How to address AI's biggest threats

Cisco has a plan to safeguard your company in the age of AI.

Transcript

00:00One of the things that comes to my mind, of course,

00:03is like the movies that we've seen in Hollywood about AI

00:06and how AI could take over.

00:09So what are the guardrails that are put in place

00:12to avoid a kind of doomsday situation

00:15where companies integrate some kind of AI

00:18in order to protect themselves against AI

00:20and then they lose control?

00:24So let me take a step back

00:26because what's happening right now

00:27is the composition of our workforce

00:30is gonna change quite a bit.

00:31So today, 100% of our workforce is humans.

00:34Tomorrow, you're gonna have augmenting of that workforce

00:38with AI agents, you might have robots,

00:41you might have humanoids,

00:42and we need to make sure

00:43that these different AI augmentations

00:47can actually work the way that we want them to work.

00:50So what we would do is if you think about a model,

00:53before a model goes to production

00:56for a specific application,

00:57let's say it's a loan processing application,

01:00we wanna make sure that that model is behaving

01:02exactly the way that you want it to behave.

01:04So we have an algorithmic way of going out

01:07and doing a level of validation on the model

01:11to make sure, and typically for an organization,

01:14it takes seven to 10 weeks, Conway,

01:16to go out and validate a model.

01:18For us, with AI defense, you can now do it within 30 seconds

01:22and so that level of compression of time

01:25and not having to worry about the details

01:27makes a huge difference in not just the velocity

01:29but also the safety and security,

01:31where you can enforce guardrails on this thing

01:34that if there's a model that's behaving in a different way,

01:36you can actually provide a compensating control on that

01:39so it doesn't behave the way that it should behave.