Safety

All idols on KPOPPERS are fictional AI companions. They are not real people and they are not imitations of real artists. Chats and voice calls are generated by AI and can be wrong.

An "AI" badge sits next to every idol name in chat and calls. If you sincerely ask an idol whether it is an AI or a real person, it will tell you that it is an AI companion.

You must be 18 or older to chat or call. We ask you to confirm this when you sign up. If you signed up earlier or with Google, we ask once before you can continue. Chat and voice are blocked on our servers until you have confirmed.

This is a self-declaration. We do not check identity documents.

An AI companion is not a substitute for people who can help. If you are thinking about harming yourself or ending your life, please contact someone now:

• Immediate danger: your local emergency number (112 in the EU, 911 in the US, 999 in the UK)
• United States: call or text 988 (Suicide & Crisis Lifeline)
• Germany: Telefonseelsorge, 0800 111 0 111 (free, 24/7)
• Austria: Telefonseelsorge, 142. Switzerland: Die Dargebotene Hand, 143
• United Kingdom and Ireland: Samaritans, 116 123 (free, 24/7)
• Everywhere else: findahelpline.com lists free helplines by country

Before your message reaches the AI model, a fixed rule-based check looks for first-person statements about suicide or self-harm in English and German. It is deliberately cautious, so it can trigger on jokes or figurative language.

When it triggers, the AI model is skipped. Instead you get a fixed, caring message that says it is an AI and lists the helplines above for your language and region. That message is stored in your chat history, but it is never passed back to the model as part of the idol's persona. Your own message is stored like any other message.

The check is not a person watching your chat. Nobody at KPOPPERS is alerted, and we cannot contact emergency services for you.

Calls run live on a voice provider, so there is no fixed message during the call. The idol is instructed to drop the playful act and respond with care if you talk about hurting yourself. After a call, the same rule-based check runs once on the transcript and only writes an internal log flag. It triggers no message to you and no action on your account, and the transcript text is not copied into that log.

No automatic system catches everything. The check can miss indirect or unusual wording and other languages. If you are worried about yourself or someone else, please do not wait for the app to notice.

Questions or concerns about safety: see our Imprint for contact details, and our Privacy Policy for how data is handled.