Skip to content
[H] Hush
  • Why
  • Features
  • Meetings
  • Privacy
  • FAQ
  • Guide
EN 日本語
Download

Privacy

Last updated 18 September 2026

Hush runs on your Mac. There is no account, no sign-in, and no analytics or tracking of any kind. Nothing in this list is inferred from marketing — it is what the shipped app does.

Audio

By default, your microphone audio and any meeting audio Hush captures are transcribed on your Mac, by Apple's built-in Speech engine or by WhisperKit. Audio is held in memory for as long as it takes to transcribe it, and it is never written to disk. Meeting audio never leaves your Mac. Dictation audio leaves it only if you switch on cloud transcription, described below.

Transcripts

Transcripts and meetings are stored locally in Hush's own database inside your Mac's Application Support folder, and your phrase replacements and settings are kept in Hush's preferences on your Mac. They are not encrypted beyond the protection macOS gives your user account, and they are not synced anywhere. Deleting an item in the app, or deleting the app's data, removes it.

Optional: cloud transcription

This is off by default. If you add your own OpenAI API key and switch on Use OpenAI Cloud Whisper — or the Compare WhisperKit vs OpenAI developer option — in Settings → AI, the audio of each dictation is sent to OpenAI to be transcribed. Meetings are always transcribed on your Mac. The request goes directly from your Mac to OpenAI, using your key, billed to you by them, and it is covered by their privacy policy and data-retention terms.

Optional: AI Polish, voice editing and meeting translation

These features are off by default. When you turn one on and add your own OpenAI or Anthropic API key, the text of the transcript involved is sent to that provider so it can be cleaned up, edited or translated. These features send text only, never audio. The request goes directly from your Mac to the provider, using your key, billed to you by them, and it is covered by their privacy policy and data-retention terms. Your API keys are stored in the macOS Keychain.

Optional: in-app feedback

If you send feedback from Settings, Hush submits what you typed, the app version, your macOS version, a truncated random install identifier, and — only if you fill them in or switch them on — your email address and recent transcription error logs. It goes through a small server we run and becomes an issue in a private repository that only the developer can read. It is never used for marketing.

Optional: crash reports

Hush includes an opt-in crash reporter that is off by default and never sends transcripts. Current builds ship without a reporting endpoint configured, so no crash data leaves your Mac even if the setting is on.

Updates and downloads

About once a day, Hush checks for updates by requesting a small file from hush.kei.kyoto. You can also check at any time with Check for Updates in the Hush menu. Like any web request, the server sees your IP address, the time, and a user agent. Nothing identifying you or your usage is included, and those logs are not used to build any profile.

If you use WhisperKit, Hush downloads the speech model you choose, once, from Hugging Face (huggingface.co), where the open-source WhisperKit models are published. Apple Speech's language assets are downloaded by macOS itself. These are ordinary file downloads: no audio or text is sent.

Children

Hush is not directed at children and collects nothing beyond what this page describes.

Changes

If this policy changes, the date above changes with it. Material changes will be noted in the release notes.

Contact

Questions about privacy? Email support@kei.kyoto, or use Settings → Feedback inside the app. Both reach the developer directly.

Hush

On-device dictation for Mac.

v0.4.2 · Built solo · Notarized

Pages

  • Guide
  • Privacy
  • Terms
  • Contact
  • Download

© 2026 Hush. All rights reserved.