Guide
Is Wispr Flow private?
It depends what you mean by private. Wispr Flow has real controls and zero-retention agreements, but it transcribes in the cloud, so your voice leaves your Mac. Here is what its own documentation says.
Short answer: Wispr Flow transcribes your speech on its servers, so your audio leaves your Mac. With Context Awareness on, which is the default, text from the active app is sent with each dictation too. It has settings to limit this and says its AI providers do not retain it. If your speech must stay on your Mac, use an on-device app.
| Feature | Precis | Wispr Flow |
|---|---|---|
| Where speech becomes text | On your Mac | On Wispr's servers. Its documentation says transcription always happens in the cloud. |
| Does your audio leave the Mac? | No. It's deleted as soon as the words are typed. | Yes, it is sent to Wispr to be transcribed. |
| What it reads to adapt the text | The frontmost app, and the tab title in Safari, Chrome and Arc, read on your Mac. Nothing is sent. | With Context Awareness on (the default): app and website, text around your cursor, on-screen text, file and variable names in coding apps, and conversation history. With it off, the app name and website domain are still sent. |
| Screenshots | Never taken | Wispr's documentation says Context Awareness works from text and does not take screenshots. |
| Stored on a server | No. History is text on your Mac. | A Dictation Cloud Storage setting controls whether Wispr stores transcripts, audio and history on its servers. |
| Used to train models | No. Nothing is sent. | The setting "Improve the model for everyone" is on by default for Free and Pro accounts, with the choice shown during onboarding, and off for Enterprise. Wispr says its third-party AI providers may not train on your data. |
| Works with Wi-Fi off | Yes | No, it needs a connection |
Wispr Flow details from its own documentation: the Context Awareness article, data controls and HIPAA article, read on 5 October 2026. Wispr changes these pages often; check them for the current wording. Product names belong to their owners.
What Wispr Flow does well
- Real controls. You can turn Context Awareness off, turn model improvement off, and choose whether dictations are stored on Wispr's servers.
- Zero-retention agreements. Wispr says it holds zero data retention agreements with the third-party AI providers it uses for dictation.
- Sensitive fields. It excludes password fields, numeric-only fields and browser address bars, and skips context reading in banking and financial apps.
- Healthcare and enterprise options. A HIPAA Business Associate Agreement is available, and Enterprise admins can enforce settings for everyone.
What no setting changes
Wispr Flow transcribes in the cloud, so your voice leaves your Mac every time you dictate. Its protections are about what happens to the audio afterwards. If your requirement is that speech never leaves the machine (client confidentiality, a company policy, a locked-down network), a cloud app can't meet it, however it is configured.
What Precis sends
Speech recognition (NVIDIA Parakeet or OpenAI Whisper) and the rewriting that shapes text for each app (Apple Intelligence) run on your Mac, and the audio is deleted as soon as the words are typed. The app goes online for only three things: downloading a speech engine once, confirming your licence key about every two weeks, and checking for updates. The privacy page lists them. Turn Wi-Fi off and dictation keeps working.
A correction worth knowing
Some articles say Wispr Flow sends a screenshot with each dictation. Its documentation today says Context Awareness works from text and does not take screenshots, so I don't claim that here. The text it does send is still substantial, as the table shows. I'm reporting what the documentation says today; I haven't audited the app's network traffic.
When a cloud app is the right call
If you dictate on Windows or a phone as well as a Mac, or your Mac can't run macOS 26.5, Wispr Flow covers that and Precis doesn't. Choosing it is a reasonable trade when you're comfortable with the controls above.
Questions
Does Wispr Flow send my audio to the cloud?
Yes. Wispr's documentation says transcription always occurs in the cloud, so your audio is sent to Wispr to be turned into text.
Does Wispr Flow take screenshots?
No, according to its current documentation: Context Awareness works from text and does not take screenshots. It can send text from the active app, such as the text around your cursor and on-screen text, when Context Awareness is on.
Can I stop Wispr Flow reading my screen?
You can turn Context Awareness off in its settings. The name of the app and the website domain are still sent, and the text around your cursor is read on your Mac only.
Does Wispr Flow train on my voice?
It depends on a setting. "Improve the model for everyone" is on by default for Free and Pro accounts, with the choice shown during onboarding. When it is on, Wispr says your audio, transcript and edits may be used to improve its models. Its third-party AI providers may not train on your data.
Is there a Mac dictation app that never sends audio anywhere?
Yes. Precis, VoiceInk, MacWhisper and Apple Dictation (on Apple silicon, for supported languages) can run on your Mac. Precis does speech recognition and rewriting on the Mac and deletes the audio after typing.
Try Precis free for 7 days
Hold ⌥ Space in any app and talk. No card and no account, then $19 once for 2 Macs.
Free for 7 days, no card. Needs macOS 26.5 on Apple silicon.