Source: https://saypad.app/is-wispr-flow-safe

# Is Wispr Flow safe?
Wispr says it's SOC 2 Type II certified and encrypts your data in transit and at rest. The bigger question is what you're comfortable sending, because Wispr's transcription always happens in the cloud. Here's what its own pages say and which settings to check. We make Saypad, a competing local app, so read Wispr's privacy and data controls pages yourself as well.

## What happens to one dictation

Wispr's privacy page walks through it. Audio is captured on your device once you press the shortcut. It's sent to Wispr's servers over TLS 1.2 or later and turned into text. Then the audio is deleted, unless you've turned on Dictation cloud storage.

## Four settings to check

You'll find them under Settings → Data and Privacy, apart from the last one.

- **Improve the model for everyone.** Lets Wispr use your audio, transcripts and edits to evaluate, train or improve its models.
- **Dictation cloud storage.** Keeps transcripts, audio and history on Wispr's servers so they follow you across devices.
- **Context awareness.** Reads limited text from the active app window.
- **Auto-add to Dictionary** (Settings → Personalization). Watches the text box after pasting, so that a word you correct gets added to your dictionary.

## What the settings don't change

- Your speech still goes to Wispr's servers on every plan.
- Wispr may collect usage statistics such as how many words you've dictated, whatever your data controls say.
- Rating a dictation with a thumbs up or down, or sending feedback, shares data from that session. It may include audio, transcripts and your edits. Wispr says it uses this to train its models and that authorized staff may review it.

## Third parties

Wispr says it has zero data retention agreements with its third-party AI providers for dictation and that they can't train on your data. It notes one exception on retention: some features, such as Notetaker briefs, rely on provider features that zero retention doesn't cover. Training is still barred in those cases.

## Work and health data

Individuals can accept a Business Associate Agreement in the app, and Enterprise admins can enforce it for a whole organization. Wispr also offers its compliance report on a trust page. Whether that's enough for your job is a question for your security or legal team, and a vendor page can't settle it. Ask where audio is processed, who can see it and whether your policy allows it at all.

## If nothing should leave your Mac

Then the answer has to be a local tool, whatever a cloud service's policy says. Apple Dictation runs on the device for many languages, and Keyboard settings show whether yours does. Saypad, which we make, recognizes speech on your Mac and works with no internet. It has no AI cleanup, no iPhone or Windows app, and it needs an Apple silicon Mac with macOS 15 or later. If you decide to stay with Flow, [Wispr Flow pricing](wispr-flow-pricing) lays out the plans.

From wisprflow.ai/privacy and wisprflow.ai/data-controls, checked 6 November 2026. Policies change, so confirm them before relying on this.

## Questions

**Does Wispr Flow record everything I say?**
Wispr says nothing is captured until you press your shortcut. After that the audio goes to its servers for transcription.

**Does Wispr Flow use my voice to train its models?**
It can. A setting called Improve the model for everyone lets Wispr use your audio, transcripts and edits. Turn it off in Settings → Data and Privacy. On Growth and Enterprise plans it's off and locked. The pages we read don't say what the default is on Free and Pro, so check yours.

**Does Wispr Flow read my screen?**
With Context Awareness on, it reads limited text from the app you're dictating into, so names come out right. You can turn that off. Wispr also says it always uses the app's name and the surrounding text box to format your text.

**Does Wispr Flow work offline?**
No. Wispr says transcription always happens in the cloud.