# Lastpass and the Security of Security Products

URL: https://securitytable.ai/lastpass-and-the-security-of-security-products/
Published: 2023-01-14
Duration: 3572 seconds
Season: Not supplied
Episode: Not supplied
Hosts: Chris Romeo, Izar Tarandach, Matt Coles

## Show notes and chapters

The gang discusses the Lastpass breach and the need for the security of utility-style security providers. We discuss Lastpass from a different angle - the responsibility of "hard security" providers.   
  
As security practitioners, we have been telling users to "just use a password manager." So what do we do now? How do password managers impact the way we give advice?

Lastpass is as "hard security" service as it can be - are security people taking things as seriously as they should? Are we too "here's your two months of credit policing, thank you very much" accommodated? We explore and reach some conclusions.

  

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 (0:01)**: Hey folks, welcome to another episode of The Security Table. This is Chris Romeo, and I'm joined by Izar Tarandash and Matt Coles. And today we're going to talk about security issues with hard security providers. But first, Matt, I heard you just got something that you bought for yourself for Christmas. What is this thing?

**Matt Coles (0:25)**: I'm, I'm wearing it right now.

**Izar Tarandach (0:28)**: Okay.

**Chris Romeo (0:28)**: So this is, I knew that after the holidays, somebody at the security table was gonna come back with new gear.

**Matt Coles (0:34)**: Yeah.

**Chris Romeo (0:35)**: From a recording.

**Matt Coles (0:37)**: Oh, well, yeah. He, he, he takes the cake on new gear.

**Chris Romeo (0:39)**: Okay. I don't even know what that is. It looks like an asthma inhaler, so I'm not that.

**Matt Coles (0:43)**: It's a Twinkie.

**Izar Tarandach (0:44)**: Oh, excellent.

**Matt Coles (0:46)**: Uh, I'm an, I'm an avid gamer, so, uh, I, I needed a new headset and, uh, I went one with, uh, surround sound and spatial audio and all that so I can hear footsteps from coming from different directions.

**Chris Romeo (0:58)**: So nice. We'll see how it goes.

**Matt Coles (1:00)**: It's not going to make me suck any less.

**Chris Romeo (1:02)**: So if one of us walks into the scene here from a different angle, you're going to know. There's no sneaking up on you in cyberspace.

**Matt Coles (1:10)**: That's right.

**Chris Romeo (1:10)**: That's what we're talking about.

**Matt Coles (1:12)**: That's the theory.

**Izar Tarandach (1:13)**: But we also have one voice on each side, so we can do the good voice and the bad voice.

**Matt Coles (1:17)**: That's true.

**Chris Romeo (1:17)**: Okay. Ooh, that could come in handy in the middle of a debate. You know, the angel on one side, the devil on the other side, kind of talking about something related to security.

**Matt Coles (1:26)**: But oddly, I'm on the one side of the image. Image here, so not in the middle where Isaur is.

**Chris Romeo (1:31)**: That's true. If Isaur had that special headset, he would be stuck with two different voices in his head.

**Izar Tarandach (1:38)**: Two more.

**Chris Romeo (1:40)**: I knew that was coming. I knew it was coming. Well, let's talk about LastPass. So, you know, the topic that we have for today, LastPass has had some security challenges here over the last few weeks. And so, I want to start with a caveat, a disclaimer. Just to say that we don't dislike LastPass. Those people that work at LastPass, keep going, keep working hard, keep trying to secure what you're building. We just want to explore the issues around what happened here, maybe ask some hard questions to help everybody get better at securing the things that they build. So we don't mean any insult here. We just want to explore the issues. But, yep. So, as security people, how many times have we said, just use a password manager. Are you using a password manager? Why aren't you using a password manager? To everyone from our colleagues to our elderly relatives in our family. Like, we've been telling everybody to do this, and then we have a security issue with it. And so, I guess, first of all, what does that mean? Like, what does that do to the people that entrust their IT needs and support to us as technology people?

**Matt Coles (2:56)**: Before we jump in and answer that question, actually, can we start with, let's talk about what a password manager is and what it gains the people who use it, right?

**Chris Romeo (3:05)**: Yeah, sure.

**Matt Coles (3:06)**: So a password manager obviously is, well, or not obviously, as the name implies, is a place where you manage credentials, passwords, or credentials, because there could be other things you put there. And these come in various forms, as we know. We talked about LastPass, mentioned LastPass. That's a common popular one that people use. There's 1Password, there's KeePass, there's Password Safe. There's also password books, which are physical password managers people write on paper with pencil and paper, which we could also consider password managers. Or sticky note under the keyboard, which we don't want people to use. So when we're talking about password managers, we're talking about ones which are electronic or online primarily, I think, for the purpose of the conversation, right? And of course, the benefit of password managers generally is the ability to store your credentials in one place. That one place can be protected by a master secret or some, uh, something that, that provides access to that without without potentially compromising, well, except for if bad things happen, like we know, like we can talk about, that master credential will store all of those others safely. And then different password managers may have various capabilities like reminding you to change your password on a regular basis. We should also call out many browsers like Internet Explorer and, or sorry, Edge and Chrome and Safari and others can also have password manager-like capabilities as well. And so there's pros and cons to each, and correct me if I'm wrong, is our, the premise of the conversation today is really we've driven users, we've been successful in driving users to use these password managers. But now we have a problem where the reason we sent them there was to secure their credentials, and now we have a case of insecurity of those credentials.

**Izar Tarandach (5:18)**: So I think that that's a starting point, right? Because the way that I was looking at it, we have a bunch of questions here that we probably want to ask ourselves and the market in general. There is the, the side that as security people, security infrastructure, or security teams or something like that, we have been giving people the, not only the password manager orientation, but a bunch of other tools that we say, "Hey, use this and you're going to be more secure." And I don't think that we quite have an answer for what happens when one of these blows in our faces. So that's one thing that I would like to, to look at. Then there's the other one of, uh, tools and systems with more of a security value, and apparently they are expected to make the same claims of application security of anything else. So my question is, should we look at these things differently and require from them more And if so, how can we do that? What, what would be a good mechanism to say, hey, you are, uh, what's the name of the, um, like electricity and water?

**Matt Coles (6:38)**: You are a utility.

**Izar Tarandach (6:40)**: Yeah, yeah, you're a utility, so you have to, to, to stand by a higher standard. So those are basically two avenues that I would like to, to explore with you guys today.

**Matt Coles (6:55)**: Okay. So, should we talk about, shall we bring up LastPass here and talk about what happened to LastPass that sort of prompted this conversation?

**Izar Tarandach (7:08)**: Yeah. So, in my view, and I probably don't have all the facts, but from what I have read, so LastPass apparently suffered a breach somewhere around August last year. And they came out and said everything's fine, nothing got accessed, and all is good. But apparently, as part of that breach, some credentials went out, and they later were used to actually go into the storage of the vaults, the encrypted material from the customers, exfil all that And later, LastPass comes forward and says, "Well, you know what? We looked at it again, and yeah, everything is out there. But it's okay because we don't have the master password, so they didn't get breached, they didn't get exfilled, and everybody with a good master password is okay." Okay, so we have a technical failure and an operational failure and bad PR practices.

**Chris Romeo (8:24)**: Incident response, incident response, right, is the process that drives what we go to the market and what we go to the comms with as far as the current status of an incident. And here's what we know right now, and here's the caveats that go with it. So it's, it's an incident response play.

**Izar Tarandach (8:40)**: But wait, there's more.

**Matt Coles (8:43)**: There always is. Why not?

**Izar Tarandach (8:46)**: So apparently when the vaults were exposed, it got figured out. Well, basically other people knew already from some time ago that not the whole vault is encrypted. The passwords, yeah, they are. But for example, URLs are not, and IPs of where you have connected to LastPass LastPass are not. So I'm looking here for— unfortunately, I closed the window— among my 150, there was one security researcher who years and years ago went over some of these issues, talked to LastPass, and apparently even though they said, yeah, we fixed stuff, not everything got fixed. And he also points at the fact that, yeah, so they have, they use the PBK, I always forget, PBKPF2.

**Chris Romeo (9:49)**: KDF, KDF. Yeah. PBKDF. Yeah.

**Matt Coles (9:53)**: For Based Key Derivation Function version 2.

**Izar Tarandach (9:56)**: Thanks for that. Uh, they use that to derive the key that they actually use for the, the, encryption decryption from your master password. And that requests a number of rounds of the function to create, to derive the— and apparently they started at a very, very, very low number of, uh, rounds, and progressively, progressively they upped their default, but they did not enforce with the customers that, hey, there have been changes, so now we need you to re-encrypt your stuff or redo your stuff so that you can take advantage of that race, right? So I think that on top of all the failures that you guys pointed out, and we are at what, less than 5 minutes in, I think that there has been a threat modeling problem here.

**Chris Romeo (10:52)**: I knew it was going to happen here eventually on the Security Table. We would get to threat modeling and we would never go back. That's where we are now.

**Matt Coles (11:01)**: Well, and so also the level set with our audience here, of course, for those who don't know about LastPass or don't use LastPass, if this was a one-time incident, that would be bad, but this is not a one-time incident, right? This is a pattern of behavior that this particular company has had over the years, but they're a high-value target. And to your point, their utility, their infrastructure, right? And we know the critical infrastructure, whether that we're talking about, you know, gas and oil or electricity or our roads and telecommunications or the stores that people use to secure all of their credentials to all their banking and social media accounts, et cetera, they're gonna be a target. And so they've had a number of issues over the years.

**Chris Romeo (11:52)**: Let's talk about the primary architectural flaw, though, because I chose and started using a password manager about 10 or 12 years ago, and I chose one for a specific reason that is the opposite of this issue that's happened here. I chose a password manager that separated the synchronization and storage of the password files from the password manager app. So, I'm not going to say which one I use because I don't want people to try to figure out what's wrong with it. But basically, what it does is it gives me the option to store my primary password file that's encrypted on Dropbox or OneDrive or Google Drive or any, all these different storage mechanisms. So, the advantage of this architectural play is that If you break into this, the company that I'm using, you don't have access. They don't have access to my files. My files are in a cloud storage provider that I've chosen, that I've placed the file where I want to. And yes, there's a little bit of security by obscurity, but there is also a lot of complexity. And you can't get— there's no giant vault of passwords that's the size of the Titanic, right? That's got every password on earth. And so, when I originally, like I said, when I originally picked a password manager, I looked at these architectural issues and said, "I don't like the idea of a password manager also having a cloud storage component that somebody, an attacker, could slip through and get all the way to the vault and grab all my stuff coming out." What do you guys think about that?

**Matt Coles (13:31)**: Well, Matt. Definitely, I agree. I went through the exact same process. We may be using the same password manager, we may not be. The, you know, and I think this, I want to tee up a part of the conversation, we don't have to solve it now, but we do, we will get to it, I think is the notion of use a password manager and what does that mean? And then allowing users or expecting users to make a choice and make an informed choice about which password manager to use. You had a set of goals in mind, right? You wanted to protect credentials, but you knew that there were attacks out there, right, or you knew that there were concerns with how the data would be stored and the fact of would you trust a company to manage the infrastructure for you or do you have to, you know, take more action, right? So when we say use a password manager, we merely mean store your credentials somewhere securely. The password manager companies either provide a utility to do so, like the ones that you and I may be using, versus provide a lot of value-added services that is very useful and makes it palatable for use by, you know, average consumers, and they take on more risk. Maybe without, uh, the informed consent part. So I agree with you. Sorry, I know I interrupted you. I'll pass it back.

**Izar Tarandach (15:07)**: So I think that both of your points touched into, into a very important point in here, is that you did do that risk analysis before you decided what to use. But in the same way that in utilities, you just stick something in the outlet and you expect it to have electricity, or you open the faucet and you have water. Most people have been indoctrinated even by us to just expect these things to work and be good enough. Right.

**Matt Coles (15:38)**: Us, you mean security people, not us, you and me.

**Izar Tarandach (15:41)**: And Chris, well, we, we are there.

**Chris Romeo (15:44)**: We are guilty as well. We know it.

**Matt Coles (15:46)**: Yeah. Yes.

**Izar Tarandach (15:47)**: Yes. The thing is that most of them not only are not equipped to do that risk analysis, but they also don't feel the need because, again, the experts told them that it's okay.

**Chris Romeo (16:01)**: Yeah.

**Izar Tarandach (16:02)**: So my question here is, when we as experts decide that we are bringing a new app into one of our organizations, we do this risk analysis. And usually, it's sending a questionnaire, getting a bunch of answers. Sometimes, you talk to their security people and you hear, "No, everything's okay. We use all the good standards. We have all the good controls. We have all the certifications and compliance and whatnot." Is that enough? Apparently not.

**Chris Romeo (16:34)**: Yeah. I mean, when you think about what all of those things you just described— well, let's talk about assurance, right? Assurance is an old word. But when I got started in security 25 years ago and assurance was our primary focus, how can we assure that something that's being built is secure beyond some reasonable doubt? And all of the things that you just described here, questionnaires and compliance-based certifications and things, let's be honest, they've moved the needle some, but do they make things secure? Like, do I get a SOC 2 certification and/or SOC 2 certificate from a company and go, Yep. They got security altogether. No, they answered a bunch of audit questions and they knew how to put the paperwork together in a certain way.

**Izar Tarandach (17:21)**: A bunch of different questions whose interpretation is open on both sides, both on the auditing and on the responder.

**Chris Romeo (17:27)**: Yeah. So, at the end of the day, that didn't give me a level of, "This thing's secure," or, "This thing is protecting privacy." And so, I think, you know, there are challenges there. But to come full circle or go, you know, push this even further out is like Underwriter Labs has tried to write security standards for IoT. I don't know how successful they've been. When I think IoT, there's a lot more constraints. I think IoT is more constrained than a cloud-based service that's doing something for somebody. Because you've got hardware on IoT, you've got things you can enforce at the hardware level to then ensure that you're getting better security functionality. And so, like, I think people have tried to solve this. I don't think anybody's been anywhere near successful in trying to say, can we create a standard that we could all get behind that if we said, if you pass this set of requirements, we would call your thing secure.

**Matt Coles (18:30)**: I think in Europe— sorry, in Europe, I think that they have been better, more successful in the European Union in setting setting rules for certain device classes of devices.

**Izar Tarandach (18:43)**: And the privacy bit.

**Matt Coles (18:45)**: And with privacy as well, yes, you're right. So they have a better regimen. It's founded in regulation and so there's, I mean, there's legal implications to it. Now, UL you're referring to is UL 2900, which is the IoT security standard. And UL 2900 has also been recognized. I'm just doing a quick look here just to see about some new updates for it, you know, it being looked at for medical device security as well. So, but that's not a regulation per se, although with the FDA on board, it may become part of the regulation. But that's— it's hard because there's a matter of who do you trust to deliver good standards, and of course we rely on folks like NIST, which ultimately ties to ISO, right, as the international standards body. And then what does that, if you comply, if you meet the requirements, how resistant really are you to attack? That's a hard problem to solve. And then what reason do vendors have to actually implement those things? If they're not going to get sued for doing, for not doing so.

**Izar Tarandach (20:05)**: Oh, oh, let's put more in the mix. So disclaimer, not a lawyer, don't play it on TV.

**Chris Romeo (20:13)**: Not a lawyer.

**Izar Tarandach (20:14)**: But there is a thing going on in Massachusetts. There's a class action in the making against LastPass. And from what I read and I saw, the class action lawsuit is demanding the court force LastPass to apply better security practices in addition to paying affected consumers damages. So we are going to get 3 months free of credit reporting.

**Chris Romeo (20:41)**: Excellent. Needed, necessary.

**Izar Tarandach (20:43)**: Awesome. Let's stack on top of the last 3 months and the last 3 months and the last 3 months. But what is this thing about demanding the courts to force LastPass to apply better security practices? Are we about to have court-mandated Application security?

**Chris Romeo (21:01)**: It would be great for our careers, but who the heck is going to be able to— how are they going to know what's— how to make something more secure?

**Izar Tarandach (21:08)**: Well, should we start teaching application security in law school first?

**Chris Romeo (21:12)**: Yes, I think so. Yes, we might have to.

**Izar Tarandach (21:16)**: We already have patent law and we— and oh, to be honest, I know some lawyers that work in some companies that are great application security professionals and they use that every day. But is that going to become a thing?

**Matt Coles (21:30)**: Well, so we already have now precedent for the, for instance, in the US, we have the US Cybersecurity Executive Order, right, from the Office of the President indicating that federal agencies and contractors who work with, you know, who sell to the federal government need to implement security in a measurable, in a defined, measurable, and provable—

**Chris Romeo (21:56)**: Repeatable.

**Matt Coles (21:57)**: Way. Well, not just repeatable, but provable more importantly fashion, and they set out the guidelines for doing so, right? And obviously that ties back to NIST and FIPS and others who are gonna be generating the actual rules you have to follow, but that's the reason for doing those things, and if the courts follow that model, in other words, what defines reasonable, what defines, you know, secure, point to the executive order or NIST requirements or whatever, then it may be okay.

**Chris Romeo (22:38)**: I'll say— What has the executive order done for us so far?

**Matt Coles (22:41)**: Nothing, because it's all about rulemaking so far, right?

**Chris Romeo (22:44)**: Yeah, I mean, you know, one of the things that I can say, I mean, I've been on this earth for a number of years and I could never look and say, "Hey, the government is here to solve my problem." I think it's a real attempt though.

**Matt Coles (23:02)**: I think the executive order though is an attempt and it's been done before. In my experience with working on IoT and medical devices and safety critical systems in the past, You know, the FDA, for instance, put out guidance around pre- and post-market cybersecurity guidelines. Now, those guidelines were fairly high level, but they were based on industry best practice at the time around, you know, having secure requirements and design and looking at your code and managing third-party components and all the stuff that, you know, that OpenSAM and BSIM prescribe. It's a beginning. It's a beginning.

**Chris Romeo (23:40)**: Yep.

**Matt Coles (23:41)**: It at least gets people talking about it, gets it into their mindset, gets it into their schedules, and it's—

**Chris Romeo (23:46)**: But in both IoT and medical, the difference here is you have a hardware component at the core.

**Matt Coles (23:52)**: Agreed.

**Chris Romeo (23:53)**: And that's the difference because if I have a hardware component at the core, this thing got a lot easier, right? Because you can say, "Oh, I need to have a trusted," what are they, Trusted Platform Module. I need to have secure boot coming off. I need to be able to validate that the software this thing is running is actually the software that we signed coming off from our CI/CD pipeline. The way I do that is with TPM and having root of trust back to the hardware core. Now we're talking about LastPass as a cloud service, though. That's a whole—

**Matt Coles (24:20)**: that's like you just launched yourself into the middle of the universe and said, "Figure out where the moon is." I will caution you, though, that cost and space and technology pressures do make it sometimes difficult, especially for IoT and some medical devices. Let's think about hearing aids, for instance, which don't have full-fledged compute and environments that maybe don't have a TPM. So it's not just simply 'cause there's hardware that that's a done deal, although your attack scenario is a lot different, right? I have to physically have access to the device or be at least local to it with respect to Bluetooth or whatever, as opposed to LastPass, which exists ubiquitously in the cloud. Yeah.

**Chris Romeo (25:00)**: And that's my point is the cloud versus a medical device or an IoT device is just, I mean, 100x more, The threat landscape is so much broader because everything's out of your control at that point. My argument is with medical and IoT, some of it is in my control as a builder.

**Izar Tarandach (25:20)**: So, to roll it back and put a bunch of things together. Okay. So, security-valued companies because they provide a service that people are building on top of to get their basic security. It can't get more basic than passwords. And let's for a second forget the incoming passwordless thing and all that good stuff. They might start getting hit, if this is a precedent, with class action suits that are going to lead to a financial cost that are probably going to move them in the direction of doing something preventative rather than detective and rather than defending themselves post facto.

**Matt Coles (26:11)**: Can I just interrupt you for a second?

**Izar Tarandach (26:12)**: Sure.

**Matt Coles (26:13)**: LastPass doesn't not do security proactively.

**Izar Tarandach (26:17)**: Oh, oh, wait, there's more. There's more. It does, but we agreed in the beginning, remember when I brought up that word? Threat modeling, that perhaps their threat model wasn't as complete as they thought it should be. So again, not a lawyer, but I understand that a very big concept in those things is best effort. If you want to show best effort in your defense efforts What is the good mechanism to put him on the table and say, see, I thought of everything that could have gone wrong?

**Matt Coles (27:00)**: I would be careful about— is best effort really mean, uh, a measure of completeness, or is it reasonable cost?

**Izar Tarandach (27:11)**: I, again, not a lawyer, but I would say that doing your best effort to figure out what could go wrong can end up in either you solving those problems or changing things so that those problems go away or accepting those problems and living with them.

**Matt Coles (27:28)**: Yeah.

**Chris Romeo (27:29)**: I mean, why does the court— the court's not going to care about what the cost is at the end of the day.

**Izar Tarandach (27:33)**: The thing is that while we can't put in front of every single customer or every single security team out there, the things that we do for security for various reasons. The one thing that we can put in front of them is, "This is my threat model. These are the things that I took into consideration when I was designing the service or this product." Yeah.

**Chris Romeo (27:57)**: And just to kind of refocus the conversation on LastPass just for a second, like, regardless of lawsuits or government proclamations, I think the market's going to dictate what happens here. I think the customer base, like, I'd love to know, I'm sure these numbers aren't public, but I'd love to know how many people canceled their service as a result of this. Because this isn't a data breach, right? Data breach, like, we're so numb to that in this day and age. Like, oh wait, I got a new credit card in the mail. Oh, I must have had a bit of a data breach.

**Matt Coles (28:28)**: Woo!

**Chris Romeo (28:29)**: But this is different. Like, I feel like, I feel like the consumer is getting this information and should be, some percentage of them have got to be going, I'm not staying here anymore. Y'all crossed a boundary and I'm not coming back.

**Izar Tarandach (28:44)**: And that's exactly the utility aspect of it, right? So I know that my, I don't know if actually mine, but I know that many water facilities, they put fluoride in the water. There was some thought in there that it was a good way to go over the problem of cavities, right? So they do the proactive thing and went there. Busted analogy, but I want to know that the people who are defending my passwords actually took all possible measures to, if the thing gets in the clear, then not only the passwords, but all the metadata that goes with it is protected.

**Chris Romeo (29:22)**: Yeah. But in this case, in your analogy here, I don't have two water suppliers for my house. I can't choose Supplier A or Supplier B to provide water to my house. I have one option. With LastPass, I have a market. There's a market that's been made and I can take my— I can speak with my checkbook. I know people don't speak with their checkbook anymore. It's such an old analogy, but I can speak with my dollars and say, I'm going to move to a different utility provider because this one has failed me.

**Izar Tarandach (29:52)**: So, I heard of this guy that uses a password manager where the actual vault is stored in a separate cloud thing that they control.

**Chris Romeo (30:00)**: His name is Matt.

**Izar Tarandach (30:04)**: So the point is, yeah, I don't have another water facility, but you know what? If I put a barrel out there and start collecting water and whatnot, or if I drill by myself, I always have that option, right?

**Matt Coles (30:15)**: True. You could get away with it. You know your analogies all break down with electricity deregulation, right? You have electricity provider choices. You have cable provider choices.

**Izar Tarandach (30:23)**: Right.

**Matt Coles (30:23)**: So I'm just saying—

**Izar Tarandach (30:26)**: generating whatever I need, right? But the, the point is that extreme as they are, there are options. My problem with waiting for the market to, to give an answer to it is that we are giving everybody the first screw-up free. Doesn't matter what the size of the screw-up is.

**Matt Coles (30:49)**: I think there's two, two things to consider. And there's sort of divergent threads, so I want to be careful to put this. So on the notion of I want to understand, or as you called that, I want to understand that all concerns have been addressed or have been included in the threat model, as an example. There is a way to solve that, and that's been solved in medical and automotive systems for I'll say, I'll go on a limb and say decades with the environments and the regulated environments in which they do analysis, engineering analysis, where they do risk analysis. And they do risk analysis, here's a set of risks, they're documented, they document the threats, they document the harms, they document all the possible scenarios, and all that engineering work is done and documented, right, as part of an engineering process. At least that's what it's— that's what's supposed to be done, right? So for safety, especially for safety systems and critical infrastructure, and, you know, if you're building a nuclear power plant or a rocket ship or, you know, a city, you're going to be doing this hazard analysis, right, as part of your work. And there are risk assessment frameworks that exist that, you know, from NIST and other places that go through this level of effort, but it's a level of effort that has a definitive cost, both in terms of engineering time as well as time to market, which makes this a problem, which leads into the other aspect of the first breach is free, as you described. So there's the cost to market or time to market issue of going through that due diligence, which means you potentially miss things because if you don't make it to market, people aren't going to be using your stuff. But also, the choices that consumers make is based on, at best, incomplete information or uninformed information. I'm looking at the LastPass website right now. I don't use it. I'm not really familiar with the technology, but, you know, it's got some interesting, you know, information on their website, right? The strong encryption using AES-256, that sounds good. Local-only encryption, all my files are mine and they don't see it, and multifactor authentication. All that stuff that security people are talking about, that sounds good. How do I know I shouldn't choose LastPass, right? And until the breach happens, how do I know that they either misspoke, lied, or didn't understand what the implications were?

**Chris Romeo (33:34)**: And I think this problem is a lot bigger, right? Like LastPass is— we keep saying LastPass, But there are hundreds, if not thousands of companies in this world that are in this same boat. Like, you only have to go— Oh yeah, sometimes worse. We've all been, yeah, we've all been places. We've all worked lots of different places and you get on the inside of working somewhere and you're like, holy cow, how do we even, how do we even operate tomorrow?

**Izar Tarandach (33:58)**: How many of those are security utilities? I mean, some— How many of them are keeping a place where they provide you basic security, basic, Lots of things.

**Chris Romeo (34:07)**: No, I mean, I get your point. You keep kind of pushing us back towards security utility. I think you're defining a new category here, by the way. But I mean, security utility is a category. And given the fact that, as we said at the top of this conversation, we as security people have told the people that trust us to use these things, and the people that trust us have reflected that trust to the LastPasses of the world, because we said do it. And so, yeah, security utility is, it needs to be a thing going forward.

**Izar Tarandach (34:46)**: But again, it needs to be a thing. It is already a thing. They are all there.

**Chris Romeo (34:51)**: Yeah.

**Izar Tarandach (34:52)**: Okay. We have companies that say that they are going to help you monitor your CI/CD. You just have to give them all credentials to every single thing that you run in your CI/CD, but it's fine. They're going to take great care of it.

**Matt Coles (35:05)**: What could possibly go wrong, right?

**Izar Tarandach (35:07)**: Exactly. And that's exactly my point. What could possibly go wrong? The one thing that I can ask these people to actually tell me is, what's your threat model? What did you think could go wrong? And give me an idea of what you did about it.

**Chris Romeo (35:20)**: Yeah.

**Izar Tarandach (35:21)**: Because me, as your customer, I can perhaps add to that threat model.

**Chris Romeo (35:26)**: I can perhaps— Yeah, but how much of that, how much are you giving away though? Right? Like how much, I mean, even under NDA, how much of your innovation and how much of your trade secrets and whatnot are contained in that document? Like, if I'm doing something innovative, I would have to describe it some way to consider the threats of it.

**Izar Tarandach (35:49)**: So if you're doing something truly innovative, chances are you have a patent out there and that patent already describes the whole thing.

**Chris Romeo (35:57)**: True.

**Izar Tarandach (35:58)**: Now, if you don't have one, okay, we have— again, not a lawyer, but I heard that NDAs may be enforceable, may not be enforceable, don't know. Signed a bunch through my life, but there must be some way to guarantee that there is a privileged discussion between peers where some amount of detail can be discussed.

**Matt Coles (36:27)**: So I— can I— sorry, I just need to interrupt.

**Izar Tarandach (36:30)**: Just one more thing.

**Matt Coles (36:31)**: Yeah, yeah.

**Izar Tarandach (36:32)**: The problems that we are trying to solve security-wise, they are not exactly things requiring special IP. They are not requiring innovation. The things that we are trying to solve in terms of everyday security are well understood. We just want to make sure that they are in place and they're being used. And more than that, that all possible bad cases were considered. I think that the first step here that me as a customer can offer a company like that, that discusses their threat model with me is, wait, have you considered blah, blah, blah? And perhaps they haven't.

**Matt Coles (37:08)**: Have we shifted focus in this conversation? I think, well, I think we, so, Companies who are looking at security vendors, right, we're talking about security vendors of security tools or vendors of security capabilities. Companies who are looking to purchase or contract with those absolutely have the power. They have the power of the purse or the checkbook, so to speak, right, in terms of buying power, but they also have legal teams to create and enforce NDAs. LastPass is a consumer security capability, right, with— I would hesitate— I would hazard a guess of million-plus subscribers who don't have NDAs, right? And in fact, buying a product, you get a software license, you get a license that doesn't even give you access, it doesn't give you control over the software you're using. They're not going to get an NDA out of that. And so the discussion I think is completely different. We shouldn't lose sight of that, right? That there are ways that companies can gather input from their consumers, like through forums and support groups and whatnot. Of course, you'd have to get past the, "Oh, I don't want to talk to the vendor. I just want their stuff and make use of it, and I want it to be good and secure and reliable." They probably don't think about security, but they probably think about reliability at the very least, or at least cost-effective, right? And so we're not talking about the businesses buying capabilities, we're talking about consumers buying capabilities, and that's really where the big challenge comes in.

**Izar Tarandach (38:50)**: Hate as I must, there are instances where government has stepped in and advocated in the name of the customer. Right, we have the BBB, we have the, uh, all kinds of different organizations as well that while they're not, they're not governmental, they're quasi-governmental and people do take them seriously. Hell, even Consumer Reports.

**Matt Coles (39:13)**: FTC. FTC would be the government agency you're talking about.

**Chris Romeo (39:17)**: Yeah.

**Izar Tarandach (39:17)**: So there is a, a, a way of putting a big voice in front of the business and say, talk to me. Chris, you were saying something.

**Chris Romeo (39:26)**: Yeah, I mean, but when— so I'm gonna, I'm gonna refocus once again. Matt, Matt, brought us back to the consumer focus of this. We started to get kind of enterprisey. We went off on an enterprisey tangent for a second. And that's really not the challenge. The challenge right now is, like Matt said, LastPass has likely over a million customers. Some of them are like us, but most of them are not in that they don't understand what AES-256 is. And so, to solve the problem for them, like even if they had a copy of the threat model, They're probably not going to look at it. They're probably— some, some of them would have the knowledge and the abilities to understand it and do something with it, but a lot of people wouldn't even know what to do with it. And so brings me back around to, you know, we've talked a lot about problem. We've kicked the can a little bit on solutions, but like, do we want to go to an industry certification standard where everybody has to get in a line? And provide their documents and their test results and that, because we've done this before and it doesn't work. So I'm already, I'm kind of leading the witness here, like, we're not going there. So then what else do we have, though? Like, what else do we have for the consumer?

**Izar Tarandach (40:36)**: You know, I keep going back to this article that I read about LastPass, and I believe that the author is Jeremy M. Gosnick, and he outlined every single failure across the system. Steps in there, right? And my point is that if these things are made shareable, and by these things I mean the what could possibly go wrong part of the threat model, and we have these citizens around that do have the expertise to look at those things and them being so juicy a target, they have an incentive to look at these things as well.

**Matt Coles (41:13)**: Okay.

**Izar Tarandach (41:16)**: Then you do have a cadre of experts that are available to look at those things and say, in the name of the people, why didn't you encrypt this thing?

**Chris Romeo (41:28)**: But look at it like this from—

**Matt Coles (41:30)**: it's—

**Chris Romeo (41:30)**: you're making me think of open source, right? One of the arguments for open source is that everybody can see, meaning experts, everybody can see the code, but people that actually understand the code and can do a security review of the code can also look at it. So you're almost describing the open sourcing of the threat models of every cloud service that exists in the market.

**Izar Tarandach (41:55)**: And I'm not saying anything new. Adam has been talking about sharing threat models for a long, long, long time.

**Chris Romeo (42:02)**: Adam Shostak, by the way.

**Izar Tarandach (42:03)**: Yeah. And it's not new. Perhaps it's just one more incentive to talk about that in a different light.

**Matt Coles (42:12)**: I would, I'm not a lawyer, but I would expect that lawyers, by the way, we should put "I'm not a lawyer" into the phrase jar that— We should rename the podcast to "We are not lawyers." But I'm pretty sure that reverse engineering a commercial product with licensing, license agreements that specifically say don't reverse engineer, even if you're security professionals, is a legal liability concern.

**Izar Tarandach (42:43)**: Yeah, but they wouldn't have to because the information will be shared with them.

**Chris Romeo (42:47)**: Yeah, I was thinking more that the company would say, "I'm gonna provide some level of threat model to the world. I'm gonna just release my threat model in some forum maybe," or whether it's just gonna be available to anybody that wants to read it. And then people like us that geek out about reading threat models, would look at them and we could, and there would just be a mechanism for some level of review versus what we have today, which is nothing.

**Izar Tarandach (43:18)**: And look, again, I'm saying I'm not setting out to read their threat model and say if it's a good one or a bad one, or if they implemented the mitigations well or not, or if they have the right controls or not. But I think that even something minimal as, Let me get into your head for a second and understand the list of things that you thought might go wrong with this thing. Because it could be that from where I see it, I see other things that you didn't see. We do this kind of thing for features all the time. There are customer advisory boards where I bought your product, now you expect me to give you my time for free or for some swag and tell you what are better features, features that you're going to add to it. Why can't they do the same thing with the threat model?

**Matt Coles (44:04)**: But how does this solve the consumer problem of which tool to choose and why to choose them?

**Izar Tarandach (44:09)**: One thing at a time. This at least solves the problem of, uh, oops, we didn't think about that thing and now all your data is out there and it's unencrypted and it does have some value even if it's not your password. Okay, because clearly the case exists. It's out there. It's happening. And somebody didn't think of it or thought of it and said no big deal.

**Matt Coles (44:33)**: Well, there is a, there is a third option that, that technology evolves over time, and what they implemented has— what they implemented at one, at point A now becomes vulnerable at point B, and they haven't kept up to date.

**Izar Tarandach (44:46)**: So that I would agree with. For example, in the, the iterations that they, they add in their, uh, uh password training thing. But again, that same person, and again, I think it's Jeremy, but they said, you know what? Yeah, they upped the number of the lower limit of iterations, but they didn't let the customer say that, no, that they had to do something. You would log in and it says nothing about your master password. We want it bigger. Change it to something bigger. You still have vaults out there with people using master passwords with 8 characters.

**Chris Romeo (45:24)**: Yeah, but would you have caught that in, if we had the threat model, would we, we, we probably, we might not even, they might not even document that level of granularity for, for if they were gonna let us look at their threat model. And so that, that, that type of thing could be still be a, still be a giant problem.

**Izar Tarandach (45:43)**: It could, but it, it would give somebody the, the, the chance to say, you haven't documented this. What have you guys done about this case, about these things that can go wrong? Have you done something? And they'll say, yeah, you know what, here, sign on this NDA and we're going to talk more about it. Or they can go, ah, we didn't think about that, good thing that you raised it. Same thing with a bug report. You send a bug report on whatever kind of software, you get back a, hey, thank you, We figured it out, we're going to solve it. It's just a very, very, very early bug report.

**Matt Coles (46:21)**: I, I, this, I mean, the problem I have with this still is that again, let's assume they have a million subscribers. How many, how much do you expect their support teams or their vulnerability response teams, because this is security related, so it doesn't get flagged as such, does somebody post on Twitter, you know, uh, how much, how How many res— how much resources should it, should an org company, uh, allocate to this type of care and feeding to solve our questions?

**Chris Romeo (46:55)**: Not necessarily direct effort.

**Matt Coles (46:56)**: I mean, it's their problem. It's their problem.

**Chris Romeo (46:58)**: Yep. It's their problem right now. Like, you know, we're, when we're looking at this right now, they're wishing that we would've somehow done a review and sent them a, a list of things. Of course, it's always easier to see that after the dawn, right?

**Izar Tarandach (47:10)**: Yeah.

**Matt Coles (47:11)**: Yeah, so, and I think what you called out though is that it isn't just that they evolved the technology to deal with the threat. So they identified a threat, they actually understood a threat, they went and made a change, but then failed to communicate that appropriately to consumers who don't understand what the change means, right? So, action required, and, you know, a big pop-up or whatever, and So, there's a UX problem there, right? There's a customer training, there's a UX, there's all these other things that are not strictly security-related.

**Izar Tarandach (47:48)**: No, the UX is how the problem shows up. The security problem is in not recognizing that, yes, you walked to that limit, but you haven't finished doing what you did because you have to re-encrypt all that stuff.

**Chris Romeo (48:00)**: I'm going to come back around to this is a free market problem. I mean, we've spent a lot of time kind of thinking through this and whatnot. But when I think about like, what's the reality of this actually happening? I think it's a free market. I think the market is going to drive— if either LastPass is going to get their security house in order or they're going to have more security breaches, which is going to continue to erode confidence, and then people are going to finally walk away from them and say, "This is a product that I can't use." Like, I don't— I would love the things we're talking about to be part of the culture of software and software as a service and all these things, but I just don't— I don't see this as being a reality.

**Matt Coles (48:46)**: I think I can— I think I'll predict that somewhere there will be an open source project that does basically what LastPass does, that does go free open, you know, opens the door, you know, to the backend because of course it's open source code or, you know, maybe it's a service built on top of open open source code where they may be more free and capable of providing that information. And open source doesn't mean insecure, right? So there's an opportunity here for an open source organization to really fill that gap.

**Chris Romeo (49:24)**: Yeah.

**Matt Coles (49:25)**: Now the question is, would a collection of uninformed consumers take advantage of that? That's the free market part.

**Izar Tarandach (49:34)**: Just to make sure that there are already open source password managers out there that, that operate on the model like LastPass does with the cloud enabled and that, that give you the capacity to do at least what you guys listed in the beginning of the, the, the episode. But again, it, it goes to, okay, so for everything that becomes a security utility, we are going to wait until somebody wakes up in the morning and writes the open source version of it. So that we can feel secure?

**Chris Romeo (50:07)**: I mean, the only other alternative is what we've tried for, like I keep coming back to, right, is you write a standard, you have a way that somebody has to certify to a standard. They have to prove that there is a set of standards for password, for cloud-based password managers. And one of those standards is you have to be able to show that your, as your security moves forward, you bring all your other users with you. You don't leave anybody at different tiers or different levels of security. That's a big undertaking though, because somebody has to— and this is where the government is the only organization that can do this level of writing a standard and then enforcing this, the compliance with that standard and using that as a mark. You know, Underwriting Labs, once again, coming back around to them, If you look on your hairdryer, I mean, you guys probably have hairdryers and need to, you know, dry your hair at different times. There is a UL stamp on the bottom of that because the UL has tested that product that you won't grab onto it and it won't just spontaneously combust and burn your arm off. Again, again, again. But that, but that came at a price. And regulation stifles innovation. So, yes, I have a hairdryer. When's the last time I had an innovation in my hairdryer? I don't know. Not in my— I can't think of one in my lifetime. I mean, there was cool and hot when I was a kid. But there's also a safety standard there that it won't burn my arm off. Like, I am 100% sure that if I go turn the hairdryer on upstairs, it's not going to burn my arm off because UL has a set of standards for electrical stuff. And I don't even know what goes on. I just know that I can trust that mark. We don't have that in security.

**Matt Coles (51:57)**: Well, so, oh, go ahead, I'm sorry. Yeah.

**Izar Tarandach (52:00)**: Speaking of standards, wouldn't it be cool if we had a standard for changing passwords so that people who have 400+ passwords in the password manager and all they have to move away from that thing because things happen, they have a way to script the way that they change the passwords and don't have to go to 400 sites and do it.

**Chris Romeo (52:16)**: Damn it. That's innovation. That's free market innovation right there. Somebody build it.

**Matt Coles (52:20)**: Yeah. If only somebody had a centralized credential manager or a centralized identity service, right? Everyone needs to use Active Directory. No, I'm kidding, and we shouldn't take names here, but anyway. So speaking of standards, actually, and UL, I just was looking and there's an ISO/IEC standard 18013-5, a standard for electronic credentials. And so this is something that the UL is, I guess, has been pushing as well. So I don't know what the timeline is, how far along it is. It doesn't have a date on it. So, but it may be something along that lines. The right standards for the right things don't stifle innovation. They support and create a framework for being innovative, right? You set a common baseline, you get a certain set of expectations and a certain set of understandings out of the way, and you get consistency in offerings in the ways where you need them to be, like not blowing up your house when you turn on your hair dryer. But then innovation can come from other places like look and feel and cost and other features.

**Chris Romeo (53:36)**: Right?

**Izar Tarandach (53:37)**: So—

**Chris Romeo (53:37)**: I'm gonna disagree with you though. In that standards become a race to the bottom, right? You get a standard and then you have, "This is what I have to comply with." And then I comply with that and then markets tend to fall like hair dryers. There's not a lot of— like you go to Amazon, there's some price deviation. It's not a lot because there's just not a lot of innovation. And so, I believe and will argue strongly that that standards will limit innovation.

**Matt Coles (54:11)**: Yeah, we can have that debate at another time. I'll just— we can talk about HTTP if you want and what standards has done for us there or not done for us there. So, but anyway, yeah, perhaps, but I think the challenges of innovation that you're talking about there come in other ways, you know, manufacturing locations and and cost of goods and other things that we're not really talking about today, right? We're talking about— Yeah.

**Chris Romeo (54:38)**: I mean, you think about like, where did LastPass come from? LastPass was a startup at one point who wouldn't have been able to comply with a standard. And if they had stopped and started complying with a race-to-the-bottom standard, then there— I'm sure there are some— I don't use LastPass, so I don't know all— I'm sure they have some cool functionality, though, so that we as security people would look at and go, "Oh, that's kind of cool." But if they had that standard, they would have just— they would have— they would have focused on the standard as their first. If that becomes the barrier to selling to the market, that's the key. If you have to have the standard compliance, then as an innovative startup, I can't come up with the coolest way to do it, but I don't meet the standard. Oh, well, we can't buy it because it doesn't have the stamp on the bottom that says it won't burn my house down. Yeah, but we can— we can digress. We're getting off on that. I think we got to wrap this. We got— we got to bring this thing to a conclusion now. Around the security table. And so, Izar, give us your kind of final thoughts on this.

**Izar Tarandach (55:38)**: Oh God, everything's a mess. Everything is burning. Let's close and start from the beginning again. I wish I had something more, something smarter to say, but Right now it's just looking at the problem and imagining smaller pieces of solution that we can piece together and bring the situation to a better place. The first thing that I would really, really, really like is for developers, and again, I'm dropping this on the developer, for them to understand that they do have a security responsibility and that they have to do something about that. And for security people on both sides of the equation, both those providing services and those acquiring services, to understand that they have to keep asking what could go wrong and what have you done about it. Mm-hmm. And educating the customer, even with the knowledge that when we educate people, sometimes that thing can go and bounce back on us. So yes, you know what, we, we're not infallible. And people who write and design software are not infallible, and things will continue happening. And we can only strive to, to lower the, the chances of those things and the impact of those things happening again and again and again, to the point where we become numb to it, as Chris said.

**Chris Romeo (57:07)**: Matt, what are your final thoughts?

**Matt Coles (57:09)**: Uh, I, I think I have two, two final thoughts. Uh, one around the consumer. Uh, keep Keep in mind the thought process we talked about at the very beginning of the episode about how Chris and I chose password managers.

**Izar Tarandach (57:22)**: Yep.

**Matt Coles (57:23)**: All the fancy bells and whistles are great. Just understand that that could cause— that it opens you up to more than maybe you want to take on. Right. And obviously it's hindsight. It's much easier in hindsight to recognize that maybe it wasn't such a great idea, or maybe not all the thought process went into it, as César mentioned. So just keep in mind that we tend to be, as security people, we tend to be technology people, and we like the fancy technology, but as security people, we'd like to be both pragmatic and restrictive and conservative in what we do, what we actually take on, because that reduces our exposure. For those who may have used LastPass or any of the other password managers that have been affected by breaches and compromises, compromised over the years. The public service announcement: go to haveibeenpwned.com and take a look and see if any of your passwords have been compromised.

**Izar Tarandach (58:25)**: And subscribe to the service.

**Matt Coles (58:27)**: And if they have been, change them.

**Chris Romeo (58:30)**: Yep, definitely. That's good advice. And I guess my final thought is, yes, things are a mess, things are a disaster across everything that's built in software today. We know it in hardware, right? Like, because we've all been in different places and seeing different things. But that doesn't mean we stop. It doesn't mean we give up. That means we just have more work to do. And we all know here that we're going to retire from this industry and the problems are not going to be solved. They're still going to be— there'll be new problems. And, and we'll retire and take up golf and shuffleboard and wear black socks with sandals and all of those different things along the way, but the problems will still be there. But You know what? We don't give up hope. Like, people should still have hope that we can make this, make these things better. We can improve security. It's just, it's not something that happens overnight. It's not something that happens with the snap of fingers. And with that, thanks for listening to this episode of The Security Table and tune in soon for another spirited debate.
