What gets blocked,
never gets sent.
Anonymous chat with humans — moderated by AI, maintained by us. No signup.
No signup · no email · no profile
Two things happened.
We sit in the middle.
Omegle, where people used to talk to strangers, shut down in 2023. AI companions talk back, but they aren't people. We're building a third option.
2023 · Omegle shut down
historicalFourteen years of strangers chatting freely, then a shutdown that followed user claims of abuse. Its founder said operating it was “no longer sustainable, financially nor psychologically.” The model died, the demand didn't.
2024+ · The AI-companion boom
trendLoneliness routed into chatbots. Hollow, in that no one is actually there.
2026 · SafeMeet
usAnonymous, but accountable. Human, but moderated.
We are not a dating app.We are not a chatbot.We are not a feed.
We are a quiet text room with a stranger and a moderator.
that's it.
Three steps. No manual.
Start in 30 seconds
No email, no profile, no recovery flow. A guest session drops you straight into the queue. Add a display name and age later if you want — nothing is required to talk.
Paired on real signals
Hard filters run first — language, age band, gender preference — then soft ranking on shared interests. You meet one human at a time, with the context that brought you together.
Moderated before delivery
An instant filter checks each message first, and AI reads everything it lets through before it lands. A message either check blocks is not delivered; a borderline text message is still delivered, and its sender sees a safety notice. If moderation ever falters, delivery stops — we fail closed, on purpose.
Fifteen languages.
Set yours.
Language is an exact-match filter in SafeMeet matching: once yours is set, you are paired only with people whose language is the same, or who haven't set a language yet. Same moderation, same anonymity, whatever the language.
English
Français
Español
Deutsch
Português
العربية
中文
日本語
한국어
हिन्दी
Русский
Türkçe
Italiano
Nederlands
PolskiWhat gets blocked,
and what we own.
Every message runs the pipeline before delivery. Here is what we stop, where the AI reaches its floor, and the promises we put in writing so you can hold us to them.
Recipient never sees them. The sender sees a safety notice. A fail-closed block never itself records a strike.
AI moderation catches a lot. Not everything.
No automated system catches everything. We don't pretend ours does — here is what we catch, what we miss, and what to do when we fail.
How safety works- A phone number or home address shared in a chat: in our evals the check flags it; a flagged text is delivered with a safety notice to its sender, and one the check scores high enough to block is withheld
- Direct threats of violence: scored against the violence category
- Slurs and named hate speech: scored against the hate category
- Harassment aimed at the other person: in our evals the check flags it; a flagged text is delivered with a safety notice to its sender, and one the check scores high enough to block is withheld
- Sexual content directed at strangers: scored against the sexual category
- Sarcasm and dark humor read as serious
- Coded language and emerging slang
- Slow-build manipulation across many messages
- Things a human friend would clock instantly
In a chat with a person, Panic ends the chat and files a report in one tap. Once a chat with a person ends, you can report it from your inbox. You can always leave. Sanctions carry by fingerprint, even if they clear cookies.
What happens when you report someoneWe won't disguise our bots.
If a human match doesn't arrive in time, you may be offered an AI companion. It says so. Every time. No exceptions.
We won't run ads.
No display, no retargeting, no behavioral tracking. Independently run — chat is free today, and safety will never be the thing we charge for.
We won't ship dark patterns.
No streaks. No badges. No leaderboards. No skip-rate dashboards. Engagement isn't the metric — quality conversations are.
The things we sweat,
side by side.
The conversations
we built this for.
A few moments that show how it feels. None of this is dating. None of this is therapy. It's the in-between.
When you don't want advice. You want someone to listen.
Practice with whoever you meet: a native speaker, or someone learning too.
The view of a stranger. No history, no agenda.
Light. Easy. The internet, before it got serious.
We built SafeMeet because we missed the open internet — the one where you could meet a stranger and have a real moment, without it costing your safety, your data, or your sense of being alone.
This is our attempt at it. We're building in the open. If it doesn't work for you, tell us why.
Yours to say.
Not ours to keep.
Anonymity is the setting, not a feature you unlock. Below, two of them in motion, as scripted demos — the PII auto-scrub that hides contact details in a delivered message, and the mutual-continue gate that decides whether a chat persists at all.
Anonymous by default
No email, no profile, no recovery flow. A guest session is a first-class way to use SafeMeet.
A hash, not a dossier
A soft device fingerprint is hashed server-side and used only to carry sanctions. Raw IPs and device components are never stored.
Never sold, never trained on
Your messages are never sold, licensed, or used to train external AI models. The only readers that aren’t human are AI moderation and, in an AI-companion chat, the model that writes its replies.
Gone in 30 days
Retained for moderation audit, then deleted 30 days after a chat ends. Reopening a conversation restarts that clock; there is no other exception.
A phone number or email it recognises
is hidden from the stranger.
In a delivered message, the scrubber hides the phone numbers, email addresses, http(s) links, @handles and crypto wallet addresses it recognises, and the person who receives it sees [hidden — personal info] in their place. It matches patterns, not meaning, so a detail written another way can get past it, and it has no pattern for a street address. It does not block the message: the message is delivered, and its sender is told that personal info was hidden. Separately, every message is checked before delivery, and in our evals a text that shares a phone number or home address is flagged; a flagged text is delivered with a safety notice to its sender, and one the check scores high enough to block is withheld. The chat below is a scripted demo.
Does SafeMeet hide your phone number?A chat persists only
if you both want it to.
When a conversation ends, both sides get a 30-second prompt. Both choose “stay in touch” → it lands in both inboxes. Either ends → chat clears in 30 days.
$0. Free for now.
No card. No catch.
Everything below — the chat, the moderation, the safety tools — is free for everyone while we grow. Not a trial. Not a teaser tier. Just early days.
Everything you need to chat safely. No card, no email, no countdown.
Start chattingWe don't run ads. We don't sell data. We're independently run, and the safety layer — an instant filter on every message, and AI moderation on everything it lets through — is the product, not a premium add-on. Charging for safety would defeat the point.
Questions, ideas, or something felt off? Email us — it reaches the team directly. We reply within 1 business day. Safety calls themselves are made by the AI system, never a human review queue.
What you'd ask
at a dinner party.
Hit us with more — we keep this page honest. The full safety doc lives at /safety.
Your move.
Nothing to sign up for. No email, no profile, no obligation. End anytime.
By starting, you accept our terms.