Security in the Age of AI: Vanta CEO on Compliance and Risk
Channel: Alex Kantrowitz
Published at: 2025-04-10
YouTube video id: uwpRnpEG-a4
Source: https://www.youtube.com/watch?v=uwpRnpEG-a4
Let's talk about how AI and automation are creating a new generation of security risks and how the technology itself might prove useful in fighting these new threats. We're joined today by Christina Cassiopo, the CEO and co-founder of Vanta to dive into these issues in an exclusive interview presented by Vanta. Christina, great to see you. Welcome to the show. Thanks so much for having me. So, I want to ask you about the security risks that we're seeing from AI and automation. Now, so much of the conversation around these technologies is all about, you know, how is it going to progress? Is it hitting a wall? Uh, but I, as far as I understand, the current technology today is actually pretty powerful in and of itself and is already upending a lot of the ways that the internet works today. So, what do you see the role of these technologies uh in the online experience? Uh, and then what are the risks? Yeah, I think there's a ton of excitement about these technologies. is I mean some of the way I think about it is they're letting builders create these magical experiences that people have thought of before but have never been able to do. So specific example in Vant's context is uh security questionnaires are just a part of selling software to enterprises. Basically you get this long spreadsheet of questions you have to answer. Um everyone has always wanted to build a product that just answers the questions correctly the first time. that wasn't possible prior to some of these foundation models, but now it is. And so just an example of of kind of the magical product experiences that can be built now, but couldn't even 3 5 years ago. Okay. So you have your automation tools, being able to fill out forms, being able to basically take action on your behalf, and I guess being grounded in really reliable information. Uh so that's the plus. Um, but then I imagine you also have an army of uh of bots that are going out and maybe doing this and for nefarious purposes or even um if we're going to have this increase in bot traffic on the on the web, uh it seems like it's going to make things more difficult for just like normal companies going about their day-to-day. So, um we talk about the opportunities, but what do you see as the risks? Yeah, there there definitely are risks. Um, the one stat I found interesting is over half of the Fortune 500 companies in their most recent uh, annual financial reports cited AI as a risk factor. Um, and so, right, it's it's kind of prevalent. I think I think about it in two ways. One, just general technology deployment. AI is a new technology. We're still figuring out its parameters. Anytime there's something new, there's a ton of excitement, but there's also just a ton of risk as we figure it out. So, there's something just general there. You know, there's a story of mobile 10 years ago, maybe cloud 15 years ago. It's AI now. Um, and then there's a specific piece of these uh tools can create incredibly accurate text and video that looks like it's real or you know, but might not be. And so there's a whole host of attacks that are now again much easier. It's sort of a different magical product experience that we might all think is actually less magical or the result is. Um, but those have also become much easier as well. Yeah. So what are these new attacks like? Yeah. So I mean we see some of these at Vanta too. There's a lot of impersonation. Um it is just easier to you know I can think the old way was the joke at Vanta was if you joined the company you would get three text messages from me asking me to buy you gift cards but things were probably like somewhat misspelled and you probably wondered you know does the CEO really need me to go out and buy gift cards? Again it was sort of this you know attack that was easy to laugh about. Um, that one still happens, but there's versions of that that are much more compelling. So, CrowdStrike has recently talked about how they had one of their customers, um, I think it was the CEO, but an executive at one of their customers actually impersonated in this credible way and, you know, go ask employees that I think over video to go do something for them. And, um, so again, it's kind of you can see some of these same attacks, but just way more compelling. Yeah, it's crazy that that's actually happening. I mean, the gift card attack, you know, we all know about that. We generally know probably not a good idea to go out and get gift cards and not hand them to your CEO. If you do it, hand them. Don't send them free. Uh but but the fact that we're now seeing video, uh you know, impersonation, pretty convincing impersonation is is wild. And I Yeah, I guess to be able to do video, you need really good audio. And that stuff is is wild what we're we're seeing today. Yeah, I think you're totally right about the, you know, audio quality getting quite good, the video quality quite good. And I think maybe just taking a step back, one of the things that, um, is so exciting about this period is like that technology has gotten so good is opening up lots of new use cases and products. So again, one of those is maybe fishing somebody else's employees and trying to get their information, which probably not a good thing for society or markets or really anyone. But another one is we're seeing companies build like wholly interactive experiences that again help them deliver their products and services even better. Um and do things you know for good in the world rather than again try to try to scam people out of their money. Um but but these technologies can be kind of used for both ways. I feel like this is a lesson we've really learned over the past 10 years of social media is there's a bunch of great things that can come out of new technologies. There's also some things that are less great or um less expected. Um, and I think we're really seeing that sort of parable play out in the age of AI as well. Yeah. And you know, it's so interesting that you bring up social media. I covered social media for a long time and no one would ever suggest that like the ways to handle social media vulnerabilities was just more social media or you know different social media. I mean maybe people people tried alternative social media platforms but they were never cred credible alternatives or credible solutions to some of the problems that we saw on social media. almost like needed the platforms to reform themselves or to change uh to be able to address those those issues. But with AI, uh, very different because this technology that's available to all the people that we're talking about using it for bad uses, uh, is also available to people who are trying to head off security concerns. Um, and also available basically for people who are trying to be good citizens on the internet and I think that is exactly where Vanta sits is being able to help them uh, use these products and would be great if you could tell our viewers a little bit about how you do it. Yeah. So, Vanta was started under the dual mandate of helping companies um initially build out their security program and over time understand it and improve it and then uh that's all a bunch of hard work, right? And get credit for that hard work of building their security program uh with the market, with their customers, with executives, with their board. And so, part and parcel, we've always tied the security work we help companies do to again customer outcomes. And so the first place we started doing this were compliance reports. So uh we help companies get sock 2, ISO 2701, GDPR, HIPPA, um tons of standards like these. We help with the security questionnaires. We help security teams demonstrate all of that hard work they're doing. Um that helps customers build trust and ultimately grows their company's revenue. And now are you using artificial intelligence tools to be able to you know put together these reports or to be able to uh address some of the concerns that we were discussing earlier on. We are in a in a bunch of different ways. Um we started building AI in from the ground up across our product. So everything from helping a company decide uh which security controls are best for them given their maturity, given what their customers are asking for. um when there's a you know error vulnerability how do you fix that uh AI is helpful for giving a a first pass or a suggestion for the team to take action on and then um a lot of the process of these audits or requirements is just a lot of documentation back to the security questionnaires I talked about um at the beginning uh there's you know you have to do the work and then you have to tell people about how you did the work uh when I talk to security professionals they usually really like doing the work that's why they got into the job They generally don't like having to tell people about the work in a bunch of different forms. So, you know, an audit, a screenshot, a security questionnaire, and a conversation. And so, one of the places we've seen AI be really effective is again to take all that work that was done, summarize it, and put it in the right format. Um, so the teams can focus on actually doing the work. Yeah. And can you talk about how that happens? So is it basically a generative AI model that's monitoring uh the people as they work and then being able to sort of translate that activity into natural language. Talk a little bit about how that how they operate that way. So it is definitely generative AI models um kind of the foundation models plus some post- training um and a ton of kind of quality improvements to make sure uh there's there's high accuracy and security compliance is a place where I think there is less tolerance for creativity and true generation and you know more more desire for accuracy. But basically take these models and not observe uh what people do but observe the outputs. Um so what have they done like what was there before what did they do look at the output rather than kind of the work itself but then take those outputs and say okay maybe you went and I don't know changed a bunch of configurations in your cloud infrastructure um in a way that's more secure. Okay we can take that new configuration uh you can turn that into a policy document. So there's a standard kind of document written in sort of legal ease or compliance that describes what there is. Take that configuration um translate it to answers for the security questionnaire. So the next time you're asked about your cloud infrastructure configuration, you got the answer there direct from from you know the reality um of what's in the system live. Uh and then finally take some of again take that same information map it to a bunch of compliance frameworks. And so if you're going through an audit or going for a particular um certification, uh all that information is there um up to date. Uh and again, no one had to kind of keep it up to date or remember, okay, after I went and changed all those configurations, now I need to go update the documentation. Documentation is just there for you. Okay. Now I'm very curious how you handle hallucinations because uh I think that's it's it seems like AI is getting better at not hallucinating but as we see even like the better models come out there are still hallucination uh errors that come in and um there was this great uh story by Ben Dictanss this analyst about how he wants to use deep research open AI's research tool but if it doesn't get him to 100% accuracy it's sort of less useful for him as a researcher now for you and your your clients uh if you have something that hallucinates security questionnaire answers, I would imagine that's kind of be going to be um you know, sort of put the entire reason for that product's existence um on the rocks. So tell me a little bit about how your team handles hallucinations and if there's a way that you've been able to get around them. I think we're so not a not a um I don't think we're doing anything that like other leading AI companies aren't doing honestly, but a couple of things. a bunch of post- training, a bunch of prompting around like, you know, just do not hallucinate. There's parameters inside some of these models to set where you can sort of tell it like how creative do you want it to be and we generally say please don't be creative at all. Um, and then we just keep humans in the loop uh uh on both sides actually. there's Vanta team on the post training um uh we have this golden data set and so you're constantly kind of checking and making sure that from the data we see not customer data but data we generate how are we doing on that and it's a metric we keep track of and then also our customers in the loop so um sometimes I describe this and people think oh you just fill out the questionnaire and then send it back to the customer and no one sees it and that would be a magical experience if these things worked but given the domain we're in uh that is not a magical experience that's a very scary experience. And so to that, you know, the way the way it works is we'll take a we'll try to answer all the questions. We put it in front of the end user and say, "Hey, what do you think of these um and actually then look at that score and uh you know they can go change things, they can go send it, they can go about their day and then our system gets smarter because um we're learning from the places where you know our end user accepted an answer versus rejected one." Um, so humans in the loop in a couple places given the domain we operate in. Well, you said you just said something that I've never heard before and I'm pretty fascinated by it. You can tell the machine not to hallucinate and it kind of knows what you're saying like be less creative and it says okay, I'll stick to for instance the spreadsheets. That's really that's effective. It is reasonably effect. It's not 100% fail safe, but it's like it's pretty good again. And it's um and again, it's it's I think sort of a nice customization that some of the foundation models have. Um I think we'll see more stuff like this over time. Honestly, I'm just I'm just guessing, but as a product person, right, because it helps them serve a a variety of use cases. Like if you you know, again, if you're in security or compliance or legal, you like don't want all that much creativity about the facts. Um versus if you're like a design tool, you actually want a ton of creativity. And so having a dot aisle um I think just makes these models uh work better for more use cases. That's pretty cool. That's why we love speaking with people who are actually building the stuff on the show and not just the people philosophizing about it because you're they're seeing how this works in action and it's great to be able to bring that uh to viewers. So you you also mentioned that it can help people sort of the technology can help people outline or or build out a plan their security plan. Uh can you talk a little bit about how that works? Yeah. So basically I think it's this again it's this experience that we can now build that actually I wanted to build in 2017 2018 when Vanto was just getting started but it was it was a bit too hard but basically I think the experience you want is you have some combination of like hey here are the best practices for a company of this size. Maybe we're a 2,000 person company maybe we're a 20 person company but like here's some general guidance on best practices. We used to call this like don't be a doofus right? You just want you want to be doing like generally the right stuff. Yep. And then you want that matched with like specifics about your company. So again, maybe you're 20 people, not 2,000. Maybe you're fully remote. Maybe you have, I don't know, all your people in the, you know, spread across the world and there's no offices. So you don't want to be asked about key carding into the office, whatever it is. But you kind of want to combine, right, best practices and company specific things. And that is what the best security and compliance professionals do. they kind of know, they know the ground rules, they know the regulations, best practices, and then they adapt them to to the specific company. Um, and I think these models are kind of a really interesting way to do that at scale. Early Vanta, we built in some rules and heruristics, but they weren't they weren't totally what you wanted. Better than nothing, but it wasn't like a 10 out of 10 experience. Um, you know, again, and if you have an individual person on the call doing that setup and scoping, you get a great experience, but you have to do the call. you need the people, you have to do a lot of these calls. And so these models kind of let you again they understand the best practices. You can get them to understand the regulations and then they can ask you some questions about um your company or you can tell them about your company and they can kind of merge all of that into a set of okay here's what you should do for your company given stage maturity and your goals. Um and we're we're now able to build that again in a way I guess six seven years ago we weren't. Yeah. Yeah, it is amazing what all the new technology is enabling uh companies to do. Yes, I guess there's the one side of it which is that it's kind of a gift for bad actors. But if you're trying to do things the right way, uh now it does seem like there's a real chance that you can actually do it much more uh easier and more effectively by relying on these models, which is which is great. Now, you're working in a space that's always changing. Uh and of course the AI world is always changing no matter where you are but you're also working with companies to help them remain compliant with let's say AI regulations and that is something that is just moving at an extremely fast pace uh in places like the US where we know that AI is being talked about within the White House and especially Europe where there's this new uh EU AI act. So I want to hear your perspective on the impact of regulations like the EU AI act, what it's doing for your clients and then sort of how you have to adapt as well. Yeah, I think we're in because this is our business. I think we are paying much closer attention to a lot of these than many of our customers. It's not to say our customers don't care, they definitely care. It's just um they they have an they have a separate business, right? And they want to do, you know, do the right thing per these regulations. But right now, a lot of these regulations are still a bit up in the air, right? Nothing has entirely landed. Nothing has been implemented. There's uh very few kind of implementation guidelines. Um, you know, the US's position on the issue has changed with the new administration. And so, I think a lot of our customers are in a like, okay, we're sort of paying attention, but we're really waiting for clarity, and then when it is clear what we need to do, given the markets we're operating, we will absolutely do it. For us at Vanta, we're paying closer attention. And I think honestly have a bit of the almost luxury of of being able to because it's our business. Um and I think what we're seeing is again people still trying to work lawmakers uh policy makers still trying to work through some of the implications here. I think everyone is you know has a high level goal of we want to protect data we want to kind of do the right thing and we don't want to stifle innovation and and needing to balance those two sides of the scale. Um what else do I think? I think um it it does seem like Europe will take a stronger stance here than the US especially especially within the new administration. Um I think we've all learned from the implementation of the GDPR over the past seven years or so, right? And seen um I think one of the primary critiques there is that the regulation is so open-ended that it's actually quite hard to figure out what you need to do. Um and again I think that probably came from good intent. it came from a desire for flexibility, for not stifling innovation. But where it sort of ended up is there's just a lot of uncertainty and it's hard to move through that. Um, and while you know the European courts will clarify that over time, it's just a lot of risk to ask companies to take on. So when we think about these new AI regulations, I think we'll see probably a little bit more guidance more than the GDPR in ways, but again still trying not to be so prescriptive as to, you know, kind of be out of date um as soon as the the ink on the paper is dry proverbally. Yeah. What could be coming with the new AI regulations? Any sense as to what they might touch? O um if you force me to bet uh I think a lot of it will focus on data security and so it'll be kind of under the you know banner of AI but it'll probably be a lot of the things we've talked about with data security and I think with that you know think through the best practices of understanding like as a a company or a service provider like what data you're collecting why where it goes when you send it to a third party. Um, one analogy a security professional gave me uh years ago was you should think of data um as sort of like toxic waste. Uh, you have it, you might need it, but you want to keep it contained and if it starts leaking out, it's very hard to um kind of clean that up. Uh, and again, you know, someone said me said this to me years ago, but I I think it's true. So I think you're going to see these regulations try to like fence in that data much much more so than um again we were doing 10 years ago as software builders. Yeah, it is interesting because with AI you go from sort of more deterministic software where you run a a a query and it hits one spreadsheet and then gives you an output. And with large language models it's probabilistic. So you have this whole wealth of data that you're touching and then you're trying to spit out an output, but you're never quite sure what it's going to go into and what it's going to share, even if you have rules. And I imagine that's going to um produce some very interesting headaches for people. You're nodding your head. I'm curious what you think. Yeah. Yeah. I also think it's like it's just a different way of building software. One thing we've noticed internally is again when it's deterministic and you have a system and it generally runs and you do the monitoring and observability but like your system's kind of your system unless I know something like an input to it changes. AI it's just not true. You cannot change the code. You can get very different answers. Um and so it's almost a different uh mentality on building. Again when it's deterministic you ship it and you can kind of move on. Again you want to keep an eye on it but like it'll keep doing whatever it it did when you first shipped it. AI systems and they will do different things at different points no matter whether or not you changed anything about them. So you have to keep you know much closer track of of kind of the quality. It's just different. Um it's a different kind of way of thinking about software development and maintainability and all of that. Definitely. Can you bring us into like the head of one of your client like of a client? You don't have to name the client's name but how do they prepare like how do they deal with all this stuff? Like what I mean they have to watch out for this regulation. They're already probably doing GDPR regulation, a bunch of other compliance that um that we've touched on at this point. Is it just like constantly monitoring for what might comes next or like what does what what is it like in in um the life of of your clients trying to keep up with this stuff? Yeah, it's really hard. I mean, our customers range from like CISOs of you know, large global organizations to like a first security or compliance hire. So someone who's like trying to do it all at a you know small to mediumsized company to a founder who's just getting started and you know each of those folks kind of thinks about or like does different things with their day but I think all of them are just trying to figure out like what's going on what do we have to do how do we keep track of all this work and how do we not let it slow the company down or our customers and revenue down and so I think over time you know people try to find ways to unify all these concerns. There's a lot of um uh sort of building or using off-the-shelf frameworks for all the things you have to keep track of and then that at least gives like a structured way to think about it all. Um I think the folks who have you know kind of larger company experience tend to do that. I think founders tend tend to be more um urgency based. I say this with a smile as as a founder myself. Um um but I think it's kind of this dual mandate of like what's going to happen next or what could happen, how bad is that? And then how do I again not slow the company down and hopefully even you know support the growth versus be be the no person or be the um well you can't do this yet person, right? And in the broader tech industry there's a debate about the wisdom of all these regulations. you know we have uh like you mentioned already GDPR which is something that people debate about um there's so much there's a lot of security uh compliance and regulations which I think those are uh yeah probably more universally acclaimed and now we might have AI regulation that's coming uh are you know you see the the impact of these regulations firsthand and as you mentioned like the security professionals they're trying to work within their organizations to, you know, make sure that it doesn't slow companies down, uh, trying to comply with them. So, just on a whole, do you think that all these tech regulations are are worthwhile? I think so. There's a, if you're, uh, if you're familiar, there's this great web comic called XKCD. It's like 15 20 years old. Stick figures. Okay. Yeah. You know, XKCD. So, of course, there's an XKCD about standards. Um, where basically, you know, you there's a you need a standard for something. The old ones don't work. So the answer is in fact to like make an 18th standard. It is never to go fix the other ones. Um in like true XKCD faction, I think this is often this is in fact how the world works whether or not it's how the world should work. Um people are constantly coming out with new standards. There's always a reason for them, right? There is a real problem they're trying to address, but it does just make the proliferation really hard. Um, I think some of the more effective efforts in the past couple of years have really focused on trying to create like an umbrella over multiple standards or at least a bunch of mapping. So, it's like, okay, fine, you're doing the 18th, but here's how it relates to, you know, 10 of the other 17. Um, so it's not like you're asking for 100% net new work because that's just that's just really tough for everyone. Definitely. Okay. I don't want to leave here without hearing a little bit more about the uh Vanta founding story and you know where you think the tech business is going today. Uh you co-ounded it. You have a very interesting story and you're right in the thick of things with a growing company um that's working with other growing companies. So why don't we end with that part? For sure. Um so started Vanta uh several years ago. The the joke that's not a joke about Vanta is I think if you want to start a security company you should start a compliance company. Um by that I mean uh but the kind of the founding thesis was security on the internet is increasingly important. Uh we'll be more conscious of it in the future not less. Uh it's also something that has real world implications even though internet security is again just just kind of online. Um but uh one of the reasons I think getting you know getting companies to have better security quote unquote is hard is because that's can sound like a nice to have or it's a nice to have until it's too late. Um, I think compliance in its best form is unlocks new markets. It unlocks new revenue when you get new certifications. You know, you can sell to larger customers, you can sell to Europeans, you can sell to healthcare. Um, and so it was sort of a way to say, hey, security team, you're doing all this really good work. How do we tie that to direct business value? So, you get credit for that, you get more resourcing, you can do more of that good work. Um and so uh that that was kind of like the thesis behind the company was again compliance uh it's kind of misunderstood but it's this great way to tie the security team's work to real revenue and real business impact. Um it was a nice idea on you know a piece of paper. It turns out it's like working quite well for us before um or now. Um so company today uh we announced like 10,000 customers around the world um uh and uh have been growing revenue at at quite a good rate. Nice. And if people want to learn more about working with Vanta and Vanta itself, where do they go? Vanta.com. We are always on the always open for business on the internet. Easy enough. vanta.com. Christina, great to speak with you. Thank you again for teaching us about what it's like putting this technology into play in a practical way and also some of the things we should all be paying attention to in terms of the downside risk and how to mitigate that. So, great having you on the show. Thanks again for being here. Thanks so much for having me. All right, everybody. Thank you for watching. We'll be back again with another interview soon and we will see you