Privacy Policy
Effective October 30, 2026
This English translation is provided for convenience. In case of discrepancy, the Spanish version prevails.
Who we are
AudioKids is a product of DJINN FOUNDRY, S.L., the data controller for the processing described in this policy. For any privacy question, write to the contact address.
- Legal name: DJINN FOUNDRY, S.L.
- Tax ID (NIF): B88711726
- Registered office: Plaza Julio Pardo 10, 46625 Cofrentes (Valencia), Spain
- Contact: hola@audiokids.org
- Registered in the Valencia Commercial Registry, sheet V-235200
What data we process
- Account: the email of the parent or guardian who creates the account.
- Child profile (created and controlled by the adult): name, age and language; and optionally: gender, interests, hobbies, favorite characters, topics to avoid, and a text in which the family describes the child in their own words (for example their character, family, friends, pets or what scares them). Older profiles may keep pet (type and name), best friend, siblings (name and age) and family members (name and relationship) until the adult replaces them with that text. Every one of these fields is entered voluntarily by the adult and is used only to write and narrate the stories. None is required except name, age and language. We no longer ask for the child's city, school or holiday place, and we have deleted those already stored along with the old free-text notes, which never reached the stories.
- Cloned voice (optional): if you choose to narrate stories with your own voice, we record a sample of your voice. This is biometric data and we treat it with extra safeguards (see the dedicated section below).
- Payments: processed by Stripe; we never see or store your card details.
- Usage and errors: product usage events (analytics on this website only with your consent) and technical error logs with internal account and story identifiers, without names or story text.
Purposes and legal bases
- Generating personalized stories and providing the service (account, plans, payments, transactional email): performance of our contract with you.
- Product analytics on the website: your consent, requested via a notice before anything activates; you can decline without losing any functionality.
- Voice cloning: your explicit consent, requested at recording time and withdrawable at any moment by deleting the voice.
We do not sell data. There is no advertising in AudioKids, so there are no ad profiles and no data shared with advertisers.
Who we share data with (sub-processors)
We use these providers to run the service, each limited to its function:
| Provider | Function | What it receives |
|---|---|---|
| Cloudflare | Hosting, database (D1) and audio storage (R2) | Service data (account, profiles, generated stories) |
| OpenRouter | Routes each text request to the AI model that writes it | The same data the models in the next rows receive; never your email |
| Mistral (Z.ai's GLM model) | Writes the text of each story, through OpenRouter. Z.ai does not receive the data: the model is run only by Mistral, in the European Union | The child's name, age, language and preferences (interests, hobbies, favorite characters and topics to avoid); the text in which you describe the child; the pet's type and name, the best friend's name, and the name and age of siblings or the name and relationship of family members, if the profile still holds them; the estimated reading level and vocabulary, and a summary of previous stories; and the topics the family searches for or types in the app, to find or create the story's topic |
| Anthropic (Claude) | Writes the text only when GLM does not answer, through OpenRouter and on Anthropic's own servers | The same data as the previous row |
| Anthropic (Claude) |
Checks the quality of stories already written with the
anthropic/claude-haiku-5.5 model, through OpenRouter and on
Anthropic's own servers: it gives a second opinion on some before they are
narrated, checks their questions and scores a sample each night. Stories
Claude writes are checked instead by Z.ai's
GLM, run by NextBit in Spain, because a model does not
check what its own family writes
| The story text, which may include the names in it, its topic and language, and the child's age; never the profile's name, interests or topics to avoid |
| Inworld | Story narration (text-to-speech) and cloning an adult's voice | The story text, including the names in it; your voice sample only if you enable cloning, under an internal name that identifies no one |
| Google Fonts | Typography for the app and the website | Your IP address when the font is fetched, as with any web resource |
| PostHog | Product analytics (EU-hosted) | Usage events; on this website only after your consent; no children's names or identifiers, and age as a range (for example, 6 to 8), never exact |
| Sentry | Technical error tracking (EU-hosted, Germany) | Error traces with internal account and story identifiers; before sending we remove names, story titles, searches and topics typed by the family, and age goes as a range |
| Stripe | Payments on the web | Payment data (handled directly by Stripe) |
| RevenueCat | Native in-app purchases (once the apps ship) | App-store purchase identifiers |
| Resend | Transactional email (verification, account notices) | Your account email |
What child data the AI receives
To write each story we send GLM, Z.ai's AI model run by Mistral, through OpenRouter, the profile data listed in the table above. To narrate it, Inworld receives the story text and, if you enable cloning, your voice sample. We never send your email or payment data.
On every request we ask OpenRouter to use only providers that do not collect the data for their own purposes, and for GLM only Mistral. If GLM does not answer, Anthropic's Claude writes the story, with the same data and under the same condition. To check quality, the text of stories already written (which may include those names, their topic and the child's age) also goes through a second model, Anthropic's Claude, likewise through OpenRouter and under the same condition.
How long we keep data
For as long as your account exists. When you delete your account (or a child profile), we delete the database rows and the associated stored audio at deletion time, and cloned voices are flagged for deletion at the voice-synthesis provider as well. Stripe retains the payment records it is legally required to keep.
Your rights
You have the rights of access, rectification, erasure, portability, restriction and objection (GDPR arts. 15-21). Erasure does not require emailing us: from the app itself you can delete your whole account or a specific child profile (Settings → Account → Delete my account), and that deletion purges the database and the audio storage. Step-by-step instructions, with what is deleted and what we keep, are on Delete your account. You can also exercise any right by writing to hola@audiokids.org. If you believe we have not responded properly, you can complain to the Spanish data protection authority (aepd.es) or your local one.
Children's data
Children do not create accounts on AudioKids: profiles are always created and managed by an adult, who decides what information to enter. Profile data is used exclusively to personalize that child's stories. We show no ads, sell no data, and do not profile children for any purpose other than adapting educational content to their level.
Cloned voice (biometric data)
Voice cloning is optional and requires your explicit consent at recording time. The sample is used solely to create your synthetic voice and narrate your children's stories. You can delete your voice at any time from the app: we remove the sample and preview from our storage and order deletion of the clone at the voice-synthesis provider, completed once the provider confirms it.
Cookies and analytics
On this website we only use PostHog analytics, hosted in the European Union, and it does not activate until you accept the consent notice. If you decline, no analytics code loads and no measurement cookie is written, and we remember your choice. We use no advertising cookies and no third-party trackers.
International transfers
We use EU hosting where the provider offers it (for example, PostHog and Sentry). Some providers (OpenRouter, Anthropic, Inworld, Google Fonts, Stripe, Resend) process data in the United States; in those cases we rely on the GDPR transfer mechanisms each provider offers in its data-processing agreement (standard contractual clauses or the EU-U.S. Data Privacy Framework, as applicable).
GLM, the model that writes the stories, is run by Mistral in the European Union, and GLM as a reviewer by NextBit in Spain. Anthropic, which writes when GLM does not answer and checks quality with Claude, processes the data in the United States.
Changes to this policy
If we change this policy, we will publish the new version here with its effective date. If a change is relevant to your rights, we will notify you by email.