40)**: Isn't emitting a CVE part of a responsible vulnerability disclosure policy?

**Matt Coles (15:45)**: It doesn't have to be. Doesn't have to be, but is it? Well, more, more importantly, more importantly, so the goal, you, you're misrepre— I think you're misrepresenting the vulnerability disclosure policy goal.

**Izar Tarandach (15:55)**: Okay.

**Matt Coles (15:55)**: And we're, and we're obviously, we're skipping around here for a bit for folks who are trying to read, read it, read along at home. So the goal of vulnerability disclosure policy is within 1 year of signing the pledge, publish a vulnerability disclosure policy. So, First off, uh, let's talk about measurability here. The measurability is, did you publish something that you didn't have previously? But, but really, shouldn't this be talking about implementing a vulnerability disclosure practice that functions, that works, that results in CVEs being generated so that you can meet the— in order to meet the next part of the goal, you have to be able to issue CVEs, which means you have to work with a CNA. Right? A naming authority for CVEs. You have to have information in your vulnerability disclosure practice. It's missing from the actual goal.

**Izar Tarandach (16:48)**: True. Yeah, I see the distinction now. And actually, when you read, another thing popped up in my head. Uh, within one year of signing the pledge, publish a vulnerability disclosure policy that authorizes testing by members of the public on products offered by the manufacturer. Isn't there a lot of license?

**Chris Romeo (17:07)**: Yeah.

**Izar Tarandach (17:08)**: Isn't there a lot of licenses out there that say in no way, shape, or form can you try to do something malicious to our product, otherwise we're going to get your firstborn and kill the gold and do all kinds of things?

**Chris Romeo (17:19)**: A decade ago, that was a problem.

**Matt Coles (17:21)**: I didn't read that. Holy moly. So they are talking about safe. So if you read further down on that, it's talking about safe harbor language. So this is the, this is the notion that we wanna make, that, that companies should not make security research on their systems a punishable offense.

**Izar Tarandach (17:41)**: Yeah.

**Chris Romeo (17:42)**: Whether that's legal license or—

**Matt Coles (17:45)**: Security research is not, is not universally accepted as an acceptable form of, of testing. Um, you, we have, it still runs afoul of many rules.

**Izar Tarandach (17:56)**: But, and even pentesters need a get out of jail free card.

**Matt Coles (18:01)**: But a vulnerability response, but a vulnerability response policy is not the place to put that.

**Chris Romeo (18:05)**: Right?

**Matt Coles (18:06)**: I mean, maybe if you're going to do a bug bounty, you might have that in a bug bounty scope. But I don't, I don't know. And maybe this is something if we have vulnerability or incident response teams out there who are listening and can provide some feedback on this, is this something you would actually put in a vulnerability disclosure policy? That doesn't sound right. Uh, you may put, you may put protections for, for safe reporting practices, meaning you'll keep, you know, the researcher confidential and you'll do coordinated disclosure. But I don't know that, that you would put in your vulnerability disclosure policy that it's okay to do security testing. That doesn't sound right. Yeah.

**Chris Romeo (18:49)**: Hmm.

**Izar Tarandach (18:49)**: There's something fishy here.

**Chris Romeo (18:51)**: Do you think everybody knows? Do you think everyone who signed this knows? They, they must have caught that nuance.

**Izar Tarandach (19:00)**: Wait, wait, wait, wait, wait. You mean that everyone read it before they signed?

**Chris Romeo (19:04)**: Well, their lawyers read it.

**Matt Coles (19:06)**: I would imagine their lawyers, yeah.

**Chris Romeo (19:09)**: So if their lawyers read it—

**Izar Tarandach (19:09)**: Why would the lawyers if it's non-binding? I can't find anything that's non-binding.

**Matt Coles (19:14)**: No, it's a PR question. It's a PR question. If you, if you didn't— if you're, if you're saying on behalf of, uh, so-and-so, on behalf of Microsoft or Google or HP or, you know, the other signatories here, hey, we agree to this pledge and you didn't run it by your corporate attorneys.

**Izar Tarandach (19:30)**: Oh, so then we agree that there is a measure of accountability here.

**Matt Coles (19:34)**: Well, there's public accountability.

**Chris Romeo (19:37)**: No, just in this, like to your, to your earlier point, some, I think Matt made the point about the EULAs that this may not be allowed inside of the EULA. And so I'm looking at this and going, could there be like something that people signed up for that's actually against their own EULA? I mean, any of those, like, that would be, that would be hilarious.

**Izar Tarandach (19:57)**: Hilarious.

**Chris Romeo (19:58)**: We're retracting our sign that we actually didn't sign. We put a fake name. We wrote Mickey Mouse on the pledge when we were at the ceremony, just in case we, uh, or are they going to give you a million dollars because they found the one guy that read the EULA?

**Matt Coles (20:12)**: Tiny letters at the bottom. So now it's interesting if you, so let's, let's touch upon this BDP one a little bit more. Cause if you read down further, there's a note. Note, due to the specific nature of this item, examples of achieving this goal are not included.

**Izar Tarandach (20:29)**: What?

**Matt Coles (20:32)**: But there is a measure, though, there is an example of demonstrating progress. Actually, Chris, so, so to your point. There is in each section, there is an example of demonstrating measurable progress. So they do indicate what might be considered as measurable progress.

**Chris Romeo (20:48)**: I read that as non-binding.

**Matt Coles (20:50)**: It is non-binding. All of this is non-binding. That's part of the problem, right?

**Chris Romeo (20:55)**: Is it's all non-binding. It's, it's the non-binding pledge.

**Izar Tarandach (20:59)**: One more reason for the goats.

**Matt Coles (21:01)**: Well, it's a, it's a, it's, it's a non-binding—

**Chris Romeo (21:04)**: you just can't let the goats go.

**Matt Coles (21:06)**: All right. So also we probably should let people know about the history, right? So this, this came on the heels of something else.

**Chris Romeo (21:12)**: History of goats? I don't know the history of goats.

**Matt Coles (21:14)**: This came on the heels of something that CISA did with, with, um, K through 12 education organizations around securing their environments.

**Izar Tarandach (21:23)**: And I guess they're saying the pledge in schools now.

**Chris Romeo (21:28)**: I pledge allegiance to, to CISA.

**Matt Coles (21:37)**: Um, yeah, so, so there's, it's modeled after, after something that, that came before with, you know, a pledge to K, K through 12 education organizations to, to improve the security of their environments. Um, and it had some measure of su— of success. And so they took that approach and are using it here. And so it's all voluntary. But there's ways to describe what, what progress means in each of the sections.

**Izar Tarandach (22:08)**: And just to—

**Matt Coles (22:08)**: so let's just flush it out because we skipped around a bit. Security patching within 1 year, demonstrate, demonstrate actions to measurably increase the installation of security patches by customers. Now, I have some thoughts on that one, but let me, let let me flesh out the rest of them just so we get to them. And the last one that we haven't talked about is evidence of intrusions within 1 year of signing the pledge demonstrate a measurable increase in the ability for customers to gather evidence of cybersecurity intrusions affecting the manufacturer's products. We have a lot to unpack in both of those.

**Izar Tarandach (22:50)**: Yeah.

**Matt Coles (22:50)**: So security patches. Yeah. Driving, driving customer behavior, driving customer behavior, but I'm going to save it.

**Chris Romeo (22:58)**: So let's do patches first.

**Matt Coles (23:00)**: Oh yeah. So patches, driving customers to install patches. So first off, it's outside, it's outside the control of the manufacturer.

**Izar Tarandach (23:09)**: Wait, wait, where is that? Oh, yeah.

**Matt Coles (23:13)**: So, so increasing the installation of security patches by customers, you can do that in one of two ways. Really one of two ways. You educate customers like, like Google has done over the years, I think, with, with Chrome. You make it, you make it stupid easy to, uh, to install the patches and you make it very clear why you need to install the patches.

**Chris Romeo (23:34)**: I mean, I'm in favor of this.

**Matt Coles (23:36)**: Or you take it out of their hands and you do it automatically.

**Chris Romeo (23:38)**: Yes, just why are we even making this a user decision?

**Izar Tarandach (23:42)**: Wait, wait, wait, wait.

**Chris Romeo (23:43)**: Apply a security patch.

**Izar Tarandach (23:45)**: Why don't we make the system so that they stop working if the patches are not in there?

**Chris Romeo (23:50)**: Okay, you're going too far now.

**Izar Tarandach (23:51)**: The goats.

**Chris Romeo (23:52)**: And now you want the system to just shut down when it should have been patched. But let's, let's, let's seriously walk this back for a second because I think Matt just pinpointed something that might, might have, might make this pledge actually have teeth. So why doesn't the pledge say within 1 year of signing the pledge, Demonstrate actions to, to automatically install security patches for customers.

**Izar Tarandach (24:20)**: Ooh, the automatically is the problem.

**Matt Coles (24:22)**: Yeah, the automatically is the problem. Let's use an example of an OT system.

**Chris Romeo (24:26)**: You know, so you can't sign the pledge if you're an OT.

**Matt Coles (24:29)**: What?

**Chris Romeo (24:30)**: Not my problem. I mean, that's the pledge. Like the pledge is if we're going to be, if we're going to get serious and we're going to start pushing the envelope here. We can, if we water things down, it doesn't do anything except a bunch of nice photo ops at RSA and handshakes and people smiling and using a ceremonial pen and goats. And there's nothing that goes with it. Like it's, it's like, but if, but if, what I just defined, RSA. No, no, this was one particular segment. Come on. But why not? Why not? Why isn't that a pledge, a goal you guys could get behind? Within, I'll say it again, within, I've just made it up on the fly, but within 1 year of signing the pledge, demonstrate actions taken to automatically install security patches for customers.

**Izar Tarandach (25:16)**: Apart from the fact that some systems you cannot have it automatically, and in some business cases you can't have it automatically because, hey, I just need that version because the other thing that I use needs that version.

**Chris Romeo (25:26)**: How long has Windows been auto-applying patches?

**Matt Coles (25:30)**: It has the option to not. I don't know.

**Izar Tarandach (25:32)**: I don't use it.

**Chris Romeo (25:33)**: It turns on by default. Now it just auto applies patches, right?

**Matt Coles (25:37)**: And they do have disabled auto apply, but it's, it's different.

**Izar Tarandach (25:41)**: It's different when you're talking about the desktop operating system. It's, it's you with the, with the machine. Perhaps you have a bunch of things running locally that might give you a headache. But when you're talking about business systems and this thing, version 3.6 has a dependency on that thing versus version 4.7, whatever, you can't just—

**Chris Romeo (26:00)**: So you gotta get, so you gotta get better at, you gotta get better at how you build software then.

**Izar Tarandach (26:06)**: And how you manage dependencies and your patch cycles and these and that and the other.

**Matt Coles (26:14)**: It's also much more closely related to the design of your system, by the way. Yeah. It is a design problem, but, but there are operational requirements that, that you, that may restrict the ability to push patches automatically.

**Chris Romeo (26:27)**: Okay. Then here's my point. I'm making a new point.

**Izar Tarandach (26:29)**: There's some places that will not push a patch until after they spent a month, uh, testing that.

**Chris Romeo (26:37)**: Right? So, at some point, we have to trust the people who are providing us with the software that their patch is not going to cause the whole, the whole enterprise to shut down. Yes, I grew up in the days when you applied a Microsoft patch too early and everybody was locked out of their desktop or out of their laptop. I watched it happen multiple times. Okay. But I feel like we're past that now. We have to be past that now.

**Matt Coles (27:05)**: Okay. But we also have, we also have systems, you know, say in data centers where if you apply a patch to a, to a database at even at 2 AM on a trade, on a trading application, You know, you're gonna, you're gonna stop after-hours trades for billions of dollars in, in, in process.

**Izar Tarandach (27:21)**: Let me put it like this. There is a very, very strong reason why we would never turn on the autopilot on a self-driving car. And I correlate that directly to the fact that no, I will not be having automatic hands-off matching any type of system.

**Chris Romeo (27:37)**: You know what you just told me?

**Izar Tarandach (27:38)**: Because no, I won't be trusting the people who wrote that thing.

**Chris Romeo (27:41)**: You know what you just told me? You haven't driven in a Tesla with Autopilot in a while. And I use it all the time. I use it all the time.

**Izar Tarandach (27:49)**: Yeah, yeah, yeah. And I watch.

**Chris Romeo (27:50)**: I watch. It's freaking cool, man. It's, it will blow.

**Izar Tarandach (27:55)**: Until you sit back and you start thinking, wow, I know a lot of people who do QA. Well, I know a lot of people who write software really, really, really well. I know a lot of people who don't. And no. No, that, that you're just asking for a level of trust that I am not equipped to give.

**Chris Romeo (28:11)**: Yeah. Maybe this is generational. Maybe you guys are just too old to see this new way of thinking for me.

**Izar Tarandach (28:16)**: Or as the new t-shirt says, I'm older than both of you guys.

**Chris Romeo (28:20)**: That's why I'm making ABTM. What's that?

**Izar Tarandach (28:22)**: Always be threat modeling.

**Chris Romeo (28:24)**: Always be threat.

**Matt Coles (28:24)**: Yeah.

**Chris Romeo (28:25)**: No, I mean, I can t-shirt idea ABTM.

**Izar Tarandach (28:28)**: The fact is, no, I, you asking for a level of trust here in things that we day to day see the dark side of. Well, for me, I've seen enough enterprises that don't do that stuff.

**Chris Romeo (28:38)**: But what's the alternative? The alternative is we just keep doing the stupid things we've been doing for the last 30 years and things never get any better and they never get any different. If you can't build me a system that can patch on the fly, hold on now. Part of the problem here is we can't patch without rebooting often. You should be able to apply a software patch while a system is operating without having to do a reboot. Computer science is not that far.

**Izar Tarandach (29:03)**: The Linux kernel got a lot of capability to start certainly.

**Chris Romeo (29:06)**: But when I think about like macOS, Windows, kind of the enterprisey desktop operating systems, neither one of them, I think they, I think macOS maybe can do it in certain situations, but often they need a reboot for the whole thing. Yeah. Like that seems like a failure in software design over the last number of decades. Like we should be able to patch something on the fly. So we don't have to reset the entire system to apply a patch. We should also have the ability to roll that patch back automatically if the system detects some fault state. So now we're building a better operating system to be able to automatically apply security patches. The problem is we haven't, we haven't moved anything forward behind the scenes.

**Izar Tarandach (29:53)**: So I think that in a lot of different domains, we have already gotten quite a lot of what you're describing. The things that doing it like that automatically will turn things more expensive, both in terms of development, testing, certification, and actually use, right? Sometimes you, I'm talking out of my behind here, but sometimes you have pieces of hardware that you have to physically go and initialize for them to get the result of a patch.

**Chris Romeo (30:26)**: Hold on. Like, are you talking about something that's been made in the last 5 years? Or are you talking about things from 40 years ago?

**Izar Tarandach (30:31)**: Matt, am I?

**Matt Coles (30:32)**: Talk about relatively recent things. Yes. This is how hardware, a lot of hardware devices work.

**Izar Tarandach (30:37)**: Matt knows.

**Chris Romeo (30:38)**: Are we talking about ICS, OT style systems?

**Matt Coles (30:43)**: In part, but not universally those. There are desktop, there are consumer grade devices that are probably in that boat as well.

**Chris Romeo (30:51)**: So they, that's, I mean, I've talked about this before. I'm just baffled that the market hasn't crushed certain failures in products to move forward. Classic products.

**Matt Coles (31:05)**: Cheap stuff.

**Izar Tarandach (31:06)**: For the love of God, we are talking about taking away default passwords and you want people to dynamically initialize and patch kernels? Seriously? Are we in the same ballpark here?

**Matt Coles (31:17)**: How much RAM do you think these things, how much RAM do you think these things have?

**Chris Romeo (31:20)**: I mean, add more RAM to it.

**Izar Tarandach (31:21)**: I knew, but I forgot.

**Chris Romeo (31:23)**: Yeah. Cause you don't have enough RAM. Burn it in the ROM would be the answer.

**Izar Tarandach (31:29)**: You add disk space, then you have virtual memory, then you don't need more RAM.

**Chris Romeo (31:32)**: Let's deal with the last one. Let's deal with this. We haven't dealt with evidence of intrusion.

**Matt Coles (31:37)**: Oh, the last one is, yeah, the last one is, well, boy, let's, let's open that can of worms, shall we?

**Chris Romeo (31:42)**: Figured it was time. Pandora's box here.

**Matt Coles (31:45)**: So this one, so I have a really simple, I have a really simple idea on this one.

**Izar Tarandach (31:49)**: Wait, wait, wait, read it first. Read it first. Yeah.

**Matt Coles (31:51)**: So, so, and it's a mouth, it's a mouthful. Demonstrate a measurable increase in the ability for customers to gather evidence of cyber intrusions affecting the manufacturer's products. So let's break this down. A manufacturer product has to be able to know that it's being intruded, intrusion, that intrusion is taking place or generating indicators of compromise that a customer can gather to know if it's, if an intrusion has occurred. So, and the context, so they added context to this one, specifically for this one, with some detail here. Now, mind you, they didn't, they didn't go into any more detail on the, on the goal. Jump right into the context. It's essential that organizations have the ability to detect cybersecurity incidents. And so software manufacturers need to enable their customers by providing artifacts and capabilities to gather evidence, such as having audit logs. And so at the, at the very, at the, at the minimum, I'm so confused. Having log generation, having log generation is a core foundation of design that we've, we've talked about for decades. If you're not already logging, WTF.

**Izar Tarandach (33:14)**: But wait, it is essential that organizations have the ability to detect cyber— who, who's organizations here? Is it the customer or is it the provider?

**Matt Coles (33:24)**: It's for customers to gather evidence of intrusions.

**Izar Tarandach (33:29)**: So, okay. So customer Let's take the example of a cloud provider. I'm a customer. What is being asked of the cloud provider to give me here?

**Matt Coles (33:40)**: That your environment that you're operating in that cloud has had intrusions.

**Izar Tarandach (33:45)**: So I need the logs for the VPC saying that nothing bad has happened and entered in there.

**Matt Coles (33:50)**: Yeah. Or they give us this— or whatever you're using.

**Chris Romeo (33:52)**: Yeah.

**Izar Tarandach (33:52)**: So people are already— SOCs out of the land are already overloaded with checking events for things that they actually wrote and are responsible for. Now we have to check the other side of the shared responsibility model as well. Yeah.

**Chris Romeo (34:08)**: Well, it's about transparency. It's about giving you—

**Matt Coles (34:10)**: Transparency. It's logging. It's logging and hunting.

**Chris Romeo (34:12)**: It's giving you the ability, it's giving you the transparency to be able to detect an intrusion against a product without the manufacturer's help.

**Izar Tarandach (34:21)**: But look, again, it's like the bomb. Yeah, let's create this thing. Who's going to consume it?

**Chris Romeo (34:27)**: Yeah, I mean, that's been my role for the last 2 years.

**Matt Coles (34:33)**: But I mean, if, if, so if a product isn't, again, if a product isn't generating logs, it needs to be generating logs. Just, it needs to happen. I'm more worried about, does the product need to have automatic detection of intrusions? Or do you, is it the customer who needs to make that determination?

**Izar Tarandach (34:49)**: I don't know, but from what I read, everything that can today do detections also has some automated capability of figuring out what they are. There are LLMs galore doing that stuff for us. I have GPT-whatever, right? Yeah.

**Chris Romeo (35:03)**: Goats.

**Matt Coles (35:04)**: Did an LLM write this pledge?

**Izar Tarandach (35:07)**: Oh. No, but interestingly enough, we have a pledge that the LLM did write. No, we have the parallel pledge that the LLM did write.

**Chris Romeo (35:16)**: All right. So Matt, what do we do with this? This last one, like, what are we— what is your take? Do you, do you see value in this? Because Izzar's saying— Izzar's kind of taken the negative side, playing my role from this whole conversation. I somehow was defending this. I just realized. Matt, what's your take?

**Matt Coles (35:38)**: So, so, uh, uh, yeah, I, I think if it was, if it was all it was cracked up to be, there would be more signatories. But I 100% agree with you. The people who, the companies who have already signed this are the leaders in these areas already. So if we're looking at the pledge signers to improve, I, I, I find it hard to know. I, I have, I find it hard to understand how they improve much more from where they are. But the people who don't or are not signature signatories to this on the, the large swath of product, product manufacturers and suppliers of, of systems and devices that didn't sign it are really the target audience for this. And there's a little room, lots of room for improvement.

**Chris Romeo (36:25)**: I would say my concluding remarks here, I should have said this from the top. CISA's trying to do what they can to move our industry forward. Okay. Probably should have been nicer about it from the beginning. So people that hung up on this conversation, we like beating on CISA.

**Matt Coles (36:40)**: It's a, it's a love-hate relationship. It is.

**Izar Tarandach (36:41)**: It is.

**Matt Coles (36:41)**: But I mean, they are trying to move the industry forward.

**Chris Romeo (36:44)**: Forward. I get it. I get they're trying to do something. And so is something better than nothing? Yes. Something is almost always better than nothing. I just, my struggle has been, it seems like they could have gone so much further with this pledge. They could have made it simpler, but yet more hard-hitting and have, have stronger teeth. Maybe they'll do that in the future. Maybe that's their goal.

**Matt Coles (37:06)**: Uh, or more entertaining at the very least.

**Chris Romeo (37:08)**: Anybody at CISA, if you want a gamified pledge dashboard, let's chat. I think that could be kind of fun, but at the end of the day, CISA, keep doing what you're doing. Like, keep trying to move the industry forward. Like, that's not— my goal is not to try to squash the movement of our industry. I think you just need some— you need some people that aren't just high-fiving you all the time to give you feedback on stuff you write.

**Izar Tarandach (37:31)**: Like, I personally, I just think that apart from the reducing entire classes of vulnerability here, the whole thing went to the books. I mean, Yeah, that stuff that, you know, if you aren't doing up to now, seriously, I don't know what to tell you. But on a secure-by-design approach, that the reducing entire classes of vulnerability, that's the one that to me means something. Still, is that measurable? Is that something that you can report?

**Matt Coles (37:58)**: I don't know.

**Izar Tarandach (38:00)**: And maybe CISA—

**Matt Coles (38:01)**: probably not. Maybe CISA try to be a little bit close, you know, name this thing properly. If you're gonna do a secure by design pledge, have a pledge that actually has secure by design items.

**Izar Tarandach (38:10)**: Yes. I, I, I think that they were not trying to have this pledge be like the greatest, greatest of all time, or as people say, the gold. But, uh, uh, uh, I think that they went for something that they can definitely reach out to and get people moving in a direction, as Chris said.

**Chris Romeo (38:26)**: Right.

**Izar Tarandach (38:27)**: And hey, positive change is, is positive. So yeah, we stand behind it.

**Chris Romeo (38:34)**: Yeah. Even though we just spent 38 minutes pinpointing a lot of flaws with, oh, we're in, we just came back.

**Izar Tarandach (38:40)**: It's not a big mountain to climb, right? You don't have to be, I don't know, a goat to do it, but it's there.

**Chris Romeo (38:48)**: Oh man.

**Matt Coles (38:49)**: There's a t-shirt in there somewhere. There is a, yeah.

**Chris Romeo (38:51)**: Goat. Picture a goat on the front with no other context. It's just gonna say goat.

**Matt Coles (38:59)**: Goat.

**Chris Romeo (38:59)**: Oh, I don't know where else we could go at this point. I think we've reached the end of this.

**Matt Coles (39:06)**: Over the cliff. We're, we're guard. We lost the guardrails and we're off the cliff.

**Izar Tarandach (39:10)**: We can always go up to the barn.

**Chris Romeo (39:11)**: Let's go to the barn. Let's see if the cow is also nearby to the goat. All right. With that, I will end this conversation and, uh, please listen again soon. To another episode around the security table.

**Izar Tarandach (39:29)**: But CISA, we love you.
