Angloteka Privacy Policy
Effective date: 2026-07-27 Last updated: 2026-09-23
In short
- In the iPhone app we create no accounts and ask for no email or phone number. The web version does need an email address — only that, and only so paid access can be found again on another device; see section 3.2. The app asks for a first name once, only to greet you, and that name stays on your device — it never reaches us. Giving it is optional
- Your cards stay on your device
- A photo is sent to our server and to a recognition service only while it is being processed. We do not store it on our servers. The recognition provider automatically deletes inputs within 30 days — longer only if their automated systems flag the request (see section 4)
- In the «Разбери конструкцию» (Break down the construction) lessons, new explanations are produced by the same provider. What goes out is the text of the lesson — the topic, the English phrase with its translation and, when explaining a mistake, the exercise question and the number of the option you picked. None of your cards, photos or progress is part of that (see sections 3.2 and 4)
- We use no advertising, analytics or tracking — not a single advertising or analytics SDK
- We do not sell or share your data for anyone else's purposes
1. Who processes your data
Controller: Dmitrii Tiurnikov, individual developer Address: Giorgi Brtskinvale Street 105-107, Batumi 6010, Georgia Contact: privacy@angloteka.app
2. Who this app is for
Angloteka (formerly Англотека) is built for anyone learning English: an adult studying on their own, and a child studying together with an adult. The interface is deliberately simple and speaks informally to whoever is studying — in part because that may be a child.
The account belongs to an adult, and an adult makes any purchase. The two decisions the app asks to be made explicitly are addressed to that adult: permission to send a photograph for recognition, and taking out a subscription. Both screens say so on their face.
What we receive from the person studying: nothing that identifies them. That holds equally for an adult and for a child. The app does not ask for age, gender, email, phone or any other detail — about the person studying or about anyone else — and creates no account. On first launch it asks once for the name of the person who will be learning, solely to greet them on the home screen; that name is kept in the app's local storage, never reaches us, and the step can be skipped, or the name changed or cleared later in Settings — see section 3.3.
The only thing tied to the device that does reach us is a random installation identifier (see 3.2). It is not derived from anybody's data, is never matched against anything else, and serves only to count recognitions: without it we could neither honour the limits nor tell ordinary use from abuse. Records carrying it are deleted after 90 days, and a button in Settings erases them immediately.
A parent may withdraw permission to send photographs at any time, erase the records tied to their device, and write to privacy@angloteka.app about any of this.
3. What data is involved
3.1. Stays on your device only
Cards, translations, learning progress, statistics and the greeting name live in the app's local storage. None of it reaches our servers and we cannot see it. Deleting the app deletes it.
3.2. Sent to our server
| Data | Why | Retention |
|---|---|---|
| The photo you take or choose | To recognise the English words in it | Not stored on our servers — held in memory during processing and discarded once the response is returned. The recognition provider, Anthropic, deletes it within 30 days; if their automated systems flag the image as violating their rules, it may be kept for up to 2 years |
| Device identifier (random, generated by the app) | To limit how many recognitions can be requested and prevent abuse | 90 days |
| The IP address the request came from | To limit recognition requests per address: the device identifier is supplied by the app itself, so it is not sufficient on its own | Not written to the database. Held in the server's memory and dropped no later than 24 hours after the last request |
| The text of a construction breakdown: the lesson topic, the English phrase with its translation and the list of phrases already covered so the breakdown is not repeated; when explaining a mistake, also the exercise question, the answer options and the number of the one you picked | To produce a new construction breakdown, or to explain why an answer in an exercise is wrong | Not stored on our servers — like a photo, held in memory during processing and discarded once the response is returned. At Anthropic the retention is the same as stated above for photos |
| Request time, amount of text processed, cost of the recognition call | Cost control and detecting abnormal load | 90 days |
| Web version only: your email address | Signing in to the web version and tying paid access to you. Without it, paid access would be lost whenever browser data is cleared: unlike the app, a browser has no secure storage that survives clearing | For as long as you have an account. Deleted on request to privacy@angloteka.app |
| Web version only: the notification delivery address issued by your browser | Lesson reminders. A browser cannot show a reminder while the app is closed, so our server has to wake it | For as long as reminders are on. Turn them off and the record is deleted at once |
What exactly goes out in a construction breakdown. It is the text of the lesson itself — ours, not yours: the name of the construction, the English phrase, its translation and the exercise question. The only part that comes from you is the number of the option you picked. No cards, no photos, no greeting name and no study statistics are included; none of that ever leaves the device at all (section 3.1).
The device identifier is a random string the app generates on first launch. It contains nothing about you, is not derived from any hardware number, and is not linked to your identity — the app has no accounts.
It is kept in the device's secure storage (Keychain) and therefore survives deleting the app. That is deliberate: this identifier is how we count the recognitions you have already used, and without it the limit could be cleared by reinstalling. You can delete the records tied to it at any time from Settings.
3.2a. Reminders in the web version: what we know and what we do not
The reminder is composed by the app on your device, from your own progress, which never reaches us (section 3.1). All our server is told is when to wake you, plus the delivery address issued by your browser. The server never receives or stores the text of the notification: it sends an empty signal, and what to show is decided by the app on your device.
The delivery address is a URL issued by your browser's notification service (Google, Mozilla or Apple, depending on the browser). It points at an installed app, not at you, and we do not link it to your email or to the device identifier.
3.3. Never collected
We do not collect your phone number, postal address, location, contacts, browsing history, search history, health data, or advertising identifiers (IDFA). On first launch the app asks once for the name of the person who will be learning, solely to greet them on the home screen; that name is kept in the app's local storage, is never sent to us, and the step can be skipped or the name changed or cleared later in Settings. If a photo carries a geotag, the app strips it before sending. The network (IP) address a request comes from is processed — see section 3.2. The app does not request App Tracking Transparency permission, because it does not track. Email is not processed in the iPhone app at all; in the web version it is used to sign in — see section 3.2.
4. Who else receives data
Anthropic PBC (United States) — the processor that recognises words in the image and prepares the breakdowns in the «Разбери конструкцию» lessons. Two things are sent there: the photo (to recognise the words) and the text of a construction breakdown (to compose a new breakdown or to explain a mistake in an exercise). Data is sent for the duration of processing. Anthropic deletes inputs and outputs within 30 days of receipt, with one exception: if their automated systems flag a request as violating their usage policy, the data may be kept for up to 2 years, and the classification scores for up to 7 years.
Nothing is sent without your permission. Before the first upload the app shows a screen listing exactly what is sent, to whom, and for how long, and asks for your permission. Until you grant it, no photo leaves the device. You can withdraw the permission at any time from the Settings screen, in the data and privacy section — the app interface is in Russian, and the row is «Отправка фото на распознавание». After withdrawal the app asks again the next time you want to photograph words. Withdrawal applies to future uploads; photos already sent are deleted on the schedule above.
Construction breakdowns are not gated by a separate consent screen — unlike photos, and the difference is not about convenience but about what is sent. Recognition sends out a photograph you took: content you created. A breakdown sends out our lesson text and the number of the option you tapped — you put none of your own content into it. So instead of a consent screen, the lesson screen permanently shows a line naming the provider, next to an expandable note carrying the same list as section 3.2: what is sent, what is not, to whom, and for how long. Staying silent about this transfer would be wrong; asking permission to send our own text would have nothing to ask about.
This processing is governed by Anthropic's Data Processing Addendum, which is incorporated into their Commercial Terms of Service and includes EU Standard Contractual Clauses for international transfers. Anthropic acts as a processor, acts on the controller's instructions, and does not use commercial product data to train models.
Railway Corp. — hosts the backend.
Apple — handles in-app purchases. Payment details are processed by Apple; the app never receives or stores them.
Dodo Payments — handles subscription payments in the web version. Dodo acts as the merchant of record: the sale contract is with them, and they issue the invoice and remit taxes. They receive your email address — the same one you sign in with — because without it we could not match the payment to your account. Card details are entered on their page; we never receive or store them. In return we receive the subscription identifier and its state: paid, cancelled or expired. On your card statement the charge appears as DODOPAY_ANGLOTEKA.
We do not sell personal data and do not disclose it to third parties for their own purposes, including advertising.
5. Legal bases (EU/EEA and UK users)
| Processing | Basis |
|---|---|
| Recognising words in your photo | Performance of a contract (Art. 6(1)(b) GDPR) — this is the service you asked for |
| Construction breakdowns and explanations of exercise mistakes | Performance of a contract (Art. 6(1)(b) GDPR) — the breakdown is produced because you asked for it, and is the service requested |
| Device identifier and request log | Legitimate interests (Art. 6(1)(f) GDPR) — protecting the service from abuse and uncontrolled cost |
6. Your rights
You may request access to your data, correction or erasure, restriction of processing, object to processing, and receive your data in a portable format. For EU/EEA and UK users these rights come from the GDPR; for California residents, from the CCPA/CPRA.
How to exercise them. In the app: Settings → Data and privacy → "Delete my data from the server". Records are deleted immediately; there is no need to contact us. If the button is unavailable, email privacy@angloteka.app with the device identifier shown on the same screen.
We respond within 30 days, at no charge.
You may also lodge a complaint with your national data protection authority.
7. Notice for California residents
In the preceding 12 months we collected the category of "identifiers" (a device identifier) and "electronic information" (a photo, transiently; not stored on our side, deleted by the recognition provider within 30 days, or within up to 2 years if their automated check is triggered).
We have not sold and not shared personal information as those terms are defined by the CCPA/CPRA, and we do not use personal information for cross-context behavioural advertising. You will not be discriminated against for exercising your rights.
8. Security
Data in transit is protected with HTTPS. The administrative dashboard is password-protected. We do not store photos on our servers, so a breach of our storage has no photos to expose. That does not mean the photo is protected everywhere: while the recognition provider still holds it (see section 3.2), it is covered by Anthropic's safeguards, not ours.
No system is perfectly secure, but we aim to hold as little data as possible — that is the primary safeguard.
9. Changes
If this policy changes materially we will update the date at the top. The current version is always available at https://angloteka.app/privacy-en.
10. Contact
Questions and requests: privacy@angloteka.app