FeaturesAI DisclosureTransparency & Data SharingPrivacy PolicyTerms of UseContact Get the App
Honest AI

AI Disclosure

VoiceScribe uses two kinds of AI: on-device AI for transcription and optional cloud AI for advanced features. This page explains exactly what's what — so you always know where your data is processed.

Last updated: 2026 · Applies to VoiceScribe for iOS

The short version: Your audio is transcribed on your iPhone with WhisperKit and never uploaded. Cloud AI features powered by the DeepSeek API run only when you choose them, and only on the transcript text you select.

What AI features exist

VoiceScribe includes the following AI-assisted capabilities:

  • Speech-to-text transcription — convert recorded or imported audio into text.
  • AI Summary — condense a transcript into key points.
  • AI Notes — generate structured notes, concepts and action items.
  • AI Chat — ask questions about a transcript and get answers.
  • AI Rewrite — rephrase or clean up text.
  • AI Translation — translate transcript text into 100+ languages.
  • AI Study Tools — study modes, explanations and memory recall.
  • AI Quiz Generation & Flashcards — turn notes into quizzes and flashcards.

Which AI features run locally (on your device)

Audio transcription is processed on-device using WhisperKit. Your audio recordings are not uploaded for transcription. On-device processing also powers live transcription, language detection and the dictation engine behind the voice keyboard.

Because transcription is local, it works without an internet connection and your raw audio stays on your iPhone.

Which AI features use DeepSeek (cloud)

The following optional features may send transcript text to the DeepSeek API, and only when you explicitly choose to use them:

  • AI Summary
  • AI Notes
  • AI Chat
  • AI Rewrite
  • AI Translation
  • AI Study Tools
  • AI Quiz Generation & Flashcards

These features require an internet connection because the processing happens on DeepSeek's servers. If you never use them, no transcript text is sent to DeepSeek.

What information may be processed

When you trigger a cloud AI feature, VoiceScribe sends the transcript text you selected (for example, the transcript of a meeting or lecture, or text you typed) to the DeepSeek API to generate the result. We do not send your raw audio recordings. You can view the transcript text in the app before choosing to process it.

DeepSeek processes the text to return a summary, note, answer, translation, quiz or rewrite. We design these requests to contain only the content necessary to fulfill your request.

How you control AI processing

  • Every cloud AI action is user-initiated. Summaries, notes, chat, rewrite, translation, study tools and quizzes only run when you tap them.
  • You can review the source text first. The transcript that will be processed is visible in the app before you proceed.
  • You decide what to select. You choose which transcript or text gets processed.
  • Nothing happens in the background. There is no automatic or silent cloud processing.

How to avoid cloud AI processing entirely

If you prefer that no text ever leaves your device, you can use VoiceScribe as a fully on-device tool:

  • Record and import audio.
  • Use on-device transcription and the voice keyboard.
  • Read, edit and export your transcripts (TXT, PDF, DOCX, Markdown, SRT, VTT).
  • Simply do not tap any AI Summary, AI Notes, AI Chat, AI Rewrite, AI Translation, AI Study or AI Quiz action.

Used this way, VoiceScribe sends no transcript text to DeepSeek.

AI limitations

AI-generated content — including summaries, notes, translations, quizzes and chat answers — may contain inaccuracies or omissions. VoiceScribe's AI features are provided to assist you and should not be relied upon as professional, legal, medical, financial or other expert advice. Always review AI output before relying on it. See our Terms of Use for more detail.

Want to see exactly what's sent and to whom? Visit the Transparency & Data Sharing page.