Privacy Policy

1. Your account and reading records

Leung Chin Ho operates TataRead and issues this policy. When you sign in with Apple or Google, TataRead receives a sign-in identifier and account details supplied by the sign-in service, such as an email address. We use them to authenticate and separate accounts. Your books, covers, chapters, lists, passages, notes, tags, favorites and text drafts support reading and synchronization across your devices. Supabase provides authentication, database and image storage. Your device keeps local data for offline reading and continued editing. Account features require a sign-in identifier. Covers, page photos and notes are optional. When you search the catalogue, TataRead sends your query to its database to find matching editions. Sign-in credentials use the device’s secure storage. Connections to the backend are encrypted, and account access controls restrict access to private records.

2. Camera, photos and text recognition

You can use the camera or select a photo. Saving a cover associates its image with a book. When you submit a page for recognition, the service sends images of the marked text and, when source suggestions are enabled, header or footer crops used to suggest a chapter or page number to Alibaba Cloud Model Studio’s Qwen service. The original page photo is used for that editing session and is excluded from saved passages and text drafts. You can enter text manually and withdraw camera permission in device settings. Before submitting, check the selection, header and footer for identifying information and leave it out. Cover correction runs on your device. Saved private covers sync to your account.

3. Subscriptions, usage and support

For store purchases, the purchasing store (Apple or Google Play) handles payment. RevenueCat and TataRead’s backend use an app account identifier, product, transaction and validity information to verify purchases, restore entitlements and allocate usage. The recognition service records operation and usage information needed to enforce quotas and avoid duplicate charges. Contact Support creates an editable email with the app version, operating system and device model. It reaches support only if you send it.

4. Widgets and sharing

If you add a widget, people who can see your Home Screen may see a passage from the source you selected. Share images are generated on your device. You choose the recipients or destination app. Recipients or destination services control the copies they receive.

5. Account deletion, retention and your rights

You can edit your reading records. To request deletion, open your account screen. You can also email lazypig2218@gmail.com with the subject “Delete TataRead account”, your sign-in method and account email, without reinstalling the app. After verifying identity, we process the request and remove account identifiers, bookshelves, private covers, passages, notes and drafts. Saved reading records generally remain until you delete them or your account. Usage, security, duplicate-prevention and legally required records are kept only as long as needed for those purposes. Deletion does not cancel a store subscription. Recipients control copies you have shared. Contact the same address to request access to or correction of personal data or to ask about its handling. After deletion, the system may retain an operation ID and a hashed verification value so other devices can confirm the outcome. This receipt contains no passage text. The corresponding completion receipt is removed when the device acknowledges the result.

6. Processing locations and recognition data retention

The Supabase database currently serving users is in Singapore. Page recognition uses Alibaba Cloud’s Singapore endpoint. Inference may run in other regions outside mainland China. Under the provider’s standard policy, inference inputs and outputs are generally retained for up to 30 days, or longer if required by law or reasonably necessary for security. Content flagged by its safety systems may be retained for up to two years. The provider states that it will not train foundation models on business data without separate consent. API and database operational logs for the current service are retained for one day.