Privacy Policy
Last updated September 6, 2026
Local transcription keeps the speech-to-text step on your device after the selected model is downloaded. Model downloads and updates require a connection. Optional external AI processing sends the text needed for that feature to the provider you configure.
The short version
- The speech-to-text step runs locally on your device after the selected model is downloaded.
- Downloading models and updates requires a network connection.
- If you enable an external provider for optional AI processing, the text needed for that feature is sent to that provider.
- Review the setting and the provider's privacy policy before enabling external processing.
Local transcription
Verbatim captures microphone audio while you dictate and performs the speech-to-text step locally using the model you selected and downloaded. Optional text processing is a separate step described below.
- Models: downloaded speech and language models are cached on the device.
- Settings: preferences such as the hotkey and selected model are stored by the application.
- History and dictionary: transcripts and custom words you choose to retain are managed through the application's local controls.
Network access and optional external processing
Network access is relevant in these cases:
- Model downloads: the app downloads the selected model files from their hosting provider.
- Optional AI post-processing and text transforms: if you enable an external provider to clean up, format, translate or rewrite dictated or selected text, the text needed for that feature is sent to the provider you configure under its terms and privacy policy.
- Updates: the application or its download source may connect to check for and deliver updates.
Microphone
Verbatim uses the microphone to capture audio while you are actively dictating. Review the microphone permission shown by your operating system and the current application settings for the platform you use.
Android accessibility service
On Android, Verbatim offers an accessibility service (AccessibilityService API) whose sole purpose is to insert your dictated text into the text field you are typing in, at the cursor, when you tap the floating Verbatim bubble.
- It uses the currently focused text field so it can place your words at the cursor.
- It skips password fields and other sensitive input fields.
- The service does nothing until you explicitly enable it in Android's Accessibility settings, after an in-app disclosure.
- If you enable optional external AI processing, the relevant text follows the provider boundary described above.
Local data and deletion
Downloaded models, application settings, and any history or dictionary entries you choose to retain are managed on your device. Available storage and deletion controls can vary by platform and release.
Use the controls in the application or your operating system's app-data settings to remove local data. Data sent to an external AI provider is governed by that provider's policy and account controls.
Children
Verbatim is a general-purpose productivity tool and is not directed at children.
Changes to this policy
We may update this policy as the app evolves. When we do, the “last updated” date above will change.
Contact
Questions about privacy: aoa@live.ca
Verbatim is developed as a fork of Handy by CJ Pais, with the upstream attribution retained. You can inspect the source and current documentation here: github.com/GalaxyRuler/Verbatim