Transcript

So you're working with a lot of companies right now to kind of help guide them on how they're building and deploying AI, right? What are you seeing? What are some of the patterns and anti patterns around security for AI build and deployment all over the place? OK. I would say in driven by sector, by size, there's some commonalities. I would say 1 is. Very few have have. Really know this that well, there's still learning. Everyone's still kind of learning. Yeah, there's that component to it. The fact that changes so quickly is a challenge. I mean, identifying and implementing use cases where it makes sense for AI is a little less straightforward than one would consider. And just kind of the whole, you know, how does it make our business more competitive is as we discussed at RSA this past year, one of the things that came out is this CEO fear of missing out where there's a, I like to call it downward pressure. From CEOs or boards of directors, from, you know, a VPS from business unit owners to go fast, fast, fast, fast. But as we discussed in our panel. Like there is little direction on how or wherefore or the. Yeah. So, so like go fast. OK, sure. So there's this kind of gap you see, or vacuuming, if we're going to call it, of how, how, how do you go fast? Yeah. And how do you go fast in a way that doesn't imperil the organizations? What kinds of things do companies are they getting right about security? Like what is it that's helping them deploy this in a secure risk managed way? And then maybe 1. Saying that they're not doing right okay doing right most health policies now so at least they'll be 1 dot O policy yeah they might have a governance council yeah or some kind of implementation group. They all look similar where they have couple folks that get it and are far ahead two or three people that think they get it but don't understand it and mix up the AI metaphors and then probably a handful of people that openly admit they don't understand it and so that's that's a pattern that we see across so. Policies, you know some implementation, castle copilots implementation or some type of proof of concept all over yeah, the badness is treating all AI the same. OK and again back referencing our panel discussion of RSA, this division between consumer grade AI and business grade AI or like I like to call it consequential the consequential nature like it are you. Implementing AI in a system that could break the world or is it Grammarly with machine learning, right like that, right, you know, so so one size might say I approach and like we like to say, and we're hoping gets implemented if AI is used for systems that can. Lead to airplanes falling out of the sky to grids locking U then a lot more mindfulness and deliberate security review and thoughtfulness should go in that design not you know so again consumer grade and and business grade or even criticality whatever you call it yeah that that's the mistake is 1 size fits all.