Privacy Policy — Mira

Last updated: 4 June 2026 (v1.0)

Mira is an AI-companion chat app. You choose a fictional AI companion (such as Nova, Sage, Pixel, or Bloom) and have a friendly, text-based conversation. This policy explains exactly what the app does and does not do with your data.

Mira’s companions are AI, not real people. Every companion is a fictional persona powered by a large language model. Replies are AI-generated and may be inaccurate. Mira is for light, safe-for-work conversation only — it is not a therapist, doctor, or advisor, and it is not a romantic, explicit, or adult-content service. If you are in crisis, call your local emergency number, dial or text 988 (US Suicide & Crisis Lifeline), text HOME to 741741, or find a helpline at findahelpline.com.

What this app collects

Mira does not collect any personal information from you. There are no user accounts; the app never asks for your name or email and does not store your data on a server we control.

DataCollected?Stored where
Name, email, account infoNo
Device locationNo
Contacts, photos, microphone, cameraNo — the app does not request any of these permissions
Your conversations with companionsSaved locally on your device onlyApp-private storage (Android SharedPreferences)
The messages you type to a companionSent as plain text to the third-party AI provider to generate a replyNot retained by the app once the reply comes back
Crash logs / analyticsNo

How AI replies are generated

When you send a message, the app sends a short plain-text prompt — the companion’s persona description plus your recent messages in that conversation — to Hugging Face’s Inference API (router.huggingface.co), with Pollinations as a fallback text provider. The provider returns the reply, which the app shows on screen.

No photo, contact, location, or device identifier is sent — only the plain text needed to continue the conversation. Your saved conversations stay on your device; you can delete any conversation at any time with Clear conversation inside a chat.

Content-safety screening

Before any message is sent to the AI provider, Mira screens it on your device. Messages containing crisis or self-harm language are not sent to the AI — instead the app shows a supportive message with helpline resources. Sexual or other not-safe-for-work requests are declined in-character and are not sent to the AI. This screening happens locally and is not logged or transmitted.

AI-generated content disclosure

Companion replies are produced by a third-party large language model. AI can be inaccurate, incomplete, or occasionally generate plausible-sounding but wrong content. Mira is for entertainment and friendly conversation only and is not a substitute for professional medical, psychological, legal, or financial advice. The developer is not responsible for outcomes resulting from following AI-generated content. On first launch you must acknowledge that companions are AI, and you can report inappropriate AI output at any time from a chat’s menu (Report a message) or from Settings → Report content.

Network connections

The internet is used for: (1) Google AdMob to load ads (see “Advertising” below), and (2) sending your message to the AI provider when you chat. The app requires an active internet connection to open — you will see a “You’re offline” screen when no connection is available.

Permissions the app requests

The app does not request photo, contacts, location, microphone, camera, notifications, health sensors, or any other sensitive permission.

Advertising

Mira shows ads provided by Google AdMob in these formats: banner (a small strip on the companion gallery), interstitial (an occasional full-screen ad after several actions), and app-open (a full-screen ad when you bring the app back to the foreground; never on first cold start). Google’s advertising SDK collects and processes limited data — for example, your device’s advertising identifier and approximate IP-based location — to serve and measure ads. This is governed by Google’s policies:

If you live in a region where consent is required (the EEA, UK, Switzerland, or a US state with applicable privacy law), the app uses Google’s User Messaging Platform (UMP) to gather consent or to serve non-personalized ads, so it stays compliant with GDPR / UK GDPR / CCPA. You can reopen these choices any time from Settings → Privacy choices.

Children

Mira is intended for users aged 13 and older. It is not classified as a “family” app on Google Play, and the AdMob ad content is mixed-audience. It is not designed for children.

Contact

If you have questions about this policy, contact the developer through the Google Play listing.