What risk am I doing or am I performing when I ask an agent that is open AI or chat GPT, how vulnerable does that make me using that very tool that's supposed to make me more efficient?

Well, I mean, it definitely increases vulnerability, especially if you're not using what we will call company-sanctioned AI.

So AI where we have it in a controlled environment, like you could have that with Copilot or ChatGPT or many others, Claude.

But what we see is that people don't necessarily distinguish between what is a controlled environment and what they prefer to use at home.

So company documentation or even classified documents that we receive from government or very sensitive information around airport security, we're very concerned about that.

We don't want people putting that type of information into an AI because as soon as you do, that information, you've exposed it.

It can be used.

What role does that play? James, let's kind of work our way down the panel here about how humans play an important role in getting back to the people who design the defenses that keep us safe.