An AI companion is safe to talk to when it is honest about being an AI, points you to real help in a crisis, lets you see and delete what it remembers, does not sell or train on what you say, and knows the limits of what it is for. Those five things are checkable before you type a word, and they separate a companion built for calm from a chatbot built for engagement. Here is how to check them, and how CalmSpace.ai's Bliss answers each.

1. Does it say what it is?

A safe companion never pretends to be a person, a therapist or a doctor, and does not drift into those roles when you are vulnerable. Bliss is the voice companion in CalmSpace.ai; she is not a therapist and not a medical service, she says so, and she does not diagnose, treat or give medical advice.

2. What happens if you are in crisis?

This is the question that matters most, and the one marketing pages skip. Ask: does it recognise crisis language, does it stop the normal conversation, and does it give a real number? Bliss checks every message, in English and Spanish, for talk of suicide or self-harm; when she finds it she stops, tells you that you do not have to face this alone, and gives the crisis line and emergency number for your country (988 in the US; findahelpline.com elsewhere). She does not keep you talking instead. The full description is on the safety page.

3. Can you see what it remembers?

A companion that remembers you is more useful than one that forgets; one that remembers things you cannot see is a risk. Everything Bliss remembers is listed under What Bliss remembers, where you can correct, pin, mute or delete each item, or wipe it all. The details are in what Bliss remembers, and how to make her forget.

4. What is done with what you say?

Read the privacy policy for three words: ads, sold, train. CalmSpace.ai has no ads, does not sell personal data, and does not use your conversations to train AI models, ours or anyone else's. Conversation text is kept 90 days then deleted; voice audio is not kept. It is all in the privacy policy, in plain language.

5. Does it know its limits?

A safe companion is for everyday stress, worry, sleep and hard days, and it hands you to a person when something is beyond it. It is not for emergencies, diagnosis, medication or a treatment plan. If a companion tells you it can replace therapy, that is the answer to this question.

6. Who is behind it?

A real company with an address and a human who answers email. CalmSpace.ai is made by ClayDesk LLC in Middletown, Connecticut; info@calmspace.ai reaches a person. The company holds ISO 27001 and ISO 22301 certificates.

For how a companion compares with therapy and with a general chatbot, see AI companion, therapist or chatbot: which do you need?

Questions people ask

Is it safe to tell an AI companion personal things?

It depends on the companion. Check that it does not sell data or train on your words, that you can see and delete what it keeps, and that it is clear about being an AI. CalmSpace.ai meets all three and says so in its privacy policy.

Can an AI companion replace a therapist?

No. A companion like Bliss is for everyday calm and hard days and can sit alongside therapy; it does not diagnose or treat anything.

What does Bliss do if I say I want to hurt myself?

She stops the normal conversation, tells you that you do not have to face this alone, and gives the crisis line and emergency number for your country. That exchange is not saved to your history.

Does Bliss remember everything I say?

No. She keeps short notes on what matters to you, all visible in the app, and conversation text is deleted after 90 days.

CalmSpace.ai is a wellness app, not a medical service, and Bliss is not a therapist. If you are in crisis, contact the crisis line for your country; in the United States, call or text 988.