Privacy Policy

Pico Picks: Vocab Learning · Last updated: 29 October 2026

This Privacy Policy explains what personal data the Pico Picks iOS app and the picopicks.com website collect, why they collect it, and what rights you have over your data under the EU General Data Protection Regulation (GDPR) and the French Data Protection Act (Loi Informatique et Libertés).

1. Who is responsible for your data

The data controller is:

• Tudual Lucas Huon, sole proprietor (auto-entrepreneur)
• 21 Dom Morice, 35000 Rennes, France
• SIREN: 999 501 703
• Contact: contact@rougegorge.app

2. What we collect

On your device. The iOS app stores the following locally using Apple's SwiftData framework. This data does not leave your device unless you explicitly share it, or unless it is synced through your own iCloud account as described below:

• The first name or nickname you entered during onboarding
• Your home language (as an ISO code, e.g. fr, es)
• Your declared CEFR level (A1 to C2) and learning preferences
• The vocabulary words you save, review or add to your library
• Your learning history, spaced-repetition stats and daily streaks
• Your last 20 translations in the translator (translation history, kept on the device only)

Short audio recordings made during pronunciation exercises are the exception: they are sent for scoring, as described below and in §3.

Synced through your own iCloud account. If you are signed in to iCloud, your learning data (saved words and lists, progress, learning history) is synchronised between your Apple devices through your own private iCloud database — a private CloudKit container that belongs to your Apple Account. This data is stored by Apple in your iCloud account, encrypted in transit and at rest, and is not accessible to us: we cannot read it, export it or delete it. Because it lives in iCloud and not only on the device, it survives uninstalling the app: it comes back if you reinstall Pico Picks while signed in to the same Apple Account, and it remains in your iCloud account until you delete it. You can delete or export it at any time from within the app (Settings › “Delete my data” / “Export my data”), and you can remove the app's iCloud data from iOS Settings › your name › iCloud › Manage Account Storage.

Sent to our backend server (rougegorge.app). When you look up a word, request a definition, use the translator, submit a pronunciation, or report AI-generated content, the app sends the following to our backend (our Cloudflare Workers and our rougegorge.app server):

• The word or short phrase you looked up
• Your CEFR level and home language (as context for the AI model to generate a level-appropriate definition)
• For the translator: the text you typed or dictated (up to 600 characters), the two languages, and your home language
• For spoken pronunciation: the word to read aloud
• For pronunciation scoring: a short audio sample and the target word
• For pronunciation tips: your pronunciation scores and the target text, not the audio
• For reports: the AI output you flagged, the reason you picked, and any optional free-form note (limited to 500 characters)

We do not send your name or your learning history, and the words you look up are not attached to a user account. To keep your saved words up to date, the app may resend the words you saved, with your level and language, to the providers listed in section 3. These requests carry a shared application API key, not a user identifier.

Automatic requests. Without any action from you, the app checks once per launch which content version our server serves, and may refresh up to 15 saved words per day.

Share extension. When you share text to Pico Picks from another app, the selected text is sent through our Cloudflare relay to Anthropic to build the card.

Pronunciation coach. After a pronunciation attempt, your scores and the target text are sent to OpenAI so Pico can phrase a coaching tip. The audio is not sent to OpenAI.

Photo lookup (iOS 27, Apple Intelligence devices). The photo is analysed on your device or on Apple's Private Cloud Compute and is never sent to Pico Picks servers. Only the recognised English word is then looked up, like a search.

Your permission to share data with third-party services. Nothing leaves your device before you give it. You give it by tapping the button of the first onboarding step (“Next”), under which two lines tell you that Pico Picks shares your texts and voice with third-party services for translations and explanations, and link to these Terms and this Policy. The providers, and what each receives, are listed in section 3. You can withdraw your permission at any time in Settings › About › Data sharing; Search then pauses.

Translator handled by Apple. On iOS 27 with Apple Intelligence turned on, translations between your home language and English may be done by Apple's Private Cloud Compute; when neither Apple Intelligence nor our backend can answer, Apple's translation on the device is used. In those cases the text is processed by Apple under its own privacy policy and is not sent to us.

Microphone and speech recognition. The app uses the microphone to let you practise pronouncing English words and to dictate text in the translator. Your voice is turned into text by Apple's speech recognition: when your iPhone supports it, this runs on the device; otherwise Apple processes the audio under its own privacy policy. Dictated text is then handled like typed text (see the translator above). We do not keep dictation audio.

One exception, for lesson downloads. When the app fetches downloadable lesson content, it calls a single endpoint, /v1/content/auth, and sends an installation identifier: a random UUID generated on your device at first launch. It is not the Apple advertising identifier, it is not linked to your name, your email or your Apple Account, and it changes if you uninstall and reinstall the app. We use it only to rate-limit access to lesson content and to prevent abuse. It is never attached to the words you look up, to your translations, to your pronunciation recordings, or to any of the AI requests described above. If you have an active subscription, the app sends the signed App Store record of your subscription instead of this identifier, so that the server can check it before unlocking the content.

Technical metadata. Like any web server, our backend logs standard connection metadata (IP address, user-agent, timestamp) when you make a request. These logs are retained for up to 30 days for security and debugging, then rotated out.

Anonymous usage statistics. So that we can tell which parts of the app are actually used, the app sends anonymous counters to our backend (/v1/events, Cloudflare Analytics Engine). Each event is the name of an action taken from a fixed list of names accepted by our server, which rejects anything else (for example: an onboarding step completed, a review session started, a word added, a paywall shown, a purchase started or cancelled, a free-tier limit reached), plus the following context and nothing else:

• the app version
• your App Store country (storefront)
• the language the app is displayed in
• your major iOS version (e.g. “27”)
• whether the device is an iPhone or an iPad
• whether you have an active subscription
• how long ago the app was installed, as a broad band (e.g. “the first day”, “the first week”, “more than a month”)

Once a day, the app also records that it was opened, which lets us count returning users as a number. These events carry no identifier of any kind: no account, no installation identifier, no advertising identifier, no device name, and your IP address is not stored with them. They carry no content: never a word you looked up, never a text you translated, never your name or your learning history. Some events carry your CEFR level, the language you learn from, short labels describing the action and one number (for example, how many cards you sorted). They cannot be linked to you, to each other, or to the rest of your data, and they are not used for advertising, profiling or tracking, in the app or anywhere else. Statistics start only after onboarding. They are on by default and you can switch them off at any time in the app: Settings › About › Anonymous usage stats.

3. Third parties we share data with

To provide the features of Pico Picks, the app sends certain requests, through our own relay, to the eight external services below. The app contains no third-party advertising or analytics SDK. No other third parties receive your data.

• OpenAI, LLC (United States) — writes the definitions, example sentences and pronunciation coaching tips you read in the app, and helps with translations, via GPT models. Receives the word or text and the context outlined in §2 (your CEFR level and home language); for coaching tips, your pronunciation scores and the target text, never the audio.
• Anthropic, PBC (United States) — writes definitions, example sentences and enrichments via Claude models, and builds the card from text you share through the Share extension. Receives the word or text and the same context.
• Google LLC (United States), through Google Cloud Translation — translates the text you type or dictate in the translator. Receives that text and the language pair, with no user identifier.
• Microsoft Azure AI Speech (European Union region) — evaluates pronunciation when you record yourself saying a target word. Receives a short audio sample and the target word. The audio is deleted after scoring and is not stored by Microsoft for training.
• Mistral AI (France / European Union) — generates the spoken audio of lesson dialogues. Receives only the dialogue text to voice.
• Apple Inc. (United States) — distributes the App through the App Store and, if you subscribe, processes the subscription via StoreKit (a transaction identifier and your App Store locale; we do not receive your payment card details). Apple operates iCloud and the private CloudKit database in which your learning data is synced (see §2). For photo lookup, Apple analyses the photo on your device or on its Private Cloud Compute, and the same applies to the Apple Intelligence translator, translation on the device and speech recognition when it cannot run on your iPhone. This data is governed by Apple's own privacy policy and is not sent to us.
• Cloudflare, Inc. (United States, edge data centers worldwide) — relays and caches the requests above through our backend (Cloudflare Workers and the Workers KV response cache). Acts as a technical intermediary. Cache entries are keyed by an SHA-256 hash of the search term, your CEFR level and your home language only, with no user identifier. Cloudflare also stores the anonymous usage counters described in §2 (Analytics Engine), which carry no identifier and no IP address.
• Hostinger International Ltd (Cyprus / European Union) — hosts the rougegorge.app server, which relays audio (text-to-speech, pronunciation scoring) and receives content reports. Acts as a technical provider and does not otherwise access the data.

Apple, OpenAI, Anthropic, Google and Cloudflare are located in the United States. Transfers outside the European Economic Area are covered by the European Commission's Standard Contractual Clauses (SCC) as published in Decision 2021/914 (for Cloudflare, under its Customer Data Processing Addendum). Microsoft Azure AI Speech is used in a European Union region. Mistral AI and Hostinger are located in the European Union.

4. How long we keep data

• On-device data: stored until you delete it from within Settings or uninstall the app. We have no access to it.
• Data synced to your iCloud account: kept in your private iCloud database for as long as you leave it there. Uninstalling the app does not delete it, and neither do we — we have no access to it. Delete it from within the app (Settings) or from your iOS iCloud settings.
• Backend cache of definitions: up to 90 days from last use, then evicted. The cache is keyed by an SHA-256 hash of the lookup (search term + CEFR level + home language) and is not tied to individual users.
• Installation identifier: kept on your device until you uninstall the app. On our side it is used only to rate-limit lesson downloads and is not attached to a user profile or to your lookups.
• Content reports: up to 90 days, then deleted. Used only to improve AI output quality.
• Access logs: up to 30 days for security and debugging.
• Anonymous usage counters: up to 3 months, then deleted automatically by Cloudflare Analytics Engine. They are aggregated figures with no identifier, so they cannot be traced back to a person, including by us.
• Pronunciation audio: deleted immediately after the score is returned to the app. Not persisted.

5. Your rights

Under the GDPR you have the right to:

• access the personal data we hold about you;
• rectify inaccurate data;
• request the deletion of your data (data held only on your device is removed when you delete it in Settings or uninstall the app; data synced to iCloud must be deleted from within the app or from your iCloud settings, since we cannot reach it; for backend data, write to us);
• restrict or object to processing;
• obtain a portable copy of your data (the app can export your library and progress);
• withdraw any consent you previously gave (the notice under the Next button of the first onboarding step, and Settings › About › Data sharing);
• lodge a complaint with the French supervisory authority, the CNIL (www.cnil.fr).

To exercise any of these rights, write to contact@rougegorge.app. We respond within one month.

6. Legal basis for processing

• On-device storage of your profile, vocabulary and learning history: execution of the contract to provide you with the app.
• Synchronisation to your private iCloud database: execution of the contract to provide you with the app across your devices.
• Backend relay to OpenAI / Anthropic / Google / Mistral / Microsoft: execution of the contract and your explicit consent, given by tapping Next on the first onboarding step, under the notice that your texts and voice are shared with the third-party services listed in section 3. You can withdraw it at any time in Settings › About › Data sharing.
• Installation identifier sent to /v1/content/auth: our legitimate interest in rate-limiting the service and preventing abuse.
• Content reports you choose to send: your consent at the moment of sending.
• Anonymous usage counters: our legitimate interest in knowing which features are used and in improving the app, with no identifier, no content and no profiling; they are on by default after onboarding and you can object at any time with the switch in Settings › About › Anonymous usage stats.
• Security logs: our legitimate interest in protecting the service.

7. Children

Pico Picks is not intended for children under 13. Pico Picks does not ask for your age and does not collect a date of birth. Our Terms require users to be 13 or older, and the app is rated 13+ on the App Store. If you believe a child under 13 has used the app, write to us and we will delete any associated data.

8. Security

All traffic between the app and the backend is encrypted over HTTPS using TLS 1.2 or 1.3, with the standard system certificate validation provided by iOS. On-device data is protected by iOS sandboxing and benefits from Apple's hardware-backed encryption when your device is locked. Data synced to iCloud is encrypted by Apple in transit and at rest within your private CloudKit container.

9. Changes to this policy

If we change this policy we will update the date at the top and, for material changes, notify you in the app on next launch. Continuing to use Pico Picks after a material change means you accept the new policy.

10. Contact

Questions about this policy or your data: contact@rougegorge.app.

Pico Picks: Vocab Learning · Tudual Lucas Huon, auto-entrepreneur · 21 Dom Morice, 35000 Rennes, France · SIREN 999 501 703