DICTATE
View on GitHub

Privacy

Updated 21 October 2026

Dictate runs on your computer. The extension and the studio send what they record only to the Dictate server on your own machine, which saves it in your project folder. There is no Dictate account, no analytics, and no server of ours that your data could reach.

What the Chrome extension records

Only while a review session you started in Dictate Studio is recording. Before you start, while you pause, and after you finish, it records nothing. During a session it records:

It never reads what you type into fields, including passwords and form values.

Where it goes

To the Dictate server running on your computer, at localhost. The extension refuses to send to any other address. The server writes the session into your project, in .dictate/sessions/: report.md, session.json, the screenshots and frames, the transcript, the raw events, and your audio. The files stay there until you delete them. Dictate doesn't upload them anywhere.

Your voice

The studio records your microphone while a session runs and saves the audio in the session folder. The transcript is made on your computer by faster-whisper.

While a note is open, the studio also shows live captions using your browser's built-in speech recognition. In Chrome and Edge, that feature may send the audio to Google or Microsoft to turn it into text, under their privacy policies. The final transcript doesn't depend on it.

Claude

Dictate itself sends nothing to Anthropic. When you ask Claude Code to act on a session, it reads the report and the images like any other file in your project, under your own agreement with Anthropic.

What stays in your browser

The extension keeps the current session's state (on or off, your key bindings, and the local server's address) in Chrome's session storage, which clears when you close Chrome. The studio remembers your preferences, such as the last address you reviewed, your microphone, and your controls, in its own browser storage on your computer.

Why the extension asks for its permissions

Downloads

Setup installs faster-whisper from PyPI and downloads the speech model from Hugging Face. Dictate itself comes from npm. Those services see an ordinary download request, as with any other package.

Sharing and selling

Dictate doesn't sell, rent, or share your data, and doesn't use it for anything except your review session. Its use of information from Chrome follows the Chrome Web Store User Data Policy, including the Limited Use requirements.

Changes and contact

Dictate is open source, so every change to it, and to this page, is public in the repository. Questions or concerns: open an issue.