Privacy Policy
Last updated: November 2026
HushPen is an Android dictation app that records speech, transcribes it, and inserts the result into text fields across apps.
Data handling
Local mode
In local mode, audio is processed on-device using local speech recognition models. Audio does not leave your device.
Cloud mode (your own API key)
Cloud features are off by default. Before HushPen sends anything to a cloud provider, it asks for your consent for that provider, and nothing is sent until you agree.
- Transcription: recorded audio is sent directly from your device to OpenAI's transcription API.
- Cleanup (optional): the transcribed text is sent directly from your device to the provider you choose (OpenAI or Anthropic). You can also use an on-device model or your own server instead.
The provider's privacy policy applies to that data. HushPen does not run a relay server for these requests.
Recordings are kept in memory only, never written to storage, and discarded after transcription. A single dictation stops after 5 minutes, or after one minute of silence.
API keys
Your API keys are stored on your device in app storage and sent only to the provider they belong to.
Accessibility Service
HushPen uses Android's Accessibility Service only to detect when a text field is focused (so the bubble appears only when you can type) and to insert dictated text into that field after you tap the bubble.
- It does not read, store or send the content of your screen.
- It never reads or types into password fields.
- It does not monitor browsing or perform background automation.
Usage statistics
Usage statistics are off by default. If you turn them on (in the first-run setup or in Settings > Privacy), HushPen sends anonymous usage statistics to plan development and the cost of a future cloud service. You can turn them off again anytime, and delete what was sent with "Delete my statistics".
Sent once a day, as numbers only:
- a random install id, not linked to you or your device, replaced at least every 13 months,
- the day (UTC),
- number of dictations, seconds of audio per mode (on-device, own API key, cloud) and number of words,
- dictation language, speech model and cleanup mode,
- app version, the country of the phone's language settings and a memory size range.
Never sent: the text you dictate, audio, the apps you dictate into, device or advertising identifiers. IP addresses are not stored.
The data is stored on Cloudflare (EU) for at most 13 months, is not shared with anyone and is not used for advertising.
Crash reports
Off by default. If you turn on "Send crash reports" in Settings > Privacy, a crash sends the place in the code where the error happened (class names and line numbers), the app version, the Android version and the phone model. It never contains your text, audio or an identifier, and it is deleted after 90 days.
Data kept on your device
The last 30 dictated texts are kept, encrypted, so that you can copy them again when inserting failed; you can switch this off or clear it in the History tab. Word counts and badges are numbers only. The terms of your personal dictionary are sent to a cloud provider as spelling hints only when you use cloud features. Uninstalling HushPen deletes all of it.
Tips
In the Google Play version, tips are one-time purchases handled by Google Play. HushPen receives no payment details.
Data collection
HushPen does not have user accounts. Apart from the optional cloud requests and the usage statistics described above, nothing leaves your device.
Contact
For questions about privacy, write to support@hushpen.app.