Liloo — Privacy Policy

Last updated: 2026-10-20 Contact: privacy@liloosafespace.com

This Privacy Policy explains what data the Liloo mobile application (application ID com.liloosafespace.app) and the website at liloosafespace.com process, for what purposes, on what legal basis, with whom the data is shared, and how you can control it. Cookies and similar technologies on the website are described in the separate Cookie Policy.

What Liloo is. Liloo is a communication-skills training app. You hold a conversation with AI characters that play the role of difficult interlocutors (for example, a conflict-prone relative or an aggressive colleague) so you can safely rehearse handling tense situations. The app also includes a supportive assistant chat ("Liloo"), scene debriefs, "Short conversation" practices, daily practice, the "Line of the day" exercise and audio meditations. For an AI character to reply, the text of your message is sent to our servers and to third-party language-model providers (see Section 4).

Key points.

  • We process what you write in Liloo so that the AI characters and the assistant can reply, and to keep the service safe.
  • With your permission, dialogue text is sent to AI model providers to generate replies; Liloo does not permit them to use it to train their models.
  • We do not sell personal information and do not collect an advertising identifier.
  • Statistics are shared with Google only if you select the separate checkbox. You can change this choice, AI-processing permission and internal quality review in Profile → Documents and agreements, and helpful emails in your profile.
  • Your data is erased with your account (Profile → Delete account) or on request at privacy@liloosafespace.com; the few records kept longer, such as purchase records and one-way hashes used to prevent abuse, are listed in Section 5.

1. Who processes your data (data controller)

The data controller is FOP Rudenko Yevheniia Yevhenivna (an individual entrepreneur registered in Ukraine), the developer and operator of the Liloo app.

Registered address: 29k Vladimira Vernadskogo St, 310, Dnipro 49000, Ukraine

For any privacy question, to exercise your rights, or to request deletion, contact:

Email: privacy@liloosafespace.com

Postal address: the registered address above.


2. Data we collect

CategoryDetails
Account identifiersAn anonymous (guest) user identifier or your email address, depending on how you sign in. Authentication is handled by Firebase Authentication and supports guest sign-in, email/password, Google Sign-In, and Apple Sign-In. When you sign in with Google or Apple, we receive your name, email address and, if the provider shares it, your profile photo.
Account restore on AndroidIf you use Liloo on Android without registering, the app keeps a random restore secret, generated by Liloo, in Google Play services' Block Store on your device. It is not uploaded to Google's cloud backup, though Android may copy it to a new phone when you transfer your data from this one. When the app is reinstalled on the device, it uses the secret to return you to your existing guest account — with your dialogues and your remaining free messages — instead of creating a new one. Our servers store only a one-way hash of the secret. For registered accounts the app can additionally keep a passkey-style restore credential with Android Credential Manager; where the device supports it, Android includes it in your end-to-end-encrypted Google backup, so the account can also be restored after a reinstall or on a new device. Our servers store only its public part. Some devices delete Block Store data together with the app, so the Android app also sends the device's Android ID — an identifier Android generates for Liloo on this device, which other apps cannot read and which changes after a factory reset — when it asks for a restore secret and when it finds none after a reinstall. Our servers store only a one-way hash of it, next to the restore secret, and use it to return the device to its guest account, to give the device free messages only once and, if an account on this device is blocked, to apply the block to accounts created on it later (see Section 3); it is never combined with the advertising ID. Neither the restore secret nor the restore credential is a device or advertising identifier.
Profile dataDetails you choose to give us: your name, gender, age group or date of birth, and interface language. Your country or region is taken from the device's region setting, not from location services, and is used to show the right crisis resources. If you do not state your gender, it may be inferred from how you write in a dialogue (for example, grammatical gender in Russian or Ukrainian) so that the AI character addresses you correctly.
Dialogue setup and avatar dataDetails you give when you prepare a dialogue in the chat (for example, the situation, including your answers to the follow-up questions of the helper that assists you in describing it; your interlocutor's name or role; their gender and age group); until the preparation is finished they are kept as a draft in your account so that you can resume it. Plus the name and settings of an AI character (avatar) you create, and its photo if you added one in an app version where that was available.
Conversation contentThe text of the messages you type or send in a dialogue, the AI's replies, the quick-reply suggestions the app offers and which one you choose, messages an AI character writes to you on its own initiative (see Section 3), and how a dialogue ended, for example when a character ended it. This is stored linked to your account.
Scene debrief, "Short conversation", daily practice and "Line of the day"The debrief of a finished scene that the AI produces: an assessment of your lines on several skills, quotes of your own lines with stronger alternatives and a line for the real conversation. Results of "Short conversation" practices and the phrases you saved. Daily practice tasks, your answers to them, the automated assessment of whether they were completed and the bonus credited. Your answers to the "Line of the day" exercise and the AI's short comment on them.
Dictation audio and recognized textCurrent iOS and Android versions use the device's system speech-recognition service. Depending on the device and its settings, Apple, Google, Samsung or another installed system provider may process audio locally or through its own network; Liloo does not receive or store that audio. Earlier supported mobile versions may upload dictation audio to Firebase and OpenAI for transcription when local recognition is unavailable. The web test-drive does not offer dictation. In every version, recognized text remains editable and is processed like an ordinary message only if you choose to send it.
Photos (camera / gallery)Photo uploads are disabled in new versions of the app. An image you captured with the camera or selected from your gallery for your profile or an avatar in earlier versions is kept in cloud storage and associated with your account. Photos are not sent to AI providers.
Safety and moderation dataResults of automated safety checks on dialogue text — for example, a signal of self-harm risk, a disclosure of real danger from a real person, sexual content involving minors, attempts to bypass a character's safeguards, or text indicating that a participant in the dialogue may be under 18 — together with the time, the direction (your message or the AI's reply) and the action taken. Any restrictions placed on an account, including those carried over to it from another account on the same device or app installation. See Section 4.
Device push tokens and timezoneYour device's Firebase Cloud Messaging push token (in the push_tokens field on your account) and your device timezone. A token is stored when you allow notifications. On iOS the app may also register, without asking, for quiet ("provisional") notifications, which appear only in Notification Center until you choose to keep or turn them off; a token is stored for these too. The timezone is used to respect notification quiet hours and to count calendar days for the daily free-message allowance.
Free-message allowance and purchasesYour message balance, the days on which you received the daily free-message allowance and when it expires, the registration gift and daily practice bonuses, bonus codes you redeem, subscription or purchase status, and free Premium access if we granted it. Purchases themselves are processed by Apple or Google; we receive the purchase and entitlement records through RevenueCat.
Technical, crash and diagnostic dataTechnical data needed to run the service (such as conversation and message identifiers, app version and platform), crash/error diagnostics and stack traces (Firebase Crashlytics, which also records the device model and operating-system version), and our server logs, which include the IP address and request details.
Product usage and Android install attributionA random Liloo installation identifier, your account identifier, app version, language, screens viewed, taps on app controls, and funnel events (for example, onboarding completed, scenario started or paywall shown), and — where we test a variant of a feature — which variant you were assigned. On Android, the app reads Google Play Install Referrer once to obtain allowlisted campaign parameters and install/click timestamps. A Google Ads click identifier, when present, is converted to a one-way hash before storage. Message text, the Android advertising ID and the complete raw referrer are not included. The installation identifier is also used so that an app installation gets free messages only once and a block on an account applies to new accounts in the same installation (see Section 3).
Email communicationYour email address and name for service emails — password resets and a one-time welcome email after you register — and whether the email was delivered, opened or a link in it clicked. If you turn on Helpful emails, also your consent to them (time and version), the emails sent and your opt-out.
In-app bug reports and feedbackThe text you write, plus your account identifier, app version, platform, the conversation it concerns and your message balance, so we can reproduce the problem. If you leave feedback with your name or email, we process those too.
Website reports and support correspondenceIf you use the website report form, we process your name, optional email address, subject, message, page URL, IP address and browser user-agent so we can investigate and reply.
Friend invitationsWhen a registered user asks Liloo to invite a friend, we process the friend's name, email address, an optional personal note, the inviter's identity, and technical delivery/interaction outcomes (provider acceptance, destination-server acceptance, open, link visit, bounce, spam flag or opt-out).
Permission choicesWhen you accept documents or change a permission, we record the time, version and language of the text shown, screen and app build, your choices for AI processing, internal quality review and Google advertising measurement, and, on iPhone, the App Tracking Transparency status for the app installation together with its identifier.

We do not access your contacts and we do not collect your precise location or an advertising identifier. The app requests permissions for the microphone and iOS speech recognition (system dictation), and notifications (push) — each used only for that feature. On iPhone the app also requests App Tracking Transparency permission — only if you selected the Google advertising measurement checkbox. Earlier versions also requested access to the camera and photo library for profile and avatar photos.

Photos of other people. If you uploaded a photo of another person in an earlier version of the app, you must have had their consent or another lawful basis. Liloo does not verify who appears in photos. We may remove the photo and the related character data if we reasonably believe that it infringes third-party rights.


3. Purposes and legal basis

PurposeWhat it involvesLegal basis
Provide the servicePrepare a dialogue in the chat, including with a helper that assists you in describing the situation; generate the AI character's replies, run the Liloo assistant, scene debriefs, "Short conversation" practices, daily practice, the "Line of the day" exercise (ordered by the topics of your recent scenarios) and meditations; keep your avatars, dialogues, preparation drafts and history; manage your account, free-message allowance, subscriptions and purchases; support dictation.Performance of our contract with you (the Terms of Use); before sending your messages to AI providers, the app also asks for your permission and allows you to withdraw it (see “Permission for AI processing and quality review” below).
Keep the service safeCheck dialogue text automatically for self-harm risk, disclosures of real danger, sexual content involving minors, requests for instructions on violence, weapon-making or other illegal activity, further prohibited content and attempts to bypass a character's safeguards; adapt or pause a dialogue, show support resources, block or replace unsafe replies; restrict accounts that break the Terms, including carrying a block over to accounts created after it on the same device or in the same app installation.Our legitimate interest, and yours, in preventing harm; where the law requires, protecting vital interests and complying with legal obligations.
Review and improve quality and safetyAuthorised members of our team may read individual dialogues, setup details and feedback to investigate problems, review safety incidents and improve characters and filters. We also run automated evaluations in which real dialogue excerpts are assessed by language models, or answered by a model we are testing, without being shown to you. Evaluation copies of model requests are kept for up to 14 days. If you turn off internal quality review, your dialogues are not opened for quality review or included in automated evaluations; review of flagged safety events continues.Your permission for internal quality review (see “Permission for AI processing and quality review” below), which you can turn off separately. Safety-incident review is limited to safety purposes and uses its own legal basis.
Protect the free tier and prevent abuseKeep your account across reinstalls on the same Android device, so that a reinstall does not grant the free first practice and starter messages again; give the starter allowance only once per device and daily free messages once per account and device per day; give the registration gift once per account and device, and the daily practice bonus at most once per day per account and device; detect when a new account continues the same dialogue as accounts that already used up their free allowance (by comparing the situation brief), and in that case give the new account a smaller free allowance; limit how often bonuses and invitations can be used.Our legitimate interest in preventing abuse of the free tier.
Send notifications and messagesDeliver push notifications — for example a reminder, a daily-allowance notice, or a message an AI character writes to you on its own initiative when you left a dialogue unfinished (the message appears in that chat and may be shown in the notification). You can turn notifications off at any time in Profile → Notifications or in your device settings; AI-initiated messages are not sent without your permission for AI processing, to dialogues paused for safety or to dialogues involving a minor, and none is sent for seven days after a message of yours signals a risk of self-harm.Your consent to notifications; for in-chat messages, our legitimate interest in helping you finish a rehearsal.
Send service emailsPassword resets, the one-time welcome email, and invitations a friend asked us to send, with delivery and open/click tracking.Contract (password reset, welcome); legitimate interest (invitations; the recipient can opt out).
Send helpful emailsOptional reminder emails about Liloo — at most two: one the day after you create an account if you have not started practising, and one about a week after you create it if you practised but have not returned for three days.Your consent: you turn on Helpful emails in your profile and can turn them off there or via the link in any such email.
Diagnose crashes and errorsUse crash diagnostics and server logs to keep the app reliable and investigate failures.Our legitimate interest in a working service.
Measure and improve the productUnderstand how people use the app and which features and variants work, using first-party product events without conversation text; test variants of a feature on a random share of users.Our legitimate interest in improving the product.
Improve the app and measure advertisingWe share technical app usage statistics with Google (Google Analytics and Google Ads) to improve the app and measure advertising: to understand which install sources and campaigns lead to registration, dialogue practice and purchases. We also use Google Play Install Referrer.Your consent — a separate optional checkbox on the first screen (unticked by default and not part of accepting the Terms), or the “Google advertising measurement” switch in Profile → Documents and agreements, where you can also turn it off; on iPhone, system tracking permission is also required. On the website — through Cookie settings.
Respond to website reports and support requestsInvestigate the issue, prevent duplicate or recurring problems, and contact you when you provide an email address.Our legitimate interest in responding to you.
Comply with the lawKeep purchase records for accounting and tax; respond to lawful requests; report child sexual abuse material where required.Legal obligation.

Where we rely on consent, you can withdraw it at any time; this does not affect processing that took place before. Where we rely on legitimate interest, you can object (see Section 7).

Permission for AI processing and quality review

The app's first screen has two checkboxes, both unticked by default. The first is optional: permission for Google advertising measurement (see Section 4). The second is required: you confirm that you are 18 or older, accept the Terms of Use and this Policy, and allow AI processing and internal quality review. Start is unavailable without the second checkbox.

AI processing. Your messages, relevant history and situation details are sent to AI services on DigitalOcean and OpenAI to generate replies, debriefs and safety checks (see Section 4). This may include personal information you provide about relationships or your emotional state.

Internal quality review. Dialogue excerpts and related records may be evaluated by AI services, including Anthropic, and authorised staff with restricted access to check and improve the service's quality and safety. This is not permission to train models or to use conversations for advertising.

You can change your choices in Profile → Documents and agreements, using “New AI replies and conversation reviews”, “Internal quality review” and “Google advertising measurement”. If you turn off quality review, your dialogues are excluded from internal evaluations and conversations continue as usual. If you turn off AI processing, new AI replies, dialogue preparation, debriefs, “Line of the day” comments and AI-initiated messages are unavailable until you allow it again. History and saved results remain available, and crisis contacts remain on the 24/7 tab, in the chat header and in Profile. Withdrawal applies to future processing. Processing needed for safety, legal obligations and your data requests continues on its own legal bases.

We record every choice (see Section 2). If what you accept on the first screen changes, the app asks you to confirm the new revision before the next AI reply. Accepting the Terms does not waive your data rights or replace a legal basis for processing.

Automated decisions

Some decisions about your dialogue are made automatically, without a person involved:

  • Safety measures — a reply may be blocked or replaced, a dialogue may be paused, softened or held open, and support resources may be shown, based on automated checks of the text.
  • Dialogues that may involve a minor — if the setup, the situation description or your messages indicate that you or the character are under 18, the AI character automatically avoids humiliating language, romance and sexual content and hands the conversation to the Liloo assistant sooner.
  • A character ending a dialogue — if an automated check finds that a message insults the AI character, the character first warns you in its role; if another insult aimed at it follows within the next few messages, the dialogue ends and the Liloo assistant explains why. Setting a boundary, expressing anger or quoting someone else does not count. This never happens when there are signs of crisis or self-harm, a possible minor or sexual content, in the introductory dialogue after you install the app or in chats with the Liloo assistant.
  • Free allowance — an account created on a device or in an app installation where another account, including a deleted one, has already had the starter allowance gets no new starter allowance. Daily free messages can go to the active account, once per account and device per day. A salted device hash, pinned timezone and last credit time remain after account deletion to prevent another daily credit on the same device that day; this record has no account identifier. A new guest account whose situation brief repeats one already used by accounts that exhausted their free allowance receives a smaller free allowance.
  • Daily practice bonus — credited automatically when the system confirms that the practice tasks were completed. The check assesses participation in the practice, not the quality of your answers.
  • Carrying a block over — if an account is blocked, the block automatically applies to accounts created after it on the same device or in the same app installation. Accounts that existed before the block are not affected.

None of these decisions has legal effects for you. If you believe one was wrong, write to privacy@liloosafespace.com and a person will review it.


4. Sharing with third parties (processors)

To run the app we use third-party providers that process data on our behalf. Your conversation content leaves your device and is processed by these services:

ProviderWhat it processes
DigitalOcean (DigitalOcean, LLC)Hosting of the backend (the API at api.liloosafespace.com, the website and the PostgreSQL database) that stores conversations, profiles and moderation records. A configured AI model running on DigitalOcean infrastructure also generates most AI replies, setup summaries and opening lines: it receives the dialogue text, the situation brief and the details you gave for the dialogue.
OpenAI (OpenAI, L.L.C.)Language models that generate AI replies when the primary model is unavailable and for some safety-sensitive steps (for example, for a day after a disclosure of self-harm risk, while a scene is held after a disclosure of real danger, when a dialogue may involve a minor, or when a scene turns sexual); scene debriefs and the comment on a "Line of the day" answer; automated safety classification of dialogue text (moderation, detection of real danger, and the check whether a message insults the character); and, for earlier supported mobile versions only, transcription of dictation audio when the device's local recognizer is unavailable.
Google Firebase / Google Analytics / Google Ads (Google LLC)Authentication (Firebase Auth), the database of dialogues and profiles (Firestore), storage of avatar photos (Storage), server functions (Cloud Functions), push notifications (Cloud Messaging), crash/error reporting (Crashlytics), and — only if you allow Google advertising measurement (and, on iPhone, tracking) — technical app usage statistics, to improve the app and measure advertising: pseudonymous technical events (for example, registration, dialogue start or a purchase with its value and currency).
Apple (Apple Inc.)Delivery of push notifications to iPhones (Apple Push Notification service); in-app purchases on iOS.
Apple / Android system speech provider (Apple, Google, Samsung or another provider installed on the device)Optional dictation audio and the resulting text when you use voice input. Processing may occur locally or on the provider's network according to the device, provider and account settings. Liloo does not receive or store this audio in current app versions.
RevenueCat (RevenueCat, Inc.)Subscription and in-app purchase management. Processes your user identifier (Firebase UID) and purchase data.
Mailjet (Sinch)Delivery and technical status tracking of email — password resets, the welcome email, friend invitations and helpful emails if you turned them on. Processes the recipient name/email and delivery/interaction outcomes.
Atlassian (Trello and Jira)Issue tracking for website reports, in-app bug reports and product feedback, as described in Section 2.
Anthropic (Anthropic, PBC)AI tools our team uses for engineering and for the internal quality and safety review described in Section 3 — for your dialogues only if you have not turned off internal quality review; these may process dialogue excerpts and related records.

The AI providers above process your data to provide their services to us. Liloo does not permit them to use conversation content to train their models. Internal quality review evaluates our service; it is not model training.

We do not sell your personal data. Sharing is limited to the processors above, Apple and Google as app-store operators for purchases, and cases required by law.

Where your data is hosted and international transfers

  • The backend application, the website and the primary database are hosted on DigitalOcean in London, United Kingdom (LON1 region).
  • The AI model on DigitalOcean infrastructure runs in Canada and the United States.
  • Firestore and Firebase Storage are located in the European Union (multi-region); Cloud Functions run in Belgium (europe-west1).
  • Firebase Authentication, Cloud Messaging, Crashlytics, Google Analytics, Google Ads, OpenAI, RevenueCat, Anthropic and Atlassian process data on their own infrastructure, which may be located in the United States or other countries.
  • Encrypted backups are also kept on storage controlled by the operator.

Where data is transferred outside the UK, the EU or Ukraine, such transfers rely on those providers' standard contractual clauses or equivalent safeguards (including, where applicable, the EU–US Data Privacy Framework).

Website analytics and cookies

The website uses Google Analytics 4 and Google Ads only if you accept them in the cookie banner; nothing is loaded before you choose. You can accept analytics and advertising separately and change your choice at any time through Cookie settings in the website footer. See the Cookie Policy for the full list of cookies and similar technologies, their providers and lifetimes.

In the mobile app, we share technical app usage statistics with Google to improve the app and measure advertising only if you select the separate checkbox on the first screen or enable it in Profile → Documents and agreements. Sharing is off by default; a choice from an older version is not carried over. This permission is not a condition of using Liloo. On iPhone, system tracking permission is also required.

After permission is granted, Firebase Analytics creates a random app-instance identifier and automatically collects standard app events (for example, first open, session start and screen view). Liloo sends Google Analytics and Google Ads only approved technical event names (such as registration, dialogue start, activation, trial or purchase), their timestamps and, for a purchase, its value and currency. We do not send an advertising ID, Firebase UID, email, profile data, scenario/persona names, dialogue text or safety records for this purpose; ad personalisation remains disabled. You can turn off sharing in Profile → Documents and agreements on either platform, which stops future sharing. Liloo's first-party product analytics remains separate. Android also reads Google Play Install Referrer once for campaign measurement, and Firebase Crashlytics is used for crash/error diagnostics as disclosed above.

Crisis-safety signals and moderation review

Liloo checks dialogue text in real time so it can respond safely. These checks always run and are not an advertising or analytics feature:

  • Self-harm risk. The role-play stops for that message, and the message itself is not passed to language models — neither for a reply nor as context; you receive a reviewed supportive response and local support resources, and then a gentle mode applies: for your next six messages in that scene and for a day in all your chats, the character does not attack you and the scene is not closed by a finale or by the character walking out. The situation description is checked the same way: if it shows signs of self-harm risk, the scene runs in the gentle mode from the start.
  • Real danger from a real person (for example, violence at home). The scene is held open instead of ending, the character may step out of role to put your safety first, and support resources are shown once. For the next 7 days, some features that could close or interrupt your other scenes are paused. The situation description is checked the same way: if it reveals real danger, the scene is held from its first line. Such messages are also recognised in the chat with the Liloo assistant.
  • Sexual content involving minors and other prohibited content is blocked, and repeated or serious violations can lead to account restrictions.
  • "Line of the day". If an answer to the exercise shows signs of crisis, support resources are shown instead of a comment. For 7 days after a crisis, and when there are signs that you are under 18, the exercise is not shown.

A critical event is retained for authorised human moderation with the source message, internal account and conversation identifiers, matched safety trigger, direction (user input or AI output), time, system action, review status and moderator notes. A real-danger event is recorded without the message text. These records are used only for safety review, incident response and improving the safety filters; they are not exported to Google Analytics or Google Ads and are not used for advertising or personalisation.

See the separate Consumer Health Data Privacy Policy for the complete categories, sources, processors, retention and rights notice, and the Safety Protocol for how the app responds.


5. Data retention

We keep your data only for as long as needed for the purposes above. The schedule below sets out how long each category is retained:

  • Account, profile, avatar and device-timezone data — kept while the account exists; erased within 30 days after you delete your account.
  • Conversations and messages — kept while the account exists; erased within 30 days after account deletion. If you delete a single dialogue in the app, it disappears from your chat list and from the app's database, but its copy on our server is kept until the account is deleted.
  • Scene debriefs, "Short conversation" and daily practice results, saved phrases, "Line of the day" answers and dialogue preparation drafts — kept while the account exists and erased with it.
  • Product usage events, allowance and experiment records — kept while the account exists and erased with it; records of individual taps on app controls are kept for up to 90 days. After account deletion, only one-way hashes of the identifiers of the app installations and devices where the account was used remain, with the date from which they had free messages, marks of a registration gift or practice bonuses given there, and no account identifier, so that deleting an account does not give a device free messages again.
  • Safety and moderation events (results of automated safety checks, including critical events) — kept while needed for safety review, no longer than 90 days, and erased with account deletion or an earlier verified privacy request. Records of account restrictions may be kept in a form no longer linked to you after account deletion, to prevent repeated abuse.
  • Evaluation copies of model requests — up to 14 days.
  • Dictation audio — current app versions do not store it in Liloo infrastructure; the system speech provider handles it according to the device and provider settings. Earlier supported versions process audio uploaded to Liloo's transcription path ephemerally and delete it immediately after transcription. Recognized text follows the conversation lifecycle above only if you send it.
  • Push notification tokens — kept until you disable notifications, the token stops working, or you delete your account.
  • In-app bug reports, website reports and support correspondence — kept while the issue is investigated and for as long as reasonably needed to identify duplicate or recurring problems.
  • Friend invitations — kept with the inviter's account so the inviter can see the result and receive the promised referral bonus. They are deleted when the inviter deletes the account. An invited person can stop later mail from the invitation or ask privacy@liloosafespace.com to delete contact data. If a recipient opts out or the address bounces, a one-way hash of the address is kept so that we never email it again.
  • Consent to helpful emails, your opt-out and records of the emails sent — kept while the account exists and erased with it. If an email to your address could not be delivered (a hard bounce), you marked it as spam or you unsubscribed through your email provider, a one-way hash of the address is kept so that we never email it again, including after the account is deleted.
  • Purchase and subscription records — retained for up to 3 years after the transaction to meet accounting and tax obligations, even after account deletion, then erased.
  • Server logs and diagnostics — up to 90 days.
  • Backups — overwritten within 30 days.
  • Permission choices — retained until account deletion. The tracking-permission status of an app installation may remain without the deleted account identifier while another account uses that installation.

When you delete your account, your data is removed as described in Section 6, except minimal records we are required to keep by law or to prevent abuse.


6. Account and data deletion

If you have a registered account, you can delete it directly in the app: Profile → Delete account. On confirmation, the app deletes your Firebase Authentication user; if you gave a reason for deleting, we keep it without a link to you. This automatically triggers a server-side cascade cleanup:

  • your dialogues, messages and dialogue preparation drafts are deleted;
  • your avatars and their associated storage files (including photos) are deleted;
  • associated backend data (PostgreSQL) is purged.

A guest account stays linked to the Android device it was used on: reinstalling the app returns to it, and so does signing out of a registered account on that device. Clearing Liloo's storage (Android Settings → Apps → Liloo → Storage → Clear storage) removes the restore secret from the device, but the device's Android ID still leads back to the guest account until the device is reset to factory settings. The secret stops working once you register — you then sign in with your chosen method — and our copy is deleted together with your account. Signing out of or deleting a registered account also removes its restore credential from the device. To have a guest account deleted, email privacy@liloosafespace.com.

If you cannot use the in-app option or want to confirm deletion, email privacy@liloosafespace.com.


7. Your rights

Depending on your jurisdiction (including under the GDPR, the UK GDPR and the Law of Ukraine "On Personal Data Protection") you have the right to:

  • access your data and obtain a copy, including in a portable format;
  • correct inaccurate data;
  • delete your data (see Section 6);
  • withdraw consent you previously gave (including by revoking camera, microphone, notification and iOS tracking permissions in your device settings, turning off AI processing, internal quality review or Google advertising measurement in Profile → Documents and agreements, or changing Cookie settings on the website);
  • object to processing based on our legitimate interest, and restrict processing;
  • ask for a person to review an automated decision (see Section 3);
  • lodge a complaint with a data protection supervisory authority — for example, in Ukraine, the Ukrainian Parliament Commissioner for Human Rights, or the authority in the country where you live.

To exercise these rights, contact privacy@liloosafespace.com. We answer within 30 days.


8. Notice for California residents (CCPA/CPRA)

Liloo does not sell personal information. In the mobile app, ad personalisation is turned off. If you allow advertising on our website — with Accept all or the Advertising switch — Google may use the related identifiers to measure and personalise ads, including remarketing, which under the CCPA/CPRA may count as "sharing" for cross-context behavioural advertising. You can opt out at any time by turning Advertising off in Cookie settings.

The categories of data we collect are described in Section 2. Conversation content that reveals a mental or emotional state may be sensitive personal information; we use it only to provide the service and for safety and quality, never for advertising (see the Consumer Health Data Privacy Policy).

California residents can ask to know, delete or correct their personal information, and to limit the use of sensitive personal information, by writing to privacy@liloosafespace.com. We will verify the request and respond within the time the law requires. Exercising these rights will not lead to discrimination.


9. Children and eligibility

You must be at least 18 years old to use Liloo; you confirm it on the app's first screen with the required checkbox before tapping Start. The app is rated 18+ on the App Store and is not directed to children. It contains simulated conflict content with mature, mild-profanity dialogue and is intended for an adult audience; the minimum age of use is 18 regardless of the store rating.

We do not verify age with documents. If the details of a dialogue or the messages in it indicate that a participant may be under 18, the app automatically makes the dialogue gentler and offers support sooner (see Section 3); we may also restrict or remove the account. If you believe a person under 18 has provided us with their data, email privacy@liloosafespace.com and we will delete it.


10. Security

All network traffic between the app and our services is encrypted in transit using HTTPS/TLS. Firestore access is governed by per-user, owner-based security rules, and the backend API uses token-based authentication. Access to dialogues and moderation records inside our team is limited to authorised staff. Data at rest is protected by the underlying infrastructure of our processors (Google Firebase and DigitalOcean), and our own backups are encrypted.

If a personal data breach is likely to result in a high risk to your rights and freedoms, we will notify you and the competent authority where applicable law requires it (including Articles 33–34 GDPR).


11. Changes to this policy

We may update this Policy. For material changes we will update the date at the top of this page and, where appropriate, notify you in the app before the change takes effect.

If a material change introduces a new purpose of processing or needs new consent, we will ask for consent before the new processing starts. Where processing relies on legitimate interest, we will update this Policy in advance so that you can object. If the change affects what you accept on the first screen or AI processing of your data, the app asks you to confirm the new revision before the next AI reply.

Previous versions of this Policy are available on request at privacy@liloosafespace.com.

Version history

DateWhat changed
2026-10-06New version of the Policy.
2026-10-13Daily free messages: at most once per day per account and device.
2026-10-20First-screen checkboxes: required age confirmation, acceptance of documents, AI processing and internal quality review; optional Google advertising measurement; switches in Documents and agreements; Short conversation, daily practice and the situation helper; helpful emails by consent; registration gift and practice bonus; photo uploads disabled.

FOP Rudenko Yevheniia Yevhenivna (individual entrepreneur, Ukraine) · privacy@liloosafespace.com · Last updated 2026-10-20