Privacy Policy for D-Schutz

Last updated: 2026-08-06

D-Schutz ("the app") is a local, offline PDF reader with an on-device AI assistant, developed and operated by Medina Labs (Max Medina), reachable at max@medina-labs.com. This policy explains what data the app handles, what (if anything) leaves your device, and why.

Summary

D-Schutz is designed to work fully offline after a one-time setup step. Your documents, your questions, and the AI's answers never leave your device. The only data that travels over the network is the one-time download of AI model files from Hugging Face, a third-party model hosting service — no content from your PDFs is included in that traffic.

What data the app handles

Your PDF documents

When you import a PDF into D-Schutz, the app copies it into the app's private internal storage on your device (not a shared or public location, and not any external server). This copy is used to render pages, extract text, and answer your questions locally. D-Schutz does not upload your PDFs, their text, or any content extracted from them to any server.

Password-protected PDFs

If you import a PDF that is password-protected, D-Schutz detects this, prompts you for the password, and uses it locally to decrypt the file. The app then saves a decrypted, unprotected copy of that PDF in its own private internal storage so it can render and search the document going forward. This decrypted copy is not password-protected, even though your original file (wherever you keep it, e.g. in your device's file manager or cloud storage) is untouched and remains encrypted as before. The password you enter is used only in memory to perform the decryption and is never stored, logged, or transmitted anywhere.

Because the app-private copy is no longer protected once imported, anyone with access to your unlocked device and file system-level access to the app's private storage (which normal apps and the user cannot access without special permissions such as root) could potentially access the decrypted content. D-Schutz does not add any additional password lock of its own on top of this. This behavior is provided on an "as-is" basis; D-Schutz and Medina Labs are not liable for access to decrypted content resulting from loss, theft, or compromise of your device. You can remove a decrypted copy at any time by deleting the document from within the app.

Questions you ask and AI-generated answers

D-Schutz includes an on-device AI assistant (built on a local Gemma language model, run via Google's LiteRT-LM runtime) that can answer questions about a document or produce summaries. All AI processing — reading document text, generating answers, and any embedding/retrieval used to find relevant passages in long documents (a local on-device "RAG" retrieval feature) — happens entirely on your device. Your questions, the extracted document text sent to the model, and the model's responses are never sent to any server. This data is only ever stored locally on your device (e.g., cached embeddings for retrieval, and your saved AI exchanges for a document) and is not transmitted anywhere.

What is NOT collected

D-Schutz does not have user accounts, does not include analytics, advertising, or crash-reporting SDKs, and does not sell or share your data with any third party. D-Schutz itself collects no usage statistics about you.

One exception is worth stating plainly: D-Schutz uses Google's ML Kit for two features (reading text from scanned pages, and the optional document scanner). Google's ML Kit libraries send Google technical metrics about the performance and utilisation of the ML Kit features themselves, and use Google's Firebase Remote Config and Firebase Installations services for configuration and diagnostics. This is Google's processing, described in the ML Kit terms, and D-Schutz cannot disable it. It covers how the ML Kit features perform — not your documents. ML Kit processes page images entirely on your device and does not send your document content, page images, extracted text, or AI questions and answers to Google.

What leaves your device, and why

The network activity D-Schutz performs is downloading AI model files, one time, directly from Hugging Face (huggingface.co), a third-party service that hosts publicly available machine learning model files:

These downloads fetch only the model weight files themselves — they do not include or transmit any of your documents, questions, or personal information. Once downloaded, the model files are stored locally on your device and used entirely offline for all subsequent AI features. No further network access is required or performed for Q&A or summarization.

Two other things reach the network, neither of them carrying your document content:

Making these HTTPS requests necessarily discloses your device's IP address and standard HTTP connection metadata to Hugging Face, as with any download from any website — no document content or personal information is included, but the IP address itself is handled under Hugging Face's own privacy policy (https://huggingface.co/privacy), not this one.

Because these downloads are large (the language model is roughly 2.5 GB, the embedding model roughly 120 MB), the app checks your network connection type before downloading, so you may see a prompt or warning if you are not connected to Wi-Fi. This check uses only generic network status (e.g., whether you're on Wi-Fi vs. cellular) and involves no transmission of personal data.

Permissions used

D-Schutz requests the following Android permissions, used only as described:

D-Schutz does not request access to your camera, microphone, contacts, location, or any other sensitive device data, and does not use storage permissions beyond its own private app storage (which does not require a runtime permission on modern Android versions).

Data retention and deletion

All app data — imported PDFs (including decrypted copies of password-protected PDFs), AI conversation history, and cached retrieval data — is stored only in the app's private storage on your device. This data is deleted automatically if you uninstall the app, or you can remove individual documents from within the app. Downloaded AI model files are similarly stored locally and removed on uninstall.

Children's privacy

D-Schutz is intended for a general audience and is not directed at children. It is not part of Google Play's Designed for Families program and is not marketed to children. The app does not knowingly collect personal information from anyone, including children, because — as described above — it does not collect or transmit personal information at all. If you believe a child has provided personal information through this app, please contact us using the information below so we can investigate (though, per the design described in this policy, no such data would be sent to or stored by us in the first place).

Your rights and governing law

Because D-Schutz does not collect, store, or transmit any personal data off your device (beyond the IP/connection metadata inherent in the one-time model download described above), there is no personal data held by Medina Labs for it to access, correct, delete, or export on your behalf under laws such as the EU/UK GDPR or the California CCPA/CPRA — the "delete" and "export" actions you'd otherwise request are things you can already do yourself, locally, from within the app (see "Data retention and deletion" above). If you are in the EU/UK or California and have questions about how these rights apply given the app's design, contact us using the information below. This policy and D-Schutz's operation are governed by the laws of the jurisdiction in which Medina Labs operates, without regard to conflict-of-law principles.

Changes to this policy

If this policy changes, the "Last updated" date at the top of this document will be revised. Continued use of the app after changes are published constitutes acceptance of the revised policy.

Contact

If you have questions about this privacy policy or D-Schutz's data practices, contact:

max@medina-labs.com