Reader Privacy Policy

Last updated: 2026-09-29

This policy applies to Reader at reader.fabuly.app and the Reader Android and iOS applications. It does not replace the policies of the separate Fabuly audiobook app, external websites, or app stores. Visiting the Fabuly marketing and policy website at fabuly.io is also subject to that website's Cookie Policy.

Quick summary

Your imported books, reading progress, and generated audio are stored on your device or in your browser. Reader has no account system or cloud-library synchronization.
Reader processes local documents, photographs, text recognition, and speech generation on your device. It does not upload those local inputs to an AI or speech service.
When you use the web app's public-link or Gutenberg import, our relay receives the requested URL and downloads the source so it can return it to your browser. Local file and pasted-text imports do not use that relay.
The web demo uses a small English speech model; the native apps use a larger model with more voices and languages. Speech models are initially served from Hugging Face. Native apps can use a verified replacement download location. Browser pronunciation resources come from Reader's own hosting; browser recognition-language files use a separate download provider. Those providers receive ordinary connection information, but Reader does not send them your book text, photographs, or generated audio.
Reader does not integrate advertising, account tracking, or its own analytics/crash-reporting service. Hosting providers receive technical request data, and Android's Google ML Kit components collect their own diagnostic and usage information as described below.
We do not sell your personal information or use your library to train models.

Who we are

Reader is operated by:

Fabuly AI SAS

SIRET: 927 524 462 00029

Registered office: Plaine Images, 99 A Boulevard Constantin Descat, 59200 Tourcoing, France

Contact: joseph@fabuly.io

Fabuly AI SAS is the controller of the personal information we process to operate Reader, its import service, and support. Third parties may be independently responsible for information they process under their own policies, such as when you visit their websites or use their platforms.

Information Reader processes

Your local library and preferences

Reader stores imported text, documents and images, book details, source links where applicable, chapters, reading and listening positions, recent-reading times, selected languages and voices, playback settings, and generated audio. It also stores downloaded speech and recognition resources and temporary import or generation data needed to complete or recover an operation.

These files are held in the native app's private storage or the browser's site storage, including IndexedDB and the origin-private file system. Reader does not send this library to us or synchronize it across devices. Other users of the same device or browser profile may be able to access it, so use your device's access controls.

Reader keeps a local visit count and local state for optional feedback or store-review prompts. Those values are used on your device and are not sent to an analytics service.

During playback, Reader can provide the book's title, author, cover, chapter, and playback state to operating-system media controls, a lock screen, or a connected Android Auto or CarPlay interface. People who can see those surfaces may see this information. Manage device notifications and connected-car access accordingly.

Public links, Gutenberg, and imported website resources

In the web app, when you request a public URL import, your browser sends that URL and the requested content type to our import relay. The relay downloads the public source and returns the downloaded bytes. Importing associated illustrations may involve additional requests for their public URLs. The relay necessarily processes the source material, which may contain personal information made available by the source website.

The relay does not receive your browser's website-login cookies or credentials, does not log in on your behalf, and does not execute scripts in imported pages. It stages downloaded material in temporary server files to validate the transfer and removes those files when the request completes or fails. If a server process is interrupted, temporary files may remain until that temporary instance storage is discarded. It does not create a permanent cloud copy of your library.

Gutenberg search matches titles and authors locally against a downloaded public catalog. Reader does not send your search text to our servers or Gutenberg. Opening or importing a result does request that book and its resources. Catalog download requests pass through our relay in the web app; native applications download public sources directly.

Source websites receive download requests from our relay when it is used, or from your device when a native app connects directly. They process those requests under their own policies. Do not put passwords, private access tokens, or other secrets in a URL sent to the public-link service.

Connections, operational logs, and security

Our website and import infrastructure use Google Firebase Hosting and Google Cloud. Providing these services involves ordinary technical request information such as IP addresses, request times, requested resource paths, browser or client information, response status, and transfer size. Book identifiers can appear in a browser page URL. A URL submitted for import is processed in the request body by the relay.

Reader's relay application logs record error categories and whether a response had started; they are designed not to include imported text, document contents, source URLs, credentials, or request bodies. Hosting and platform logs are separate and can contain technical request metadata. We use operational information to deliver the service, investigate failures, protect it against abuse, and comply with legal obligations.

Model and recognition downloads

The Android and iOS apps use pinned Supertonic speech models and voice presets, initially served by Hugging Face, including its download infrastructure. On an explicit Download or Retry, the native app checks a small source-configuration file hosted on Google Cloud Storage. A valid configuration can change the download location for the same verified model files; Reader retains the last valid location locally and otherwise falls back to the built-in source. Installed voices do not require this request.

The browser's Web demo uses a separate, small TinyTTS English speech model pinned to Hugging Face. After the page first renders, it starts preparing the demo voice in the background, including the model download if needed. You can pause or retry preparation; audio starts only after you choose to play a book or sample. Its English pronunciation dictionary, neural pronunciation resource, and separate model-source configuration are served with Reader from our hosting. That configuration can relocate the same verified TinyTTS model files without changing the model. A sample preview does not create a library book.

Model and resource hosts receive your IP address, the requested asset, and normal connection information. These downloads do not send your book text or generated speech to the providers. Speech synthesis runs locally on every platform. The browser and native libraries remain separate.

For browser OCR, Reader downloads requested recognition-language files from tessdata.projectnaptha.com and caches them locally. That provider receives normal download information and can see which language resource is requested. Your document images and recognized text are processed locally and are not sent with the download.

The browser's speech engine, PDF tools, OCR worker software, and fonts are otherwise served with the Reader web application. Downloading a model is different from sending content to a remotely hosted AI service; Reader has no cloud speech or cloud OCR fallback.

Android-specific services and permissions

Android uses Google Play services / ML Kit for optional on-device text recognition and, on supported devices, ML Kit's on-device generative features for metadata suggestions. Inputs and results are processed locally. Google may receive SDK diagnostic, configuration, performance, and usage information, including device/application information, identifiers, recognition or inference events, errors, and input/output size information, as described in its ML Kit Android data disclosures.

Google may contact its services to check availability or download and update required platform resources. Reader's optional generative metadata feature uses a compatible on-device model already available to it. Reader does not use a server fallback for these suggestions.

Reader requests access to documents, selected photographs, the camera, or playback notifications when needed for the corresponding feature. Permission and picker choices are controlled by Android. Camera or photo inputs are used locally for the import you initiate.

iOS-specific services and permissions

iOS uses Apple's system document, image, Vision text-recognition, and, where available, on-device Foundation Models features. Reader processes the selected content locally and has no cloud fallback for recognition or metadata suggestions. Apple may process platform, app-distribution, diagnostic, or service information according to its own policies and your device settings.

System document/photo pickers and camera permissions control which inputs Reader can access. You can manage applicable permissions in iOS Settings.

Websites opened in the native import browser

Android and iOS include a browser view for importing a webpage you open. Unlike the web app's public-URL relay, this view can run the website's scripts and use website cookies and local storage. If you sign in to that website, you provide information directly to it under its own terms and privacy policy.

When you choose to import the page, Reader captures and processes its available page content locally. Website cookies and browsing data can persist in that embedded browser. Reader provides a control to clear its website data; you can also use the relevant app or device settings. Do not assume that deleting an imported recording also clears a website login.

Support, feedback, and store reviews

If you choose to send feedback, Reader opens your email application with our contact address. Reader does not automatically attach your books or diagnostic reports. Sending the email is your choice. We receive your email address, message, and any attachments you choose to send, and use them to respond and investigate the issue.

Speech diagnostics shown in Reader remain local unless you choose to copy or share them. Review their contents before sending them to anyone.

If you follow an app-store, source, or other external link, that service receives the request and applies its own policy. Store-review prompts open the relevant store only when a link is configured and you choose to continue. We do not receive your store credentials.

How we use information and our legal bases

Where data protection law requires a legal basis, we rely on:

Providing the service you request: processing necessary to perform our agreement with you, including delivering Reader, responding to an import request, and providing support about the service (GDPR Article 6(1)(b)).
Legitimate interests: maintaining reliable and secure infrastructure, preventing abuse, diagnosing faults, and responding to general enquiries, while balancing those interests against your rights (Article 6(1)(f)).
Legal obligations: responding to valid legal requirements and preserving records where the law requires it (Article 6(1)(c)).
Consent where required: for any optional processing that legally requires your consent. You may withdraw that consent without affecting earlier lawful processing (Article 6(1)(a)).

Choosing a local feature or acknowledging this policy does not authorize unrelated advertising or disclosure of your library. Reader does not use your reading activity for advertising profiles or make decisions about you with legal or similarly significant effects.

Third-party services

The main providers relevant to Reader are:

Google Firebase Hosting / Google Cloud: website delivery, the browser import relay, and operational infrastructure. See Firebase privacy information.
Hugging Face: speech-model and voice-asset downloads. See the Hugging Face Privacy Policy.
Project Gutenberg and requested source websites: public catalogs, books, webpages, and associated resources. See Project Gutenberg's Privacy Policy and the applicable source website's policy.
The Project Naptha / Tesseract language-data service: browser recognition-language downloads from tessdata.projectnaptha.com. The application does not send OCR inputs to that service.
Google Play services / ML Kit on Android: platform resources and local recognition or metadata features, with SDK diagnostic processing described above. See Google's Privacy Policy and ML Kit terms and privacy information.
Apple and Google app platforms: application distribution, operating-system services, permissions, and optional store reviews. See Apple's Privacy Policy and Google's Privacy Policy.
Email providers: delivery and handling of support messages you choose to send.

Third-party code running locally, such as PDF.js, Tesseract.js, or ONNX Runtime, does not by itself mean that its authors receive your documents. The network connections made by Reader are described above.

When we share information

We share information only as needed with hosting and support providers, the source or download services you request, and the platform services described in this policy. We may also disclose information when legally required or necessary to establish or defend legal claims or protect rights and security.

If the business operating Reader is transferred, service-related information we hold may be transferred subject to applicable data protection law and appropriate notice. Your local-only library does not become a server-side library because of such a transaction.

We do not sell personal information or share your reading history for advertising.

International processing

Your local library remains where your device stores it. Network services may process technical information or requested public imports outside your country. Reader's import relay is configured in Google's us-central1 region in the United States; hosting and third-party download infrastructure can operate internationally.

For transfers we arrange that are subject to EEA or UK transfer rules, appropriate safeguards must apply, such as an applicable adequacy decision or approved contractual clauses. You can contact us for information about the safeguards applicable to our service providers. Third-party websites and platforms explain their own arrangements in their policies.

How long information is kept

Local library and generated audio: kept on your device until you delete it or application/browser storage is removed. Reader can delete generated audio separately while preserving the book and reading progress. There is no server copy that we can restore.
Local model, catalog, recognition, and temporary caches: kept as needed for the feature, reuse, updates, or recovery; they can be replaced or removed by Reader or by clearing the relevant application/site data.
Relay source files: temporary for the request and deleted as described above. The public Gutenberg catalog may also be cached by delivery infrastructure; that catalog does not contain your search query or library.
Technical logs: retained under the infrastructure's configured retention settings for the period needed to investigate operational or security issues and manage the service. Relevant records may be preserved longer for an unresolved incident, legal obligation, or claim, and are not intended to form a permanent reading history.
Support correspondence: retained as needed to resolve the request, handle related follow-up, and meet applicable legal or dispute-resolution requirements. Avoid sending a whole document when a short description of the problem is sufficient.

Third parties may retain their own connection, platform, or account information under their own policies. Contact us if you need details about information held by us for a particular request.

Cookies, browser storage, and backups

The Reader web application uses site storage for its local library, models, preferences, and operation. It does not integrate analytics or advertising cookies. Clearing that storage can remove all locally saved books and progress; rejecting or disabling necessary browser storage can prevent Reader from working.

This statement does not apply to external websites you open, websites inside the native import browser, or the separate fabuly.io website. Those services may use cookies under their own policies and consent choices.

Reader excludes its native private library from the normal app-managed operating-system backup paths it configures. This is not a guarantee against copies made by device-management tools, a device owner, or other backup software. Original files you select from another app or cloud drive remain subject to that provider's storage and backup settings.

Your choices

You can:

Read and import local text or files without downloading speech models in the web app.
Avoid public-URL imports and use local files or pasted text instead.
Cancel model downloads and generation, and delete recordings or generated audio in Reader.
Remove browser site data or native app data to remove remaining local caches and settings. Keep original documents before doing so.
Clear website data in the native import browser and manage device permissions in system settings.
Decide whether to send feedback, copy diagnostics, open external links, or submit a store review.

There is no Reader account to delete and no Reader cloud synchronization to disable. Deleting a local recording does not delete the original file you imported, a website's copy, or an email you previously sent us.

Your privacy rights

Depending on applicable law, you may ask to access, correct, erase, restrict, or receive a portable copy of personal information we hold about you. You may object to processing based on legitimate interests and withdraw consent where processing relies on it. These rights are subject to applicable legal conditions and exceptions.

Contact joseph@fabuly.io. We may request information reasonably needed to verify and handle your request, without asking you to upload your library unnecessarily. We cannot access or identify local-only books from your email address; use Reader or your device settings to manage those copies.

You may complain to the CNIL in France or another competent data protection authority, including the authority where you live or work in the EEA. These rights do not depend on making a complaint to us first.

Children

Reader is intended for users aged 15 and over, consistently with its Terms of Use. We do not knowingly solicit children's personal information through Reader or use it to market to children under 15. If you believe a child under 15 has sent personal information to us, contact us so we can review and address it. Local content on a child's device must be managed on that device by its user or guardian.

Security

Reader uses private application/site storage, HTTPS connections, validated model downloads, and limits on imported resources. The public import service restricts source addresses and request sizes to reduce misuse. Access to our infrastructure is limited to people and providers who need it for their work.

No system is completely secure. Protect your device and browser profile, keep software updated, and keep original copies of important documents. Other extensions, apps, device administrators, or websites operate under their own permissions and are not controlled by Reader.

Changes to this policy

We may update this policy when Reader's features or data practices change. The date and version above identify the current text. We will provide additional notice or seek consent when required by law. For questions, contact joseph@fabuly.io.