YOUR WORDS, YOUR CHOICE

Privacy policy

Dictore processes dictation on your device by default. This page explains the information stored locally and the optional features that contact other services.

Last updated: September 9, 2026.

Which versions this covers

This policy covers Dictore for Mac and Windows, the Dictore Chrome extension, and, where described, pre-release iOS builds. The Mac App Store version is being prepared for review. Its dictation and rewriting run locally; Dictore-hosted cloud dictation, cloud translation and cloud accounts are disabled in that release.

Local dictation and writing data

Microphone audio is processed by Apple's on-device speech recognition or an installed Whisper model on Apple devices, and by local faster-whisper on Windows. On-device recognition depends on language support and installed system components. Dictore does not upload audio or text while you use these local paths.

Your transcripts, history, custom dictionary, snippets, writing preferences, correction learning, and usage statistics are stored on your device. A writing-style profile may include vocabulary that identifies you; we do not describe it as anonymous. You can disable personalization and remove saved items from the app.

On compatible Macs with Apple Intelligence enabled, Dictore may use Apple's on-device model to polish your text. This does not send your transcript to a Dictore server. Other local cleanup engines also operate on your device.

Recordings and permissions

Recording begins after a user action and requires microphone permission. On Mac, Accessibility permission enables inserting text into other apps. Meeting recording can also capture system audio after you grant the relevant macOS permission. Obtain any permissions you need from other participants before recording them.

Meeting recordings and transcripts are saved locally until you delete them. Ordinary dictation audio is retained only when you enable audio retention; retained audio follows your selected retention settings. Transcript history and saved recordings can be managed in the app. Exported files and copies you paste elsewhere remain under your control in their destination.

Keyboard and browser extensions

The iOS keyboard uses a shared on-device app container for transcripts, dictionary terms and snippets. Full Access enables this shared data flow; basic typing and switching keyboards work without it. The keyboard does not send keystrokes, audio or transcripts to a Dictore server.

The Chrome extension’s standalone mode starts recording after a user action and transcribes locally with WebGPU. First use downloads model data from Hugging Face into browser storage. Optional Desktop companion mode communicates with the app through an authenticated connection on the same computer. These extension paths do not upload dictation content to a Dictore service.

Optional services you choose

• Model downloads: user-initiated downloads contact the model host, such as Hugging Face or GitHub. No dictation content is included in a model download, but the host receives ordinary connection information such as an IP address.
• Mac and Windows meeting or text AI providers: optional summaries, titles and other supported text operations may send the relevant text to the provider you configure. Local Ollama stays on your computer. A remote OpenAI-compatible provider receives the text directly using your own credentials and applies its own privacy and retention terms.
• iCloud: where enabled, personal dictionary terms and snippets can sync through your private Apple iCloud database. This does not route through a Dictore account or Dictore server.
• Pre-release iOS cloud features: if explicitly enabled, cloud transcription or voice transformation sends audio, and cloud rewriting or translation sends text, to Dictore's Cloudflare processing service. Requests include an access token used to verify access. Optional writing-style information is sent only with the corresponding enabled cloud setting. These are not part of the Mac App Store release.
• Optional connected exports: in editions that offer them, exports send the selected content to the service you connect, using your provider credentials. Local file exports and clipboard copies do not themselves upload content.

Remote services can have their own processing and retention policies. Leave remote features disabled if you want your content processed only on your device.

Purchases, support and website visits

Apple handles App Store purchases and Microsoft handles Microsoft Store purchases. Dictore does not receive or store your payment-card details. When you email support, we receive the email address and information you include and use it to respond to your request. Please avoid including recordings or sensitive transcripts unless they are necessary for the issue you want help with.

This website is hosted by Cloudflare, which receives normal web requests and connection information to deliver and protect the site. We do not add advertising or cross-site tracking scripts. Dictore does not sell your dictation content or use it for advertising.

Your choices and contact

You can revoke microphone, Accessibility and recording permissions in system settings, turn off optional providers and sync, disable personalization, and delete local history or recordings. To ask about privacy or information you sent to support, contact [email protected].

Updates to this policy will be published on this page. Features and privacy disclosures will be reviewed again before additional platform versions are released.