Whisper One
All guides

Troubleshooting & your data

History, audio, and privacy

Understand local history, cloud transcription, temporary audio, and external AI tools.

What history contains

History stores transcription text, time, and recording duration locally on this Mac, separated by account. The three newest saved dictations also appear on Home. This is not a cloud history sync feature. Generated answers or translations may not be saved with the recognized speech, so copy a needed result before its floating window closes.

Where audio goes

Speech recognition uses a cloud speech service, so recording is not an offline-only workflow. Fn dictation can write a temporary audio file on this Mac. Current source includes cleanup after use and stale-file cleanup, but installed versions and abnormal termination can differ. Do not treat temporary files as a promise that audio never touches disk or is always deleted immediately. Consult the published privacy policy for the applicable service terms and data handling.

Recover or remove saved text

  1. Open History and locate the saved transcription.
  2. Select it and use Command + C to recover the text if insertion failed.
  3. Use the available history deletion controls to remove local entries you no longer need. Check the result afterward.

Removing local history does not retract a message already pasted elsewhere or delete data from another service.

External AI tools are separate

When you submit dictated text to Claude or ChatGPT, you also share it with that service under its own account settings and policies. Only include material you are allowed to share. Dictation does not automatically read the rest of the conversation, your screen, or attachments.

A privacy-conscious practice example

Draft a meeting invitation using placeholder names. The topic is a project check-in. Leave the date and meeting link blank.

Use fictional names when learning. If an optional analytics control is available in your installed build, review it in Settings; changing analytics does not turn off the cloud processing required for speech recognition. For the website's current privacy terms, use the Privacy policy link in the footer.