Privacy and permissions.

An explanation of current data handling and permissions, not a formal privacy policy.

What is processed online?

Speech recognition and text editing require a network connection. Available text around the cursor is used for editing, and dictionary entries can provide recognition hints. Local mode does not mean speech and text are processed only on the device.

What is stored locally?

Text history, dictionary entries and retained audio use encrypted local storage. You can set history retention, delete entries or clear history. History and recordings cannot be assumed to remain available when no retention is selected.

Why are permissions needed?

Microphone permission is used to capture speech. Accessibility permission is used to attempt to insert text where you started. You must grant these permissions yourself. Password and other secure input fields are excluded.

What are the limits of deletion and export?

Local deletion is not proof that every online processor has deleted its data. Audio conversion, retry and export can involve controlled temporary unencrypted files. Exported audio is a regular usable file that you need to store appropriately.

What still needs to be clarified?

A formal privacy policy has not yet been provided. Online logs and third-party retention periods still need to be explained. The current information does not support promises of zero retention across all services or no use for training.