Privacy Policy

Last updated: 5 July 2026

This policy explains what data lisper (“we”, “the app”) processes, why, where it goes, and the choices you have. We tried to keep it plain and honest rather than legalese.

Who is responsible

The data controller for lisper is Pierre Lorenz, Hohe Straße 12, 56244 Hahn, Germany. For any privacy request, contact bl4ckpipe@proton.me. (See our Imprint.)

What we process, and why

Voice audio. When you dictate, your microphone audio is streamed to our server and forwarded to a speech-to-text provider so it can be turned into text. Audio is processed in transit and is not stored by us — unless you switch on the “Support lisper” slider (see below).

Transcribed and dictated text. The recognised text is sent to a language model that cleans it up (and, in agent mode, decides which actions to take). We do not keep a record of your dictated text on our servers — not even when “Support lisper” is on (that shares audio only).

Account data. If you create an account we store your email address and a securely hashed password (bcrypt). A hashed device identifier is used to run the free trial.

Payment data. Subscriptions are handled by Stripe. Your card details are processed by Stripe, not by us; we only store a customer reference and your subscription status.

Usage and diagnostics. We log minimal technical data to run the service and prevent abuse: request counts and token usage per account, and, for our download links, the IP address and referrer of the request. Our app log on your Mac stores only lengths and timings, never the content of what you said.

The “Support lisper” slider (opt-in)

You can choose to help us improve recognition by sharing the audio of your dictations — audio only, never the transcribed text. This is the only case in which any of your dictation content is stored. It is off by default, entirely voluntary, and you can switch it off at any time in the app. While it is off, nothing is collected.

Where your data goes (sub-processors)

Which providers see your audio and text depends on the mode you choose in the app:

Privacy (EU) mode keeps processing within the EU: speech-to-text and text cleanup run on Mistral / Voxtral (France).

Fast mode uses providers outside the EU for lower latency: speech-to-text via Deepgram (USA, EU endpoint) and text cleanup via DeepSeek. If you require EU-only processing, use Privacy mode.

Lifetime / BYOK: with your own API keys (e.g. Google Gemini, OpenRouter), dictation cleanup goes directly from your Mac to that provider, under your own account and their terms — our servers are not involved.

Other processors: Stripe (payments) and our own server infrastructure in the EU. Screenshot-based actions, when you use them, send that screenshot to Mistral (EU) for reading.

Agent actions run on your Mac

Agent capabilities (files, apps, shell, etc.) are off by default and act locally on your machine after you enable them and grant the relevant macOS permissions. The app asks for confirmation before it changes anything. We do not receive the contents of your files unless they are part of what you send for a specific request.

Retention

Account and billing records are kept while your account is active and for as long as legally required afterwards. Diagnostic logs are kept only as long as needed to operate and secure the service. Opt-in quality samples are kept until you withdraw consent or we no longer need them.

Your rights

Under the GDPR you can request access to, correction of, or deletion of your personal data, restrict or object to processing, and request portability. To exercise any of these, or to withdraw consent, contact us at bl4ckpipe@proton.me. You also have the right to lodge a complaint with your local data protection authority.

Changes

We will update this page when our processing changes and revise the date at the top.