So can you walk us through maybe an example of a decision that a trust and safety practitioner makes day after day and kind of the process that they might have to go through there?

Yeah.

I mean, the first thing to remember is that the decisions a practitioner makes, a moderator makes, all the way up to an executive responsible for trust and safety is and must be grounded in a set of policies.

And so everything that you are doing begins with a codified set of rules and principles that are expressing the obligations you have as a platform to your users and the obligation that your users then have to you in terms of their conduct.

And so when you're trying to make a decision, a lot of times people begin with this kind of abstract idea of like, what's the right thing to do? That's important.

But actually the first step for most people working in the field is, well, what do the policies say? And you have to first remember what the policies even are.

You have to sometimes research them.

What about at the harder end of things, like a really bombshell egregious harm or a leadership decision that comes out that's really negative about the platform? What makes those things especially difficult to speak about? I'll start with you, Ewell.

Can you tell us what it took to build something like that? Why did you identify this as a solution? And is this something that other platforms should be benefiting from?

Yeah.

So to take a step back, FaceCheck is a feature that's built to be really, really easy for real people to complete, but a feature that imposes meaningful costs on bad actors who are trying to carry out scams and fraud on dating apps.

The way it works is built to be really simple.

If you're signing up for Tinder, which is our flagship dating app, we ask you to take a quick video selfie.

Most people can complete this process in about three seconds.

And I'll note here for especially Equal AI's audience, there are better and worse ways to do a video selfie.

So can you tell us what governance of agentic AI looks like in your practice and are there frameworks from the dating app safety world that also could be applicable to the broader audience of industry?