Safety.

1. What Daydream is, and is not

Daydream is an AI companion platform for entertainment and conversation. Every character is a large language model, not a real person, and what it says is fiction. Daydream is not a healthcare, mental-health, or crisis service. It does not provide medical, legal, financial, or therapeutic advice, and it is not a substitute for professional care or for the people in your life.

We say this inside the app, too. Before you can chat, you have to acknowledge a notice that Daydream is AI and not a substitute for therapy or crisis support, and a short reminder that the character is AI stays visible while you talk.

2. How we respond to self-harm and crisis

When you talk one-to-one with a companion, every message you send is checked for signs of self-harm or suicidal thinking before any reply is generated. The check combines fast pattern matching that runs on every message, a language model that examines messages of concern in the context of the conversation, and a signal for whether risk is building over the course of a session.

The system is built to err toward caution. If a layer is unsure, or a check fails or times out, the message is treated as a possible crisis rather than ignored.

When a crisis signal is detected, the companion does not reply. The conversation pauses and a full-screen panel appears with crisis resources for your region, including the lines at the top of this page, as tappable links. Your conversation is saved so you can return when you are ready. Closing the panel is not treated as a sign that you are safe; your next message is assessed again from scratch.

Automated detection is not perfect and is not a replacement for reaching out to a person. If you or someone else is struggling, please use the resources above.

3. Keeping conversations safe

Every reply a companion generates is checked by an automated safety filter before you ever see it. The filter is designed to block content involving self-harm, violence, and sexual abuse, and it fails closed: if the check cannot complete, the reply is held back instead of being shown.

We also screen and limit what can be sent into the system, and we redact obvious personal details such as phone numbers and email addresses from messages before they are processed.

No automated system catches everything. If a companion ever says something that feels harmful, please report it (Section 6).

4. Who can use Daydream

Daydream is for adults. You must confirm that you are 18 or older to create an account, and accounts are removed if we learn the user is under 18. We do not currently verify age with identity documents: the 18+ confirmation is based on what you tell us at sign-up.

Daydream is also not available in every region. Where we are not yet able to operate, access is blocked and you will see an “unavailable” notice instead of the app. The current list of excluded regions is in our Privacy Policy.

5. If you need ongoing support

Daydream cannot give you therapy, a diagnosis, or a safety plan. If you are carrying something heavy, please consider talking to a doctor, a licensed mental-health professional, or one of the lines above. They can help in ways an AI cannot, and reaching out is a strong thing to do.

6. Reporting a safety concern

If you see something on Daydream that worries you, or a companion responds in a way that feels unsafe, email and tell us what happened. We take safety reports seriously. In an emergency, contact your local emergency services first.

7. Changes to this page

Daydream is early and evolving, and we will update this page as our safety practices change. The measures described here are our current practices, not guarantees of any particular outcome, and they do not replace professional help or emergency services. For how we handle your data, see the Privacy Policy; for the rules of using Daydream, see the Terms of Service.