Privacy Policy

Vocaburn · com.coco.vocaburn

한국어 버전

Vocaburn (the "App") respects your privacy and handles data as described below.

The conditions for using the service are set out in the separate Terms of Use.

Summary: Signing in is optional. Without an account, your word list and study history stay only on your device. If you sign in with Google (or, on iOS, with Apple), your word list is stored on the developer's server (Seoul, South Korea region) so it syncs across your devices, and you can delete your account and the server data linked to it at any time from Settings (see 5 for what remains in study groups and in processing records for synced words). The words, meanings and notes you enter are never sent to advertising or analytics services. However, if you use word auto-fill (on by default, can be turned off in Settings), the term you type is sent to the developer's server to look up its meaning whether or not you are signed in, and if the server has no result, it may be passed to Anthropic to generate meanings and examples (1-5). For words you sync while signed in, regardless of this setting, if the server has no result, the term may be passed to TypeSafe AI (word check), OpenAI or Anthropic (information generation) and Google Cloud Text-to-Speech (pronunciation audio) to prepare word information and pronunciation audio (1-5). If you join a study group, the messages and word sets you add are shown to members of the same group with your nickname (your test results are shown with your nickname to members of the same group only in groups where the owner keeps test results public, and to the owner regardless of that setting), and notifications about new messages and word sets are sent to your device (1-6; section 9). If you use AI Tutor (sign-in required), the word information, the messages you enter and the recent conversation are passed through the developer's server to OpenAI and TypeSafe AI to generate answers, and your conversation history is stored on your device and, while your AI Tutor pass is active, also in your account to sync your study history, including earlier conversations still on this device (kept for 90 days, 1-7). If you create a PDF in Make a test sheet (sign-in and AI Tutor pass required), the questions the App picked from your word list are sent to the developer's server to create the PDF; the server does not store them or pass them to AI or anywhere else (1-9).

1. Information we collect

1-1. Account information (when you sign in with Google or Apple)

1-2. Synced data (when you sign in)

When signed in, the following data is stored on the server and synced with other devices signed in to the same account.

ItemDetails
Word listWords, meanings, notes, example sentences, importance flag, per-word correct/incorrect counts, last studied time
Groups and collectionsNames, language tag, membership
Daily statsPer-day studied word count, correct/incorrect counts, goal completion
Study history (Lifetime users only)Per-attempt records (word, direction, result, level change, time). For other users, study history stays on the device only.
Device infoA random device ID generated by the App (not a hardware identifier), platform, app version, last sync time, Firebase Cloud Messaging device token (an identifier for receiving push notifications; used only to send study group notifications about new messages and new word sets and word-info-ready notifications — see 9)

1-3. Shared word sets (when you publish on the web version)

1-4. Automatically collected information

None of this includes the words, meanings or notes you enter.

1-5. Word auto-fill (whether or not you are signed in)

When you type a term while adding or editing a word, the App sends the following information to the developer's server (Supabase Edge Function) to fill in meanings, notes and examples automatically. Generating or checking for updates to AI study notes on the word detail screen works the same way. This feature is on by default and can be turned off in Settings > Auto-fill words; when it is off, these lookup requests from the App (Information sent and AI generation in the table below) are not made. This setting applies only to requests the App sends; downloading pronunciation audio (Pronunciation audio in the table below) and server processing of words you sync while signed in (last row of the table) happen regardless of this setting.

ItemDetails
Information sentThe term you type and a random device ID generated by the App (the server uses it for this request only to check the format and does not store it)
AI generationIf the server has no stored result, the term alone may be passed to the Anthropic API to generate study information such as meanings, examples and pronunciation. Your device ID and account information are not passed on.
Stored on the serverLookup requests from the App store only the term and the generated result on the server, in a cache shared by all users. The cache itself is not linked to any device ID or account (for processing records of synced words, see the last row of the table). Word auto-fill has no daily usage limit, so no per-device lookup count is recorded.
Pronunciation audioRegardless of the auto-fill setting, when a user with an active AI Tutor pass plays the pronunciation of an English word, the audio file is downloaded from the server (if there is no file, your device's speech engine reads it). The request URL contains the term (for example sentence audio, a hash of the sentence) but not your device ID. The audio files are shared files the server prepares in advance from terms and AI-generated example sentences; when creating them, only the text to be read is passed to Google Cloud Text-to-Speech (see 6). For other users (including users who only have Lifetime), pronunciation is read by the device's speech engine.
Server processing of synced words (when signed in)Regardless of the auto-fill setting, words synced to the server while you are signed in that have no AI study notes (words in an English word group or not assigned to any group) may be added to a processing queue on the server if the server has no stored result. When a newly synced word is added to the queue, the term is recorded in a processing record together with the ID of the account that first added it to the queue (see 5). The server ① sends the term to TypeSafe AI's Jev model to check whether it is a correctly spelled English word or a typo (the same route as AI Tutor), ② if it is a valid word, may pass the term to OpenAI, or the operator may prepare it using Anthropic's AI (Claude) (including checking terms whose validity is unclear), to generate study information such as meanings and examples, ③ creates pronunciation audio files from the term and the generated example sentences (Pronunciation audio above), and ④ when processing is finished (including when the word could not be recognized), sends a notification to users who saved that word without a meaning or AI study notes (see 9). The results are stored in the shared cache described under "Stored on the server" above. Only the term (for audio, the text to be read) is passed to TypeSafe AI, OpenAI and Google Cloud Text-to-Speech; your account ID, device ID and email are not passed on.

1-6. Study groups (when you sign in)

When you create or join a study group, the following information is stored on the server.

ItemDetails
GroupGroup name and description, invite code, invite password (stored only as an irreversible hash)
MembershipRole (owner, editor, member), join time, last-read message position, per-group notification setting, removal records (the owner who removed the member and when)
ChatMessage text, sender and time; notices announcing joins, leaves, removals, added or updated word sets and test results (including nicknames)
Group word setsTitle, description and language of word sets uploaded to the group, a copy of the words (words, meanings, notes, examples, pronunciations and AI study notes) and the uploader
Group test resultsMode, direction, correct answers, number of questions, time spent, missed words and when the test was taken. Group tests are not recorded in the study history or daily stats in 1-2.
Reports and blocksReports (the reported message or user and the reason) and the list of users you have blocked
Real-time delivery logSignals for new messages and notices, deleted messages, reads, removals, dissolved groups and changes to group settings, member roles and test settings left in Supabase Realtime while chat is delivered in real time: their times, message numbers, the account IDs of members who read messages or were removed, and group IDs (deleted within about 4 days). Message text and the content of notices (including nicknames) do not remain in this log.
Notification sending queueNotification text for new messages and word sets (the group name and the sender's nickname with the beginning of the message or the word set title)
Rate-limit countersPer-account and per-group counts used to prevent abuse of chat, password attempts, reports and similar actions

1-7. AI Tutor (when you sign in)

AI Tutor provides chat on the word detail screen, "Why was I wrong?" explanations after a wrong answer in study (dictation and multiple choice), and feedback on sentences you write in the chat's sentence-writing mode. You must be signed in to use it. There is a daily usage limit (2 turns free, 30 turns while your AI Tutor access period is active; a wrong-answer explanation counts as 0.5 turn). Answers are generated automatically by AI and may be inaccurate (this is also always shown in the App).

ItemDetails
Information sentWith each request, your account (sign-in token), the current word's term, meanings, pronunciation, examples and part of its AI study notes (definitions, synonyms, antonyms and collocations), the message you enter and up to the last 8 turns of the conversation about that word are sent to the developer's server (Supabase Edge Function). For a wrong-answer explanation, the question direction, the answer you gave and the multiple-choice options are also sent; in sentence-writing mode, the English sentence you wrote is also sent.
AI generationTo generate answers, the word information, message, recent conversation, wrong answer and sentence above are passed to the OpenAI API. To check whether a message is a word-study question, classify the type of wrong answer and score sentences, the term and meanings, the current message (or wrong answer or sentence) and the generated AI answer are passed to TypeSafe AI's Jev model. Your account ID, device ID and email are not passed to either.
Stored on the serverWhile your AI Tutor pass is active, your conversations, including earlier ones on this device, are stored in your account (see "Where it is stored" below). ① Account ID, date (UTC) and number of turns used — to apply the daily usage limit. ② AI Tutor access period — an expiry time in your account profile. Server logs do not contain message text or words, only the mode, response type and classification probabilities.
Where it is storedYour conversation history (messages, answers, wrong-answer labels, sentence scores and times for each word) is stored on your device. While your AI Tutor pass is active, it is also stored in your account to sync your study history; earlier conversations still on this device are stored in your account at that time as well. Conversations stored in your account are synced to your other devices signed in to the same account and are deleted by a weekly cleanup after 90 days. While you have no pass, conversations are not stored on the server and stay only on your device.
ReportsIf you report an AI answer by long-pressing it, your account ID, the mode, the term, your preceding message, that AI answer, the report reason and the time are stored on the server for review by the developer and are not shown to other users.

1-8. Scan a book (camera, Android and iOS)

The App asks for camera permission only in "Scan a book" when adding words, to recognize text in a book and suggest words from it.

ItemDetails
Camera imageThe App does not save the camera image or send it anywhere, and text recognition is done on your device.
What is savedOnly the words you choose to add are saved to your word list (the sentence a word came from is shown only on the confirmation screen before adding and is not saved), and word auto-fill (1-5) applies to the words you add.
Text recognition engineIt depends on the operating system. On Android, the App uses Google ML Kit, whose recognition model is downloaded to your device by Google Play services. On iOS, text is recognized on your device with Apple's Vision framework instead of ML Kit.
ML Kit usage statistics (Android)As described in Google's ML Kit data disclosure, ML Kit on Android sends Google usage statistics for diagnostics and analytics, such as device information (manufacturer, model and OS version), the app's package name and version, a per-installation identifier, performance metrics such as processing time, settings such as image format and resolution, and error codes.

1-9. Test sheet PDF (when you sign in)

Make a test sheet on the Study tab turns words from your word list into a test sheet (questions and answer key) as a PDF. You must be signed in and have an active AI Tutor pass to use it.

ItemDetails
Information sentWhen you create a PDF, your account (sign-in token), the test sheet title, language and output options (answer key, name field and what to print), and the questions the App picked from your word list (terms and meanings, multiple-choice options and answers) are sent to the developer's server (Supabase Edge Function). If you also create an answer key, the same questions are sent once more for it.
Server processingThe server checks your AI Tutor access period (the expiry time in your account profile, 1-7), then creates the PDF and returns it right away. The questions are used only to create the PDF and are not passed to AI or anywhere else.
Stored on the serverThe questions received and the PDF created are not stored. Server logs do not contain the test sheet title or question content, only processing information such as the number of questions, sets and pages, file size, processing time and error type.
The PDF you receiveIt is handed to your device's share sheet as a temporary file and goes only to the app you choose (for printing, saving files, messaging and so on). The App does not send the PDF anywhere else.

2. Information we do not collect

3. Data stored only on your device

If you are not signed in, this data cannot be recovered when you delete the App or switch devices (it may be included in your Google account backup if Android Auto Backup is on, and on iOS in your iCloud backup if iCloud Backup is on or in a backup of the device made on a computer). Lifetime users can use CSV export in Settings to back it up to a location they choose; the App does not access exported files.

4. How we use the data

5. Retention and deletion

If you can no longer access the App, you can request account deletion by email. See Delete Your Account and Data for details.

6. Third-party services

The App uses the following services, each of which may process limited data on its own servers:

You can reset your advertising ID or opt out of personalized ads in your device settings. On iOS, the advertising identifier (IDFA) is used only if you allow this app to track in your device's tracking settings (Settings > Privacy & Security > Tracking); if you do not, ads are shown without the IDFA. See the Google Privacy Policy for how Google services process data. We do not sell your personal information to third parties.

7. International transfers

The Supabase server that stores synced and shared data is located in South Korea (Seoul). Google Sign-In, AdMob, Firebase (including delivery of study group notifications), Google ML Kit usage statistics (Android, 1-8) and Google Cloud Text-to-Speech run on Google's global infrastructure, and Sign in with Apple, App Store purchases and APNs push delivery (iOS) run on Apple's infrastructure, so your data may be processed on servers outside your country. Terms passed to Anthropic and OpenAI by word auto-fill and server processing of synced words (1-5), and information passed to OpenAI by AI Tutor, may be processed on servers in the United States, and information passed to TypeSafe AI by AI Tutor and by the check of synced words (1-5) may be processed on servers outside Korea.

8. Security

All communication between the App and the server is encrypted (HTTPS). Server data is protected by row-level access control so that only the signed-in owner can access their synced data, and shared word sets are exposed only according to the visibility you chose. The server returns study group data only after checking that the requesting user is a member of that group (the information shown in group search and invite-code previews can be seen by any signed-in user, as described in 1-6, and for real-time chat, membership is checked when the connection is opened), and invite passwords are stored only as irreversible hashes.

9. Notifications

Study reminders, word alerts and the autoplay media notification are local notifications scheduled and shown on your device. The push notifications sent from the server are study group notifications about new messages and new word sets, and notifications telling you whether information for words you saved without a meaning or AI study notes is ready (or could not be recognized) (at most twice a day), delivered through Firebase Cloud Messaging. You are not notified about messages you send or word sets you upload, groups you have muted, or messages and word sets from users you have blocked. You can mute group notifications for each group on the group screen, and turn off all notifications at any time in your device settings.

10. Your rights

You can view, edit and delete your data in the App (editing your word list, CSV export (Lifetime), account deletion), stop syncing by signing out, delete your messages, leave groups, mute group notifications, and block and report users in study groups, and stop auto-fill and AI study notes lookup requests by turning off word auto-fill in Settings (pronunciation audio for English words is still downloaded from the server when a user with an active AI Tutor pass plays it, and server processing of words you sync while signed in still happens, regardless of this setting — see 1-5). For any other access, correction or deletion request, contact us at the email below and we will respond without undue delay.

11. Children's privacy

The App does not knowingly collect personal information from children under 14. If we learn that such information has been collected, we will delete it promptly.

12. Changes to this policy

If this policy changes, we will notify you through this page and an App update.

13. Contact

For privacy questions or deletion requests, please contact:

Email: philleeran@gmail.com

Effective date: 2026-10-02