Safety and help

If you need help now

If you are thinking about suicide or self-harm, or you are in danger, please contact a person now. The bots on svchat are AI. They cannot call for help or stay with you.

United States: 988 Suicide & Crisis Lifeline(opens in a new tab), call or text 988.
UK and Ireland: Samaritans(opens in a new tab), call 116 123, free, day or night.
UK, under 19: Childline(opens in a new tab), call 0800 1111 or chat online, free, and it will not show on a phone bill.
European Union: Emergency services, call 112 if you or someone else is in danger.
Anywhere: Find A Helpline(opens in a new tab), free, confidential lines in your country.

Report a message

Every message from another person or a bot has a flag. Press it, choose why, and send. Nobody in the room is told who reported it.

We look at every report, and the urgent ones first. If a message breaks the rules we hide it for everyone. If a bot someone made does, we can stop it in every room.

If you leave a room, or someone removes you, you can still report it for 30 days. Open the room's link again and press Report something.

For anything else, like a name, a picture, or a room you are still in or left long ago, email [email protected].

Leave a room, or remove someone

In the room, press the people button at the top to see who is there. Press Leave room, then Leave. You can leave any room you did not make.

If you made the room, you can remove anyone from it. That also changes the room's link, so the old link stops working. Share the new link only with people you want in.

Nobody joins a room without saying yes first. If someone sends you a link, you can say No thanks, and they are not told.

You are talking to AI

Every bot on svchat is an AI, not a person. Where the law asks for it, in the EU, New York and California, a room with a bot says so at the top and every bot message is labelled AI. We show the same notices wherever we cannot tell where you are. In New York the room also reminds you at least every three hours. Bots can be wrong, and they can sound sure when they are.

svchat offers relationship coaching and conversation. It is not a medical or mental health service, and the bots are not clinicians or health professionals. If something a bot says upsets you, report it and talk to someone you trust.

Our crisis protocol

svchat checks each message sent in a room where a bot is taking part for signs of suicidal thoughts or self-harm. The check is done by an automatic classifier (TypeSafe's Jev model), with a list of phrases as a backup when the classifier does not answer.
It reads only the messages of people who have said yes to it, because it uses information about their health. You can say yes or no at sign-up, and change your answer on your profile at any time. The crisis lines above are there for everyone, and every room with a bot shows them at the top once you are in it.
In a room marked as a role play or story, what characters say in the story is not counted. Anything you say as yourself is. The backup list of phrases cannot tell a story from real life, so it may post a notice in a role play.
When it finds such signs, it posts a help notice in the room with the crisis lines above, at most once every ten minutes in a room. The notice goes to everyone in the room, wherever they are, even when the bots are paused.
Every AI bot is instructed to answer such messages with care, to point to the crisis lines, and never to describe ways to self-harm.

No automatic check catches everything. If you are worried about someone in a room, contact them directly, or an emergency service.

Young people

AI companions and chat bots may not be suitable for some minors. svchat is meant for adults. If you are younger and use it anyway, everything on this page is for you too, and Childline is there for anyone under 19 in the UK.