Privacy Policy for the PikaBear app
In force from: 16 September 2026
Last changed: 22 September 2026
This is a translation of the Slovenian original. The Slovenian text is the reference version.
This policy explains what data the PikaBear app processes, why, for how long, and what rights you have. It is written so that it can be read to the end - because the app is intended for children, this is part of our responsibility, not a courtesy.
In short: the app has no accounts, no sign-in, no ads, no analytics and no tracking. A child's progress stays on the phone. The only thing that leaves the device is what the child says out loud in the Talk and Voice translate modes - and only if a parent allows it.
1. Who processes the data
| Controller | Credo.app računalniško programiranje Petra Šuligoj s.p. |
| Address | Polana 49, 3273 Jurklošter, Slovenija |
| info@pikabear.app | |
| Registration number | 7586132000 |
We have not appointed a data protection officer (DPO), because the conditions of Article 37 GDPR are not met. For any question about data, write to the email address above; we answer within one month, as the law requires.
2. Who the app is for
PikaBear is a learning app for foreign languages. It has four groups of users: children (4-8 years), teenagers, adults and business users.
Because the app is also intended for children, stricter rules apply. Under the Slovenian Personal Data Protection Act (ZVOP-2, Article 8), a child's consent for information society services is valid from the completed age of 15. For younger children, consent must be given or authorised by a parent or guardian.
For users in other countries, the age limit set by the law of that country applies (in EU countries it is between 13 and 16 years). Where that limit is higher than 15, a parent's consent is needed up to that age.
That is why the app asks an adult on first launch, before anything is sent from the device.
3. What data is processed
3.1 What stays on the device and never leaves
| Data | When it is created |
|---|---|
| Chosen language and user group | on a tap |
| Progress: points, daily streak, daily goal | during play |
| Words the user captures with a photo | when the word is saved |
This data is stored in the app's memory on your device (localStorage). We
have no access to it. We cannot see it, we do not transfer it, and we cannot
delete it remotely - you delete it yourself, with the Delete my data button in
the app or by uninstalling the app.
Cloud backup of this data is switched off. Android can copy app data to the phone owner's Google Drive and move it to a new phone; this app does not allow that, because it would be a transfer of a child's data to a third party. The price of this is that progress does not move to a new phone - that is deliberate.
3.2 What leaves the device - only in the conversation modes
In the Talk and Voice translate modes, this happens:
- The microphone listens to what the person says.
- Speech recognition is done by your device or by Google. For this, Android uses its own speech recognition service; as a rule it sends the audio to Google, unless an offline recognition pack is installed on the device. We have no influence over this - it is a setting of your phone.
- The recognised text (not the voice recording) is sent to our server, which sends it to Anthropic, the company that runs the language model.
- The answer comes back and is displayed. Neither the recording nor the text is stored anywhere on our side.
What this means in practice:
| Data | Where it goes | Stored by us |
|---|---|---|
| Voice recording | speech recognition on the device / Google | no |
| Transcript of what was said | our server → Anthropic | no |
| The child's name, if they say it | in the transcript, like everything else | no |
| IP address | recorded by our server provider | in the provider's logs |
| Random device tag | our server | only the total amount of usage, for two days |
About the device tag. On the first conversation, the app creates a random number on your device and sends it with the request. It serves one purpose only: so that daily usage is counted per device and one user does not use up the daily capacity for everyone else.
What this tag is not matters more than what it is. It is not a device identifier, not an advertising identifier and not a sign-in. It is created in the app, it is random, it says nothing about you or the device, it is not linked to anything else, it is passed to no one except our server, and the Delete my data button deletes it. Next to it we keep only the total amount of usage in dollars, and that record is deleted after two days.
Important about the name. In the Talk mode the bear asks the child what their name is. If the child says the name, it goes into the transcript and with it to the cloud. We store it nowhere and link it to nothing - but it has been sent. If you do not want that, the question does not have to be answered; the conversation continues also with the Next button.
3.3 What the app does not do
We list this because for children's apps it is the most important part:
- it does not use analytics (no Firebase, no Google Analytics, nothing similar),
- it does not show ads and contains no advertising libraries,
- it does not use trackers or advertising identifiers (no AAID, no device ID, no device fingerprint),
- it has no user accounts and no sign-in,
- it does not determine location,
- it does not access contacts, messages, the photo gallery or files on the device - in Words from everyday life the app does not read the gallery; it receives a single picture that the user takes or selects themselves, and what happens to it is described in section 3.4,
- it does not sell or share data for marketing - to no one, ever,
- it does not use the voice to recognise a person (voice is not biometric data within the meaning of Article 9 GDPR, because we do not identify anyone with it).
3.4 The photograph in Words from everyday life
In this mode the user photographs a menu, a sign or packaging, and the app extracts words for learning from the picture.
The picture does not leave the device. The text is read on the device; everything needed for this is inside the app and no internet is needed for reading. Until 20 September 2026 the app downloaded the reading components from the jsDelivr network on first use of the camera, which saw the IP address when doing so - this is no longer the case, and that network is therefore no longer listed among recipients.
The app does not read the gallery. It receives a single file, the one that the user takes or selects in the system picker. It does not request permission to access photos and does not have it.
What stays on the device. When the user photographs with the camera, Android writes the picture to this app's folder. The app deletes such pictures at the next launch; so they remain at most until then, are accessible only to this app, and go with it if you uninstall it. They are not sent anywhere and are not stored anywhere on our side.
What is stored. Only the words the user selects and saves themselves - and these stay on the device, like everything else in section 3.1.
3.5 Subscription
The app is free to download. Talk and Voice translate are free for up to twenty questions per device; whoever wants to talk more can buy a subscription. Everything else - games, exercises, alphabet, numbers, vocabulary, the bear's voice and words from a photo - stays free and unlimited.
Payment is made by Google Play, not by us. We do not see and cannot see card numbers, billing addresses or other payment data; Google processes this data as the seller under its own terms.
What we receive. On a purchase, Google Play issues us a purchase token - a long random string that says that a subscription exists. The app sends it to our server, which asks Google a single question: is this subscription paid, and until when. The token contains no name, no email address and no Google account, and it cannot be used to find out who the buyer is.
Why we store it at all. To count how many conversations the subscription has already had this month. The record is a single one - the token and a number - and it disappears after seventy days at the latest. Without it, one subscription on three devices would mean three times the use.
Purchasing is for parents. Before the prices there is an arithmetic task that a small child cannot solve; the subscription screen is in the settings and not among the games.
Cancellation is done in Google Play, where you bought the subscription. The subscription is valid until the day to which it is paid.
4. Why we process data and on what legal basis
| Purpose | Data | Legal basis (GDPR) |
|---|---|---|
| Running the games and saving progress | data on the device | no processing on our side - the data does not leave the device |
| Conversation with the language model and translation | transcript of what was said | consent (Article 6(1)(a)); for children, a parent's consent (Article 8 GDPR, Article 8 ZVOP-2) |
| Protecting the server from abuse | IP address | legitimate interest (Article 6(1)(f)) - so that a publicly accessible service is not abused |
You can withdraw consent at any time under ⚙ Privacy and data, where you also switch it on. By default it is switched off: until a parent switches it on, nothing leaves the device. Withdrawal does not affect the lawfulness of processing before withdrawal. After withdrawal the two conversation modes do not work; all games, exercises and vocabulary keep working, because they do not need the cloud.
5. Who the data is passed to
To no one except these, and each only for what is described above:
| Who | What it receives | Where it is | Role |
|---|---|---|---|
| Cloudflare, Inc. | IP address, content of the request | USA / network worldwide | processor (runs our server) |
| Anthropic PBC | transcript of what was said | USA | processor (language model) |
| Google LLC | voice recording during recognition, download of language models | USA | independent controller for Android services |
| Google LLC (Google Play) | payment data when buying a subscription - we do not see it | USA / EU | seller of the subscription, independent controller |
Reading text from a photograph sends nothing anywhere - details in section 3.4.
With a subscription the path goes in the other direction: Google receives the payment data and we do not; from Google we receive only the answer whether the subscription is paid. Details in section 3.5.
With Cloudflare and Anthropic we have concluded a data processing agreement (DPA) under Article 28 GDPR.
Transfer to the USA. Cloudflare and Anthropic are in the United States. The transfer takes place on the basis of the European Commission's standard contractual clauses or the EU-US Data Privacy Framework, where the provider is part of it. Despite these safeguards, the level of protection in the USA is not necessarily the same as in Europe - we say this plainly, because for consent it is essential information.
6. How long data is kept
| Data | How long |
|---|---|
| Progress, words, settings on the device | until you delete them yourself |
| Transcript of what was said | not stored - exists only during the request |
| IP address in Cloudflare logs | under the provider's rules, as a rule up to 7 days |
| Daily usage record on the server | 2 days; contains only the total in dollars and a random device tag, without the content of conversations |
| Subscription conversation counter on the server | up to 70 days; contains only the purchase token and a number, without name and without the content of conversations |
7. Your rights
Under the GDPR (Articles 15-22) you have the right:
- of access - to find out what data we process;
- to rectification of inaccurate data;
- to erasure ("right to be forgotten");
- to restriction of processing;
- to data portability;
- to object to processing based on legitimate interest;
- to withdraw consent at any time.
How to exercise them in practice. Because the app has no accounts, we cannot recognise you and cannot look up your data - because we do not have it. Therefore:
- data on the device you delete yourself: in the app under ⚙ Privacy and data → Delete my data, or by uninstalling the app. Deletion removes progress, points, saved words, settings, the device tag and the consent itself - after it the app asks you again on the next launch;
- transcripts of what was said cannot be deleted, because they are not stored - there is nothing to delete;
- for everything else write to info@pikabear.app.
Right to lodge a complaint. If you believe we process data unlawfully, you can complain to the supervisory authority:
Information Commissioner of the Republic of Slovenia Dunajska cesta 22, 1000 Ljubljana gp.ip@ip-rs.si · www.ip-rs.si
If you live in another country, you can also complain to the data protection supervisory authority in your own country.
8. Security
- The access key to the language model is stored as a secret on the server and is not in the app - it cannot be obtained from an installed app.
- All traffic between the app and the server runs over an encrypted connection (HTTPS).
- The server has a limit on the number of requests and a daily usage limit, which protect against abuse.
- Data on the device is not additionally encrypted, because it contains only game progress and a list of words. It is protected by your device's lock, like all other apps.
9. Artificial intelligence - what you must know
The bear's answers in the Talk and Voice translate modes are generated on the fly by an artificial intelligence language model (Claude, Anthropic PBC).
This means the answers are not written in advance and we cannot review them in advance. We gave the model instructions to stay with topics suitable for children, not to ask for personal data and not to carry out instructions from what is said. These instructions work reliably, but not with one hundred percent certainty. Therefore:
- for younger children we recommend that an adult is nearby;
- answers are not professional advice and are not a substitute for lessons;
- if the bear ever says something inappropriate, please report it to info@pikabear.app - every such report is valuable to us.
10. Changes to this policy
We may change the policy. Material changes - especially those that concern children's data - will be published in the app before they take effect, and we will ask for consent again if needed. The date of the last change is at the top.
PikaBear is a learning app and does not replace a teacher, a course or adult supervision.