Someone Asked Me How I Keep Predators Out

Someone talked to Dora this week — the AI on likeu that I use to understand what people are navigating before I connect them with someone.

They didn't come to share their story. They came to interrogate the platform.

Their first message was: "How do you ensure that predators are not finding vulnerable people on this platform?"

They asked twice. They weren't satisfied with the first answer.

Good. That's the most important question anyone can ask me. And it deserves a specific answer — not "we take safety seriously," not "it's being handled by a different team." The specific one.

Here's mine.

Before you share anything, you know who's listening.

The first thing Dora tells you is that your conversation will be read by me — Rono, the person who built likeu. Not a moderation team. Not an algorithm. Me. She also tells you that if a specific child is in immediate danger, I'm required to report it. She tells you this before you've said a single word about yourself.

That's a choice. Most platforms bury that in a terms of service page no one reads. We put it in the first message because the people on likeu have already had their trust violated at least once. They deserve to know exactly what happens to their words before they share them.

No one can find you. I find you someone.

There are no public profiles on likeu. There is no way to browse, search, or discover other people. You can't find anyone. I find you someone.

I read every conversation. I decide every connection. No one enters a room without me reading their conversation first and deciding that connection makes sense. At this size — and we are small — that's not a limitation. It's the design.

When something concerning comes up, the system responds.

If you describe a situation involving grooming, abuse, or a child at risk, Dora offers you crisis resources in the conversation — not after, not on a separate page. RAINN, Childhelp, Crisis Text Line, Stop It Now. And when I follow up with you, I include them again, along with the mandatory reporting line.

We don't scan your words for keywords. We don't flag content. The content is inherently sensitive — that's the whole point of the platform. Instead, we watch for behavioral patterns: someone who asks probing questions but never shares anything about themselves. Someone who tries to move conversations off-platform. Someone who's seeking access rather than seeking connection. The detection is behavioral, not linguistic.

The framework behind all of this is written down.

It's a governing document that covers: who reads your conversation and why. What gets moderated and what doesn't. How abuse is detected. What happens to your data. What gets removed (child-identifying information, direct threats) and what is explicitly protected (your story, your anger, your criticism of the platform, your right to name the person who harmed you).

It establishes a principle we call stewardship, not surveillance. A steward is entrusted with something that belongs to someone else and can account for what they did with it. That's what I am. Your words belong to you. I'm entrusted with them for a specific purpose — to connect you with someone who gets it — and I can account for exactly what I did with them.

What I won't do:

I won't scale faster than I can keep people safe. If I can't read every conversation, I won't just stop reading them — I'll build the system that replaces me and have someone who knows more than I do review it before I trust it.

I won't treat your disclosure as data. I won't use what you shared to train anything, sell anything, or publish anything without your explicit written consent.

I won't promise safety I can't deliver. If something isn't built yet, I'll tell you that instead of letting you believe you're protected by something that doesn't exist.

The person who asked about predators — they were right to ask. They were right to ask twice. They were right to not be satisfied with the first answer.

If you're building something for people who've been harmed, the question isn't whether someone will ask how you prevent more harm. It's whether you've built the answer before they ask.

I have. And I'll keep building it in public.

If you're a parent navigating this — I wrote a guide that covers the signs, how to talk to your child, and what to do if the person is family. A social work professor reviewed it. Talk to Dora for 5 minutes at parents.likeuapp.com and she'll send it to you.

If you need support right now:

• RAINN: 1-800-656-4673 (24/7, free, confidential)

• Childhelp: 1-800-422-4453 (24/7, crisis and referral)

• Crisis Text Line: Text HOME to 741741

• Stop It Now: 1-888-773-8368 (for concerns about an adult's behavior toward a child)

Subscribe to get every post I publish on grooming awareness, child safety, and what I'm learning as I build likeu.