Is free AI dictation private?

Not automatically. Price does not decide privacy. What decides it is where your speech is recognized and whether the audio or the transcript is kept, shared or used to train a model. A free tool that runs on your device keeps your voice there. A free tool that streams audio to a company gives that company your audio, and the company's own terms decide what happens to it next.

Dictation that runs on your device keeps the audio local

On a Mac, Apple says that voice inputs and transcripts for general text Dictation are processed on your device and not sent to Siri servers ¹. You can check which mode you are in. Apple says your device shows in Keyboard settings whether dictated audio and transcripts are processed on the device or sent to Apple servers ². Look there before you dictate a sensitive document, because a setting can change after an update.

Free apps that run a local model can offer the same protection. The free plan of Superwhisper uses local Whisper models without AI processing, and the company says nothing dictated leaves the device ³. The company's statement is a claim you can test, not a guarantee, so read its current privacy page as well.

Dictation in the cloud sends your audio to a company

When you use a cloud service, the audio leaves your device. OpenAI says that when you press the microphone icon in ChatGPT, the recorded audio is sent to its models to be transcribed ⁴. The text can be edited before you send it, but the audio has already left your phone or computer.

Storage and training are a second question. OpenAI's dictation FAQ says that consumer users who have chosen to share audio to improve its models may have their dictation audio used for training ⁵. A consumer account that never shares audio is treated differently from one that does, so check the sharing setting in the app.

Apple's own dictation has a similar split. If you allow an app to use Speech Recognition for transcription, Apple says the audio to be transcribed may be sent to Apple ⁶. The permission you give a third-party app is what allows that transfer, so read each permission request before you accept.

On Windows, the feature you use decides the answer

Windows has two different speech tools, and they do not share a privacy answer. Windows voice typing uses online speech recognition, which Microsoft says is how it converts speech to text ⁷. Device-based speech recognition in Windows 11 is different: Microsoft says it processes the voice on the device and sends no voice data to Microsoft ⁸.

The difference matters for a free tool. Voice typing that you start from the keyboard is not the same as the device-based recognition in the speech settings, so confirm which one you are using before you assume your words stay local.

It depends: the same name can mean different privacy

The table shows that a free tool can be in either column. The name of the feature tells you little, so the mode and the setting tell you more.

What to do before you dictate anything private

1. Find the processing setting. On a Mac, open Keyboard settings and read the indicator for dictation ². In other apps, look for an on-device or offline option in the settings.
2. Read three answers in the privacy page. Ask where the speech is processed, whether the audio or transcript is kept, and whether it is used for training. If the page does not answer all three, treat the tool as cloud-based.
3. Decline any option to share audio for model improvement, unless you have decided that sharing is acceptable for that material.
4. Check the permission before you let a third-party app use speech recognition, since the audio may go to the platform's company.
5. Keep sensitive material on the device. A free cloud app is not a good place for client, patient or legal material. A free app that runs locally, or the built-in dictation with its processing set to the device, is the safer route. For a longer answer on the risks of cloud dictation, see whether AI dictation apps send your text to the cloud, and on the trade-off between local and cloud recognition, see whether offline speech recognition is more private.

For iPhone specifically, see is iPhone dictation private, which covers the details.

Yap, which we make, runs its dictation on the device on Mac and Windows, so it sits in the on-device column above. Check its current details on the download page before you rely on it for anything sensitive.

For the broader trade-off between who pays for free dictation and what you give up, see the guide to AI dictation costs, and the question on whether voice typing is free for the built-in options.