# Software bill of materials -- what is it good for?

URL: https://securitytable.ai/software-bill-of-materials-what-is-it-good-for/
Published: 2023-01-02
Duration: 3165 seconds
Season: Not supplied
Episode: Not supplied
Hosts: Chris Romeo, Izar Tarandach, Matt Coles

## Show notes and chapters

The gang considers the software bill of materials (SBOM) approach and asks hard questions about what SBOM is for and whether it improves security. Note the gang believes in SBOM. We ask the hard questions to help us all expand our minds and truly understand the value propositions.

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 (0:00)**: Welcome to another episode of The Security Table. My name is Chris Romeo. I'm joined by Tanya Ward, Matt Coles, and Izar Tarandash. And just like all of our other episodes, we're here to discuss something in the realm of application software security and have a little bit of debate, perhaps maybe a little bit of disagreement, hopefully not a lot of agreement, but you never know. The topic for today is SBOM, Software Bill of Materials, which has gotten a lot of attention in our industry here in the last, I'd say, maybe 18 to 24 months. And so I thought I would just start us off with a definition. So CISA has a definition of SBOM, and they say a software bill of materials, SBOM, has emerged as a key building block in software security and software supply chain risk management. A SBOM is a nested inventory, a list of ingredients that make up software components. And so I guess my first question for the group here, what do you think of that definition? Is that the essence of SBOM according to what you think of it, or is there more, is there less? Somebody gimme your thoughts on SBOM.

**Matt Coles (1:17)**: Technically correct.

**Tanya Ward (1:19)**: I'll say.

**Chris Romeo (1:20)**: Okay. Okay. Why is it technically correct?

**Matt Coles (1:24)**: Well, so I think it's technically correct because I mean, that's what, that's what an SBOM is, right? An SBOM, its definition is, is an expression of a collection of things. And specifically the collection of the things that go into something in its makeup and its composition. So that's technically correct, but I think it misses the mark of what SBOM is supposed to mean. Right? SBOM is supposed to be a level of transparency. Right? So it isn't just— isn't technically just an inventory, at least in my mind. I mean, yes, it's a data format that expresses a set of things, and that those set of things are supposed to be the things that comprise or composition of the whatever the SBOM is describing, but the SBOM really expresses— it provides a level of transparency that can be shared to others. I'll leave it at that.

**Tanya Ward (2:25)**: But isn't— I mean, this isn't a first in the industry, right? In the security industry, we've always had to have a bill of materials for potential open source so that we understand the licenses. So This has been around for quite some time, and if you even go back in time, think about the FDA, when you have food ingredients, like when you have a packet of food, you have to put all your ingredients in. So I mean, it's not really a new concept.

**Matt Coles (2:55)**: Well, actually, and just, you touched upon something actually I was thinking about before we started this session, was we have— so we started with transparent— I'll say transparency when we say, well, we use open source, right? Software developers would say we use open source, or we'd assume it was there, right? And so you'd have an understanding of, well, we don't build our own components, right? That was first. And then we went to, okay, now because of licensing, they built the licensing rules that we have to then tell people, well, we're using these licenses. SBOM in that sense is an elevation or escalation of that transparency because now it's no longer, well, you're using GPL or using AGPL or a FARO or whatever licenses. Now it's, well, what components are you using? And that's not just what components, but what specifically what versions are you using and potentially what versions are they using? And so it's taking the veil off of software composition. Even further than it originally has been. Bizarre.

**Tanya Ward (4:03)**: Bizarre. And you were—

**Izar Tarandach (4:07)**: Grenade in the room. Ah, transparency is good, but by itself, if it's close to not worth anything— I mean, let's go back to the food thing. So you're looking at your latest, uh, snack that you're buying and you learn that it has orange 3, red 4, extra 4, bit of I don't know what. And now you have all the transparency in the world and you have no clue what the hell you did with that.

**Chris Romeo (4:38)**: And you're still gonna eat it. You still put it in your mouth and eat it after.

**Tanya Ward (4:42)**: And you don't know what it means either. You don't know if it's gonna— anyway.

**Izar Tarandach (4:47)**: My point is that while it's funny that we started talking about SBOM as if it was one thing, I think that myself, and I don't consider myself very well informed on the subject, but I think that there's about, what, 4 standards running around for SBOM?

**Tanya Ward (5:06)**: 3?

**Matt Coles (5:06)**: 3 standards.

**Tanya Ward (5:08)**: Yeah.

**Izar Tarandach (5:08)**: SPDX, CycloneX, and believe you me, I won't remember the other 2.

**Matt Coles (5:13)**: CycloneDX.

**Izar Tarandach (5:13)**: Not that they are less important, it's just that I am ignorant of it at this point. But my point is that by itself, I don't think that SBOM is giving us much. The processes that are being built around SBOM, those I think are giving a lot. And I think that Matt and I have, uh, have some experience on that. Once upon a time we were called to build this, uh, code signing, uh, system. And we ended up figuring out that things without a manifest were not worth much, and manifests not being signed were not worth much. And, uh, in, in that very small view that we had at the time of looking at the, this small problem of provenance and making sure that things are what they say they are, I think that we learned a lot. And again, I, I think that the, the fact that you have the transparency by itself doesn't give you a lot. The processes and the, uh, levels of assurance that you may get from that transparency, if well closed, if well looped, are probably what's actually worth there.

**Chris Romeo (6:23)**: You're always chasing, you're always chasing with SBOM, right? And so, let me give you a new working definition and then, Tania, I'm sorry, I didn't mean to cut you off, but then give me a reaction to this. Here's a new working definition, which might get me in trouble. Some people might come after me on the internet or something.

**Tanya Ward (6:39)**: I don't care.

**Chris Romeo (6:39)**: An SBOM It won't be the first time. So, I mean, it's, it's, they're coming for me. So now here's a new working definition of SBOM. An SBOM is an out-of-date inventory of the software that's included in something that I bought.

**Tanya Ward (6:57)**: Because it's always being revved, right? And changed and updated. Yeah.

**Chris Romeo (7:02)**: And my point is, can a, can a human keep up with the SBOM? Velocity, right? So, like if you're using a modern SaaS product, let's say one of the, you know, we'll say one of the big ones like Etsy or one of these people that was always the model of DevOps because they pushed 50 times to 200 times a day. Amazon apparently does the same thing. So, how valuable are SBOMs in that world if you're pushing software 50 times a day?

**Izar Tarandach (7:31)**: Or 30 times? I mean, but at some point you ask, can a human keep up with that? I wouldn't expect a human to keep up with anything.

**Tanya Ward (7:38)**: But I think your point, Chris, is that if you're constantly refreshing the SBOM and always updating it, then why do you really need to track it? Is that your point?

**Chris Romeo (7:47)**: Yeah, that's my, I guess, my hypothesis. Like, why do we have to track it if, or does it really add that much value if it's always changing?

**Tanya Ward (7:58)**: I will kind of put a spin on that. What's the point of SBOM if you don't have good vulnerability practices or disclosure practices, right? Because that goes back down to the benefit of SBOMs, right? So SBOMs, we have a component, we have a set of vulnerabilities. Now vendors have these components, they should be following, you know, per, you know, the ISO standards 30111 and 29147. These best practices around vulnerability disclosure, if they are not doing that, I would question what's the value because you don't really know if they are vulnerable or not.

**Matt Coles (8:33)**: I want to say something here because this is— so we've had— SBOM is not new. I mean, yes, SBOM is new. The concept of— and I've said this a number of times in the past, you know, this is the concept of having a component inventory right, is not new. You can't manage what you don't know.

**Izar Tarandach (8:55)**: Yes.

**Matt Coles (8:55)**: Right, and so I— did we skip a step, right? What's the value of SBOM and relating to vulnerability management? Is the expectation that just because somebody creates and publishes an SBOM to their consumers that they're going to use that SBOM to do vulnerability management practices? If that's the case, Why weren't they using SBOM internally?

**Izar Tarandach (9:19)**: Yes.

**Matt Coles (9:19)**: To do vulnerability management. Why do they have to make it public? And so, right, so we're skipping a step, right? We're jumping really far ahead. We're saying, we're basically saying, vendor, we don't trust you. Show us.

**Chris Romeo (9:34)**: Who's the SBOM for then? Who's the consumer of the SBOM? Is it for a company or is it for somebody who buys your stuff?

**Matt Coles (9:42)**: Well, I'm going to— I'd like to say it's two potential targets there, right? For their SaaS provider, maybe that's a process that's determining, you know, assurance or capability of the connected systems. For IoT devices, it's potentially just the direct consumer, right? And I think that's the counter here, is an SBOM for an IoT device that never changes after it gets, you shipped to a customer and it never gets software updates, it has value. You have time to process information and make a decision about whether you want to put that on your home network or use it in your car, right, and it doesn't change. When am I going to buy a new one? Well, when my risk appetite, you know, obviously as a consumer I'm not going to use those terms, but I'm going to say, well, that smart toaster that I bought 10 years ago has these now out-of-date components. Granted, it's in a format that I can't understand as a home user potentially unless you've got a tech background. So, maybe that needs to change. You know, nutrition labels has been considered, right? We've already talked about it here. So, maybe that's what SBOM needs to be paired with in order to make it functional in that environment. But in the SAS case, you know, that could be automated processes, right? When you have connection protocols and you have onboarding protocols and you have sort of regular ongoing maintenance. That may be process-to-process communication, not process-to-human communication. Isar, I know you've been trying to jump in.

**Izar Tarandach (11:14)**: Okay, so two things. First one is your description going to IoT brought me to mind like two guys in an aisle in a big store looking at the latest home assistant things and going, well, this one, this one has LIPXML 2.0, but this one has LIPXML Wait, don't you make your technology generation that way? Doesn't help a lot. But on the point of who gets anything out of SBOM, first of all, I think that Matt is totally right in the the way that, that he put it with the, the two orientations. But again, I'm thinking in terms of processes and two ways that you can involve SBOMs in the process of releasing and selling software. And somebody's going to take advantage of it is, first of all, in your internal process, that if you figure out that what you declared at the time that you built the SBOM and what's actually out there in front of customers. It's not exactly the same thing. It's a great indicator that something got broken somewhere. On the other hand, we live in this time, and if you don't— we— if you, the, like, the generic you don't live in the time, then you should— where we trust but we verify. So for those people who perhaps don't know how to interpret the results or the contents of an S-BOM but have the processes in place to run their own scan on that thing and say, hey, here are some red flags that perhaps you should look for, and that perhaps your own use policy— like something that happens to all of us a lot of times, we, we are called to, uh, look at a piece of software, at a piece of something, and say, is this good enough to bring inside our own companies and environments, right?

**Chris Romeo (13:22)**: Is that what people are doing with it though? Is that what they're doing with it? Are they, are you getting, if you're a company and you buy software from somebody and you're demanding an SBOM, are you feeding that into a pipeline somewhere that's running a scanning tool on it? And then you're saying, ha, vendor, fix that issue or I'm no longer doing business with you.

**Tanya Ward (13:39)**: But, but that only works again, if it goes back to the disclosure practices, 'cause scanners are designed to pick up vulnerabilities.

**Izar Tarandach (13:46)**: But wait, I don't think that people are doing that. But the important thing is that now they can do that if they so choose. So someone who says that process of bringing software into my company is a bottleneck and it takes a lot of people time to actually do that and look at it and tell you the truth, we are not really sure. Now we have a spark of something that we can use to start automating some of that thing. Is it going to give us 100% certainty? Not.

**Tanya Ward (14:13)**: So can I— so I think we're— I guess what I would like to understand is what problems is it really trying to solve? In my head, when I think about SBOM, this sounds crazy, but I just think of the Russian doll analogy. We have a little doll inside another doll inside another doll inside another doll inside another doll. Like, how many layers is there? And so if the goal is to be able to understand the risk that these components bring to your system that you're selling, will you ever, will you ever obtain that nirvana? Because you're relying on individual, each of these little NAFTA dolls disclosing, saying, hey, I'm in this component, I'm in this component. So I actually think it becomes a very complex problem that we're trying to solve. And in transparency, are we really truly, like, transparency is not, you know, transparency is such a loaded word. Like, are we 100% transparent? Are we as transparent as we can be based on what we know. I think there's— yeah.

**Matt Coles (15:11)**: So, Tonya, I think you've hit upon a really important point. SBOM expresses a single relation, right? Contains. My thing contains X. What it doesn't express is my thing contains X and it has impacted in some way, right? Just because I'm containing a component oh, I contain libssl, right, and libssl whatever version 1.1.1 something has a vulnerability. Let's just say it's a medium vulnerability just for the sake of argument. Espon tells me this thing that I bought or this thing that I'm using contains libssl and libssl contains a vulnerability and that vulnerability is medium. And that's it. It doesn't tell me how that vulnerability impacts the software that I'm using. Where in the chain it fits may help provide that information, you know, is it one level down or multiple levels down, but it doesn't really give me anything other than the ability to make a decision about should I, A, do I trust the vendor because did they disclose it, right, to your vulnerability disclosure practices, are they keeping it up to date, meaning that they have good development practices, or do I take the risk because they haven't expressed to me how that vulnerability affects me, and so what do I do with that information? Now I'm gonna, I made a little sign for this, and I'm gonna get in a lot of trouble for this I'm sure, but—

**Chris Romeo (16:51)**: My threat model is better than your SBOM, that is Yeah, that's a, that's, I mean, I don't think any of us are gonna argue against that. But Matt, on your point, I was thinking of one more facet of that too. One of the things that the SBOM is missing is we don't know if the code, the, the library that is, that's listed in the SBOM, that code might not even be used. That's, that's I think is the biggest unknown thing in open source right now is how much of that software that we're claiming that there's a vulnerability in is even being called. Within. So much of that is dead code that doesn't ever get called in your path, and it's not even accessible.

**Matt Coles (17:32)**: Contains does not mean accessible, right?

**Tanya Ward (17:34)**: That's where— but that's where when you have SBOM and you start thinking where the industry is going with VEX, the Vulnerability Exploitability Exchange System, is then you look at these components and say, how does this impact my system? Do I use it? And so you start building up, up this intelligent way of looking at how you can consume your products— or sorry, these components that are in your, um, your products. But it still, in my mind, goes back to you can still have a library. That library could consume another third-party component that they're not disclosing. And so you could think that you've fixed an issue, and then you just go back and realize that when that vendor actually discloses the vulnerability, then that component can still be vulnerable. So It's— I think again it goes back to— I'm going to keep knocking on the door. It's about vulnerability management and just good vulnerability disclosure practices. We'll never hit transparency unless everyone who builds components that goes into things are being transparent with their disclosure practices and the vulnerabilities they're fixing.

**Matt Coles (18:36)**: VEX is not a panacea either, right? Because it doesn't— vulnerability exposure is contextual.

**Tanya Ward (18:42)**: Yeah. So It's a change. Yeah, it's not static.

**Chris Romeo (18:45)**: Just real quick, for those that don't know what VEX is, just you mentioned it's Vulnerability Exploitability Exchange. Is this a system somewhere that I can go and query it as a public user, or what? Tell us a little bit more about VEX for those— I don't know what it is, so tell us just for others that might not know as well.

**Tanya Ward (19:04)**: I mean, the system is like SBOM itself, of these components that VEX has really, you know, been able to take all of the vulnerabilities that you, you know, as it relates to these components. So, you know, what you fixed, what you, you know, are still potentially investigating when it comes to these vulnerabilities, when it's, you know, those components that you may not use that are not exploitable. And so there's many different— there's different things out there, but OASIS-CSAF are doing a really good job in building machine-readable advisories that have a VEX profile that can allow you to do that. So you take the SBOM, you check it against MVD, find all the vulnerabilities, and then VEX is another layer that goes on top with the intent of being able to really understand the residual risk. But Matt, your point is good. It is, you're absolutely right, it's dynamic. Vulnerabilities are constantly dynamic. If you look at MVD and how many vulnerabilities are disclosed, you know, time over, it's on an increase.

**Matt Coles (20:03)**: Well, and how you use it, right? So if in VEX you would describe, say, libssl again, right, how libssl is used gives you context that describes how or helps inform how a particular vulnerability is going to be used and how it's going to, or how it's going to be manifested and how it could be leveraged by someone. And so, you know, this is, and so VEX is useful, don't get me wrong, the same way that SBOM has a use. but it can't be taken out of context.

**Izar Tarandach (20:38)**: So a question, uh, Chris took it to— it took what Matt put down and took it to the point of, is the code being used? But does SBOM include the environment where that code is being built? I don't know. Does any of you? Because let's say that my problem is not in my open source. My problem is my compiler. My compiler is the problem. It has this horrible habit of putting bad code inside whatever I compile, and there have been those in the past. So is the SBOM telling me this code is compiled, or this code was downloaded as is from its repository, and here's the provenance, and here's the hash for it? Is SBOM telling me you built this code, so now you are more responsible for it than you were before. It be also giving me more assurance.

**Chris Romeo (21:34)**: It's a good point because you just, you just hit on something else I'd never thought about with SBOM, but an SBOM doesn't prove that I got the right component, right? There's no component of SBOM that it does or doesn't.

**Izar Tarandach (21:48)**: So I, I'm almost sure that some of the formats out there at least do have a hash.

**Matt Coles (21:55)**: That only tells you— that tells you that you can, you can do a comparison, right?

**Tanya Ward (22:00)**: Yeah, right.

**Matt Coles (22:00)**: But you can't necessarily know that it's a still a— I mean, dependency injection still is still a thing, right?

**Chris Romeo (22:07)**: Yeah.

**Izar Tarandach (22:09)**: So let's see, SPDX.

**Matt Coles (22:14)**: And it would be up to a user to decide, you know, to figure out things like typosquatting, right? So I get from a particular site, right, uh, provenance data, you still have to make a judgment call about whether it's still the valid site.

**Izar Tarandach (22:27)**: Yeah, you see, you're going to the site for SPDX, and the first thing that they say is why, why use it? To facilitate open source compliance by accurately communicating licensing information and making such information available in a consistent, understandable, and reusable way. To me, that means lawyer too.

**Matt Coles (22:46)**: Only the licensing portion?

**Izar Tarandach (22:47)**: No, that's the first one. The second, uh, reduce redundant work in determining software license information. The third, help users comply with the licenses of open source packages.

**Matt Coles (22:59)**: And which— sorry, you're reading from which?

**Izar Tarandach (23:01)**: I'm reading from SPDX. Sorry, my bar is a bit— SPDX.dev.

**Matt Coles (23:15)**: Well, that's an interesting take on it then, right? So obviously we're talking about SBOM now because here in the US at least it showed up in the presidential executive order on cybersecurity as being a requirement. And as a cybersecurity directive, we make the assumption as security experts, hey, hey, of course they're talking to us, right? We're using this SBOM to express security information. Not strictly licensing information, right?

**Izar Tarandach (23:46)**: But then I found this article by Ryan Narain, December 7th, 2022, on Security Week, which is titled Big Tech Vendors Object to US Government SBOM Mandate because of the transparency issue, no doubt.

**Matt Coles (24:06)**: right? Because now what you're— because basically what they're saying is, basically what the executive order has said is you'll produce an SBOM for your components that we consider critical. And in doing so, they're basically, at least in my opinion, I don't know if this was actually the intent, but what they're effectively saying is big enterprise or any software vendor for that matter, we don't We don't trust you to express to us the information that we need, right, your, maybe your vulnerability management practices don't exist or they're weak or slow, right, issuing a security update once a year isn't going to help us defend our network, and we're talking about critical infrastructure networks here. So in doing so they're saying help us get ahead of your lack of care, so to speak, and express to us component information that we can use to do something with. Now that do something is kind of vague. Are they looking at automated vulnerability detection? Like, oh, these components have these types of problems or these types of components, we should be scanning them more frequently and looking for vulnerabilities.

**Tanya Ward (25:15)**: Can I just interject there?

**Matt Coles (25:16)**: Yeah, absolutely, please.

**Tanya Ward (25:18)**: So SBOM, so what is it that SBOM gives you that scanning doesn't give you?

**Matt Coles (25:24)**: Internal information.

**Tanya Ward (25:26)**: Like, give some examples.

**Matt Coles (25:30)**: Internal composition, scanners.

**Tanya Ward (25:34)**: So, are you disagreeing? Are you?

**Izar Tarandach (25:37)**: It's a loop. You get the SBOM, you get the scanning, you compare both outputs, and you see that you still have something that you can recognize.

**Matt Coles (25:46)**: Well, but keep in mind, I don't disagree with the process. I guess what I'm saying is, in an SBOM, An SBOM— so what a vulnerability scanner will find, let's say, let's take, well, we're not supposed to use names, but I'm going to— there's a couple of big vendors out there for network vulnerability scanners that do authenticated scans against systems. If you're building a system, a compiled piece of code in any form, whatever form that is, the vulnerability scanner can identify a set of things, a subset of things that are included, right? Right, it can look at open ports and services, it can do banner grabs, it can potentially figure out through DLLs what you're using if it has that capability, but what it doesn't— isn't able to do is the next level down, right. It can't do the makeup of a thing because that thing has been compiled into a binary piece of code and hides that information. So an SBOM provides a level of transparency about what does that thing I'm looking at that I'm scanning, what does it actually contain or what does the vendor say it contains because we don't, we aren't necessarily saying that there's a way to verify or prove that that SBOM is accurate and complete, right. So the vendor expresses what they tell, you know, they express some information that says this thing that I'm working on contains the set of components and I can then as a as a consumer, use that information to match it to NVD, right, to look for vulnerabilities. My scanner isn't going to find all of those components.

**Tanya Ward (27:25)**: But okay, so, but why— but could a scanner, could it, could it eventually find all of those components? Because I'm trying to figure out again what's the output of both of these systems.

**Matt Coles (27:36)**: There's—

**Tanya Ward (27:37)**: and are they the same?

**Chris Romeo (27:38)**: Is it to find vulnerabilities and be able to think, you know, You look, we're looking at it from two, there's two different use cases and we're kind of, we're muddling two use cases together here. There's the use case of I am a vendor, I build a product and I need to have a way to check my open source. I generate an SBOM and I use that to somehow do some, you know, tracking, checking of my SBOM, of my open source inside the company. The second use case is I'm a consumer, which is what the executive order is pushing. The reason the executive order was written how it is, is saying, Hey, I'm a consumer of software and I want you to send me this SBOM so I have some visibility because I can't go inside the walls of your company and look into your build systems and look in your source code control systems. You're not gonna gimme that access. The SBOM, to Matt's point, Matt said this, the transparency is the word of the day from Matt's kind of definition. But I get now kind of what you're saying. It's that it's the external party that wants to be able to look inside the organization that they can't, they're never gonna have access to be, to look at the source code. The SBOM gives them a view. And so, Tanya, I get what you're saying on the internal. I get the, and I agree with you on the internal use case, like the vulnerability, this is about vulnerability management. SBOM is about improving your approach to mitigating vulnerabilities in the things that you build. But when then I think when we flip to the other side, the other use case is SBOM is about helping me to see, to gain visibility. Now granted, as I said from the very beginning, It's gain visibility into the past. I'm not looking in the future, I'm looking towards the past as far as what you did before you sent it to me.

**Izar Tarandach (29:16)**: Okay. Again, if we look at the food and the pharmacy example, it's attestation and it's assurance. The vendor is attesting that here are the things that I think are inside the thing that I'm selling you. And the buyer is having some assurance that they have, again, some transparency, some understanding of what, what they bringing in, right? Now, both sides, perhaps the, the maker less, but both sides are, are free to not use that information as they see fit. But that information exists somewhere that everybody agrees has a certain degree of being official that it could be used for attestation or assurance, right? right? Now, what these guys, all these vendors are basically saying in their letter to the government is that, where is it? At this time, it is premature and of limited utility for software producers to provide an SBOM. We ask that OMB, that's the government agency, I believe, discourage agencies from requiring artifacts until there is a greater understanding of how they ought to be provided and until agencies are ready to consume artifacts that they request. Not a lawyer, don't like to play a lawyer, love lawyers, but to me it sounds like, okay, you're asking us to give you something that you're not going to use anyway, so why don't you stop asking us?

**Tanya Ward (30:44)**: So do you think when it was asked, I mean, we all know that to be able to produce an SBOM, it's a lot of work, right? To be able to automate that pipeline is is there's a lot of work involved. And so when the mandate came out, do you think that they thought it would be potentially— when I say they, like the recommendations by the government— they thought it would be very easy to press that button? And do you think there's enough guidance out there to help vendors? I mean, what do we actually put in the SBOM? Do we put IP? I don't think we want to put our own You know, I mean, we have to protect our own IP of, you know, for our own company. So, you know, I'm interested in your thoughts.

**Izar Tarandach (31:29)**: I'm going to disagree with you that it is so hard to create an SBOM, right? There is stuff like OWASP Dependency Track that, that's an amazing system that does it. What I think happened, and again, I'm, I'm saying this as an observer, I'm not involved into SBOM efforts, I don't work work of SBOMs day after day. But, uh, uh, it became a bit of a shiny thing. You know, everybody that, that I hear, every time that I hear SBOM, to me it sounds like jazz hands, you know, SBOM, we're going to solve all the problems with SBOM.

**Chris Romeo (32:03)**: So I sit on a, I know some of you do as well. I sit on a major conference committee's program selection process, and last year was the year of SBOM. Like every other submission was like, SBOM will save the world. SBOM into the future. Like everything was SBOM, SBOM, SBOM. This year it was all about supply chain, but last year it was definitely over. It was a lot of SBOM.

**Izar Tarandach (32:25)**: Supply chain and SBOM go hand in hand, right? Because SBOM is going to save us on the supply chain. There's a lot of people with a lot of good ideas out there and very practical ideas, I must say, which is a rarity sometimes in some subjects. But, uh, the point is, are we seeing that impact? Now, to be totally transparent with everybody, I thought that we were. I, I thought that S-bonds— okay, first of all, I think that the idea of S-bonds is a good idea. I think that we need them. I think that we, we have to have the transparency. I thought that, uh, uh, S-bonds took us to a better place. And, and that's why I was probably so surprised when we decided that this was going to be a topic of conversation here because I was like, wait, everybody agrees that there's something good out of S-Bonds, right? But then I found this thing and I said, well, apparently not everybody. But I do like the idea. I do like the fact that it gives me something to go and compare reality against. Like, hey, here's the list of the things that you said that I could expect to be here. But if I go and look, it's not actually so. But then I stopped asking myself, if The point of contention here is IP. Uh, it seems to be concerning itself mostly with licenses. It seems to be concerning itself mostly with open source. It seems to be concerned itself through VEX with CVEs and things that we know are bad, and we are trying to tell people how bad they are.

**Tanya Ward (33:55)**: I, I—

**Izar Tarandach (33:56)**: exposure here.

**Tanya Ward (33:57)**: I, I think it's a good opportunity as well for us to kind of look at the kitchen, you know. It's like, you know, is this a messy kitchen? How many How many versions of OpenSSL do we actually embed in our product? Do we need to embed all those versions? How many could be potentially end of life? So I think there's, you know, it's like if you're going to, you know, serve a nice dish to your customer, you want to make sure your kitchen's clean, you know.

**Chris Romeo (34:19)**: So yeah, that's a good way of thinking about it. And I would say one other thing just to kind of close my thought on the executive order is that I've never seen security innovation kind of pushed from a government document or standard. I've never seen the government take us to some type of security, you know, grand place where we feel like that's where every, every, all the goodness is coming from. And so I don't love the fact that, and I get what these tech vendors are saying as far as like, hey, you're pushing this thing that you don't even really understand. You don't even really— aren't really going to use. But yet, coming back to Tonya's point, I'm with Tonya on this. It is— to roll SBOM out in an enterprise is— to roll SBOM out for one project is easy. To roll it out for an enterprise is— I need an army— not an army, but I need 5 people probably to figure it out. And it's probably going to take them 2 years to roll it out across a large enterprise.

**Tanya Ward (35:19)**: And Chris, I think it's also based on the complexity of what you're building. If it's a simple application, maybe, you know, to your point, maybe not so much. And it depends on how many components you consume as well, right?

**Matt Coles (35:31)**: So, so that's what I want to touch upon. SBOM is a forcing function, right? It forces somebody to think about component management in a way that we've been screaming about for decades, and only now people are taking an strong action on it.

**Chris Romeo (35:50)**: What forces it though? Like doesn't the person, the people that are building the product have to already have a good attitude about simplifying their product or whatever? Like there's nothing— SBOM doesn't force me. I can have a crappy SBOM, right? There's nothing that prevents me from having crappy SBOM with 12 versions of OpenSSL.

**Matt Coles (36:08)**: Well, you're making a judgment call on good or bad, right? Versus complete or not. Right? So I'm going to— I want to sort of table the judgment call there for a moment. In mandating that companies have SBOMs, they're forcing companies to have good component management practices, which has always been a security— a secure engineering practice that teams— that companies should always have been doing, but nobody does really well. Unless you take the time to do it, right? I imagine there are companies like, you know, Raytheon or Lockheed or other government contractors who know everything that goes into the systems that they're building because they have security assurance practices and system management practices and do proper systems engineering, but they're not building, you know, latest greatest SaaS services that get released on, you know, 800 commits a day kind of thing. Right? And so the tooling is a help, you know, producing an SBOM over an enterprise, it's sort of like manufacturing, right? You have to set up one and then you can cookie-cutter that into other, into additional lines. That initial setup is hard, I 100% agree, but it's hard because if those practices are lacking, you have a code base that has stuff in it that you don't know about and you haven't been keeping track. Right, and that's I think the challenge here. Ultimately for me it's a forcing function. It says, it's the government saying we need this level of information even though they don't, or maybe they don't, but they're forcing a practice into an environment and as a result there's a scramble now. How do we address this? This challenge.

**Chris Romeo (38:00)**: I want to come back around and challenge you again on this idea of— you stated that this is going to make component management better. Those weren't exactly the words you said, but it was like you said—

**Matt Coles (38:12)**: I don't think I said that.

**Chris Romeo (38:13)**: You said that it was going to— I don't remember what the exact words were, but you said something—

**Tanya Ward (38:17)**: Good component practice, yeah.

**Chris Romeo (38:18)**: Good component practices. And my point is SBOM doesn't force that. No, I agree. That is the relationship. Okay, so I don't remember what exactly you said, but— I think it was about—

**Tanya Ward (38:31)**: Can I jump in? So I like good component practices because it forces you to look at what you're consuming. And again, it's what kind of practices are they instituting? Are they, again, are they following good vulnerability disclosure practices? Are they being ethical in how they're building these components? Like, I think it just opens the door for more conversations about how you're consuming things in the supply chain. And I remember giving a talk, I think it must have been several years, on something similar about Titanic. If you think about Titanic, when the boat went down, it was just one nail— like, it was a series of nail bolts that were actually purchased to be able to hold the hull together. And the quality of that bolt was actually— was found to be very high in sulfur. Prone to fracturing in temperatures in icy waters. So I think it's, it's, it's good that we look at suppliers' practices because ultimately we're only as strong as our weakest link.

**Matt Coles (39:36)**: 100%. And then that's— I think that's what I was— sorry, I just want to kind of— I want to, I want to close out on what, what Chris was saying. I think what I said was there are companies out there that have good component management practices. They keep track of what they, of what of what they're including, of what they're building in, and they have a history of doing so, then that, and that really helps them with what they're producing, and that SBOM is going to drive those behaviors. I'm not saying it's going to make somebody have good component practices, and obviously they have to catch up if they don't already have them, but it's going to, it should help, or it's the intent, I think, is to kickstart those activities.

**Chris Romeo (40:20)**: So it would, are you saying it would help because let's say you're buying software, you look, I, you're buying software from me. I send you the SBOM. I have 12 versions of OpenSSL in a, uh, 1,000 lines of code product that I provide to you, 12 versions of OpenSSL. So is it, is it you giving pressure as my customer pressuring me that you think is gonna make that where SBOM would enable better component practices? Is that the driver?

**Matt Coles (40:46)**: No, what, what you just described was you as a vendor had to produce that SBOM. And you had to produce an SBOM that you weren't outright lying to me as the consumer. So you created component management practices to make that happen. You've improved. Now, that wasn't the intent. The intent was the transparency, or the original intent, I think, was transparency. You as a vendor tell me what you're using, because I don't I trust you to tell me what you're using previously, but now in order to produce that you have to build systems and practices and processes to actually produce the document, which means you have to have a handle on your components. Now, I was looking back through some old SafeCode publications and I can't find it because I think they pulled the latest version, or the oldest version, but a component inventory has always been, or has been a long-standing secure development practice for a number of companies, maybe not everyone, certainly not everyone, but these are not new to keep track of what you embed. And so for a company that isn't— has either lost sight of that or a company that isn't already up to speed on component and asset management, then in order to produce an SBOM, you have to have some of those practices and that If anything, it's what I was getting to is it will force companies who want to be— who need to be transparent and force companies who want to have integrity and correctness in that document to have those or to build those practices. And that drives this whole— I think that drives some of the commentary that's coming from some of the larger tech companies especially. We have such massive systems that we're producing. We don't yet have a good handle on this, and if you're not going to look at the information, then why should we spend time and effort and money and resources to build out this infrastructure? It's an interesting conversation, and I want to be careful that, you know, I, like some of us here, work for an organization that may or may not have been in that list, and I want to be careful, of course, to not make anyone upset on that side.

**Tanya Ward (43:00)**: Nope. Matt, you focus a lot on what's being provided to the consumer, like that level of transparency. And I think this goes back to your point, Chris, we can't ignore the other scenario, which is as me, as a provider to my consumer, there also needs to be a level of transparency about what's coming to me. You know what I mean? It's a chain of transparency. I can't be transparent with you, Matt, Unless Izzar's being transparent with me.

**Matt Coles (43:31)**: Yeah. So there's a level of trust. There's a, there's a trust chain that exists.

**Tanya Ward (43:35)**: Exactly.

**Izar Tarandach (43:37)**: There isn't because there's nobody enforcing signatures and all that good stuff. There is a chain of, uh, we would like to trust you going on. But I, I want you to draw back to the point that both Tanya and Matthew keep using words like practice, right? And basically what I think you both are pointing at is that this thing doesn't live in a vacuum, it lives in a process. And if you don't continuously improve that process, then you're not getting the most that you can get from this S-bomb thing, as you're not getting from any other thing basically in, in, in AppSec or, or in what, what we do. But, uh, Daniel, just the last point, you pointed to the Titanic and the boats. But think about it, one of the most, uh, careful organizations in the whole history of mankind, NASA, with practices around knowing every single thing.

**Tanya Ward (44:33)**: Yeah, exactly.

**Izar Tarandach (44:36)**: Okay. They still had exactly the same case as the Titanic in the Challenger tragedy. So at some point you have to look at it and say it's human nature. We, we, we make these mistakes. How can we stop ourselves from making these mistakes? And that's when Matt raises the, the point of upset models better than your S-bomb. So the S-bomb is only one more piece of mitigation, one more tool that we have in the arsenal of things that we put forward to lower that, uh, I don't want to call it risk. Oh my God, I hate that word risk. To lower the bad things that might happen at some point and have some impact in the thing.

**Chris Romeo (45:16)**: Which is the definition of risk. We're just about out of time, but I'm going to stir the pot here because I can't resist. I can't resist stirring the pot one more time. Is an SBOM a compliance play? Is this— is an SBOM a compliance thing? Is this something— is it a security thing or is it a compliance thing?

**Izar Tarandach (45:40)**: Is somebody on compliance right now saying stop everything, you're going to have an SBOM, create one right now? I don't think so. It's probably—

**Matt Coles (45:48)**: I think the executive order did.

**Izar Tarandach (45:50)**: What?

**Matt Coles (45:51)**: I think the executive order did.

**Izar Tarandach (45:53)**: Yeah, but does it have a date? Like it already said, I'm not going to buy anything from you if you don't have an SBOM by X date. I don't know, I didn't read it. I, I don't know where I've seen that yet, but I mean, unless it has that date and unless it has the mechanisms in place to figure that out, which these people who are complaining about it apparently are saying there aren't, then it can't be a compliance thing, right? I would say that it's a nice thing for security, but again, it's a nice thing.

**Tanya Ward (46:23)**: I mean, but this goes back to your point about SafeCode. I mean, we've been talking about component inventories for a long time and third-party component vulnerabilities. I, I think it's, it's always been there. It's always been a security concern. I think compliance is just now, you know, and you can argue about what compliance is, but if we were to use that word in this context, I think it's putting a mandate that this is some— it goes back like you can't be— you're only as secure as your weakest link. And if you're, you know, you— unless you have that peripheral in terms of what they're doing, you know, when I mean they, that the components that you're consuming, um, it's, it's very hard to get that it's really, I think, hard to understand the overall security posture of what you're building.

**Izar Tarandach (47:07)**: So I think that what's pointing out is that, as always, we are looking for assurance. Like, give me something that can elevate the trust that I have in these things that I'm building and these things that I'm using to build things.

**Tanya Ward (47:18)**: But I think it's a security problem as well.

**Izar Tarandach (47:21)**: Yeah, no, no, I, I mean assurance as part of the security.

**Tanya Ward (47:24)**: Okay, got it. Yeah, yeah, right. Matt, what's your thoughts? You're quiet.

**Matt Coles (47:34)**: I'm thinking. It's like thinking. It's, you know, yeah. As I think I've sort of alluded to, an SBOM as a component inventory, and the fact that you have to have material that you can use to create that document or that inventory. I've been saying document, but it's not a piece of paper, right? It's a data structure that lives in memory somewhere. In order to produce that, and in order to produce that, you have to have data that you can use to produce that. You have to have information about the components that you use and being able to express that Right, so you have to have some component practices in order to make it complete and accurate. You have to have a good process around that, around how you embed components and how you bring them into a software system, and how you manage those components. And so, and so what, sort of, SBOM is the forward-facing, like the front door of a house, and I'm going to use a horribly bad example, maybe a better example here is the SBOM is the— oops, we lost Nizar. The SBOM is the dish that you get at a restaurant and the instruction or the information that comes with it, you know, the nutrition facts that comes with it. In order to produce that though, there was a lot of stuff that happened in the backend to produce that. And if SBOM is a— becomes a driving function for those activities and those processes, then SBOM has a security value. If SBOM is solely going to be a, you will produce this in order for us to buy or continue to use your product, and they're not going to actually use that information to make important business decisions other than use or not use, then it's a compliance problem and really has no security value whatsoever.

**Chris Romeo (49:43)**: Yeah, I think from the executive order perspective, it's set up to become a compliance step. Like, you know, I mean, I kind of grew up in security in the world of Common Criteria of going through this evaluation process and then getting a certificate, and then that allowed you to sell to global governments around the world. And yeah, we did improve security in pockets using those type of processes, But at the end of the day, we weren't doing it to improve security. We were doing it so that we had a license to sell to the federal governments of the world, which, yeah, when you work for a big company, that's a lot of, that's a lot of comp, a lot of nations and a lot of people that take that thing seriously. But at the end of the day, I was pushing compliance paperwork across the board. And so I think when I look at SBOM, I think it does have a lot of, it has a lot of potential upside. But it also could be just another compliance rubber stamp of, do you have your XML file attached to your purchase request or whatever?

**Tanya Ward (50:45)**: Yep.

**Matt Coles (50:46)**: And I'll also add that, you know, we're talking a lot of— we're making an assumption about the systems that we're talking— that are being discussed in this are primarily software-based systems. I have another sign.

**Chris Romeo (51:00)**: HBOM should be a thing is what the sign says.

**Matt Coles (51:04)**: HBOM is a thing. I actually looked it up. CycloneDX has a format for just expressing hardware information. And so when we're talking about mixed systems, having an inventory of all of the components that go into a system, whether software or hardware, can drive those practices. Well, it can drive those practices and allow people to make informed decisions. Decisions, and ultimately maybe that's the ultimate goal is allow people to make informed decisions, but I agree with Chris, you know, this— there's a danger here that this is going to become a legal thing and not a security thing.

**Chris Romeo (51:38)**: Yeah, I think that's a good place to end our conversation today. I do want to put one caveat in here. We do have, all of us, we have a lot of friends that are really working to push SBOM forward. I'm not trying to take away from anything that they're doing. Thank you for, you know, I think of like Steve Springett and everything he's done in, in SWIFT cycle and DX and, and supply chain and, and stuff within the OWASP world. My, my intention's not to take away from anything that those folks have done. My, my thought in this conversation is really just, let's really truly understand what's the value proposition here. Like what, if we're gonna do all of this stuff, is it really worth the effort? But, um, to those that are working on this, keep going. Like I do, At the end of the day, if push comes to shove, I do think this is the right thing to do. I think it does add value. Some of the things I was bringing up were just, let's push the edges and see if we can defend the value that this is generating. So folks, thanks for listening to another episode of the Security Table and tune in soon for another very interesting AppSec or software security conversation.
