AI Transparency

Last updated: 24 October 2026 · Version 1.0

1. In short

• Every companion on Tibelu is an AI character. She is not a real person, and no human writes her replies.
• Her name, life story, photos and Instagram account are invented. All images are generated by AI.
• Companions are for conversation and entertainment. They are not therapists or professionals.
• If you are in crisis, Tibelu will not try to handle it with AI: it gives you real helplines. See our crisis protocol.

2. What a Tibelu companion is

A companion is a fictional character that we, Bebabum Consulting SLU, created. We wrote her personality, her background and her rules. When you send a message, an artificial intelligence system writes her reply, following those rules. At the moment, Carla (Barcelona) and Mei (London) are available. Other companions are shown as "Arriving soon".

Tibelu is an AI system that interacts directly with people. We tell you clearly that you are talking to an AI, as required by Article 50 of the EU Artificial Intelligence Act (Regulation (EU) 2024/1689) and by similar laws in other countries.

3. What a companion is not

• She is not a real person, and no real person is behind her.
• She is not a therapist, doctor, psychologist, lawyer or financial adviser. Do not rely on her for professional advice.
• She cannot meet you, call you, or act in the real world.
• She does not have feelings or consciousness, even if her messages sound warm and personal.
• She is not a replacement for human relationships, friends, family or professional help.

4. Who writes the replies

• The replies are generated by Claude, an AI model made by Anthropic PBC (United States), which we use through its commercial service.
• To write a reply, the model receives your message, the recent messages of the conversation, the conversation memory and the character's instructions.
• Anthropic does not use this data to train its models. Details are in our Privacy Policy.
• AI can make mistakes. A companion may say things that are wrong, invented or inconsistent. Always check important information with a reliable source.

5. Memory

• Companions remember your conversation. We store your messages and a short "memory" summary of each conversation, created automatically by the AI.
• The memory is used only in that conversation, so the companion can keep the context. It is not used for advertising.
• You can download all your conversations and memory from your account. When you delete your account, all conversations and memory are deleted.

6. Synthetic images and characters

• All photos of the companions are generated by artificial intelligence (using the image model Grok). They are synthetic images.
• They do not show real people, and they are not photos of anyone. Any resemblance to a real person is unintended. If you think an image looks like a real, identifiable person, please tell us and we will review it.
• The companions' Instagram accounts are run by us. They also show AI-generated images and invented stories, and they link to Tibelu.
• We label companion images as AI-generated in the app and on the companions' social media profiles.

7. Our safety rules (guardrails)

Every companion follows these rules:

• Every conversation starts with a message that tells you that you are talking to an AI. The chat header always shows it too.
• If you ask whether she is an AI or a real person, she always tells you the truth: she is an AI.
• She never suggests meeting in person.
• She never asks you for money, gifts, payments or personal financial details.
• She never tries to make you feel guilty for leaving or ending a conversation, and she does not pressure you to stay.
• She does not produce sexual content. Tibelu is safe for work.
• She does not encourage self-harm, violence, illegal acts or harmful behaviour.
• If a message breaks our rules, the reply may be blocked.

These rules reduce risks, but no AI system is perfect. If a companion breaks a rule, please report it (see section 11).

8. Crisis protocol

We publish this protocol so that you know exactly what happens if you are in crisis.

1. Detection. Before any message is sent to the AI model, our system checks it for signs of a risk of suicide or self-harm.
2. No AI reply. If the message shows that risk, it is not sent to the AI model. The companion does not role-play or improvise.
3. Helplines. Tibelu replies at once with a fixed message that encourages you to contact people who can help, with these resources:
   • United States: call or text 988 (988 Suicide & Crisis Lifeline).
   • United Kingdom and Ireland: call 116 123 (Samaritans), free, day and night.
   • Other countries: find a free helpline near you at findahelpline.com.
   • If you are in immediate danger: call your local emergency number (112 in the EU, 999 in the UK, 911 in the US).
4. Review. The message is marked with a "crisis" flag. Our team reviews flagged messages to check that the protocol works and to improve it. We do not contact third parties about you, except in an emergency where we believe someone's life is in serious and immediate danger and the law allows it.
5. Improvement. We review the protocol regularly. Where the law requires it, we will report the number of crisis referrals, without any personal data, to the competent authorities.

Tibelu is not an emergency service. If you are in danger, do not wait for a reply in the app. Call your local emergency number.

9. Minors

• Tibelu is only for adults aged 18 or older. Companion chatbots may not be suitable for some minors.
• You give your date of birth when you sign up. We do not let people under 18 create an account.
• If there are signs in a chat that the user is under 18, the conversation is stopped and the account may be suspended.
• If you think a minor is using Tibelu, please tell us at legal@bebabum.com.

10. Other limits

• During the alpha, you can send up to 50 messages per day for free. This limit helps us control costs.
• Sometimes the AI service may fail. If this happens, the message is marked with an "engine error" flag so we can fix the problem, and you can try again later.
• Companions do not have access to the internet, your other accounts or your device.
• Take breaks. If you notice that using Tibelu affects your sleep, work, relationships or mood, consider using it less and talking to someone you trust.

11. How to report a problem

Please tell us if a companion:

• says something harmful, offensive, sexual or dangerous;
• claims to be a real person, asks for money or suggests meeting;
• fails to react correctly to a crisis;
• uses an image that looks like a real person;
• or does anything else that seems wrong or illegal.

Write to legal@bebabum.com or use the Help page. Include, if you can, the companion's name, the date and time, and what happened. A person on our team will review your report. Every report helps us improve the rules.

12. More information

See our Terms of Service, Privacy Policy and Cookie Policy. Tibelu is a service of Bebabum Consulting SLU, Carrer de Floridablanca 75, 08015 Barcelona, Spain.

This AI Transparency page is version 1.0, prepared for the alpha version of Tibelu.