To get Windows 11's Narrator feature (on non-Copilot+ PCs) to describe an image, you have to use the Copilot app, which defeats the whole purpose.
With
Windows 11 version 26H2 out, I figured I should do some quick guides, just to put the new features that Microsoft highlights as important parts of the update, even though they were actually released before, in front of people. It should’ve been easy work for me, just talk about a feature, list the steps, and tell you why it’s useful.
But as it turned out, writing guides on the latest Windows 11 novelties isn’t the walk in the park I expected it to be. Because just as I’m about to write a guide, I find a limitation that renders that guide obsolete for a big chunk of users.
My initial idea was to start with a guide on how to disable the AI agent in Settings, as not too many people are thrilled with it. As it turns out, the majority of users can’t even disable it, because the policies for doing so are only available to Windows 11 Enterprise and Education users. So, if you’re on Windows 11 Pro or Home, you just have to accept that AI search is now a permanent part of the Settings app. The Settings AI agent situation prompted me to do this report.
Then, I wanted to test the Narrator app, which can now use AI to describe images. But actual voice description is only available on Copilot+ PCs, because it uses the device’s NPU to process the image locally and read the generated text back. On a regular Windows 11 PC, though, the experience can be very different, and the way Microsoft handles it prompted me to write this article.
Since local AI generation doesn’t work on regular Windows 11 PCs, and Microsoft still wanted to give users at least something, it decided to hand image descriptions over to the Copilot app. So, if you press
Narrator key +
Ctrl +
D on a focused image, Windows will open Copilot with that image attached.
This already isn’t very useful, but it gets worse. On my non-Copilot+ PC, Copilot opened with the image attached and the following prompt already entered:
“Review this image attached and suggest five ways you can help me work with it. Keep each suggestion to a one-line bullet point.”
As you can see, that isn’t a prompt that produces an image description.
Narrator just passes images to the Copilot app So, if you want to get an image description, you have to open Narrator, focus the image, use Narrator’s image-description shortcut, get thrown into Copilot, delete or replace the existing prompt, and then ask Copilot to do the thing you originally asked Narrator to do.
At that point, why involve Narrator at all? If the image goes through Copilot, why not open the Copilot app itself, attach the image, and ask the assistant to describe it? Why even bother going through the entire Narrator workflow at all? And assuming this feature is mainly built for users with vision problems, I imagine they’d have a great time doing all of this.
The funny part is that Microsoft could make this workflow at least more bearable by just changing the default prompt. If Copilot opened with something like “
Describe this image in detail” already filled in, the user could simply press Enter once more. Copilot would generate the description, which could then be heard through Narrator or Copilot’s own Read aloud feature. Literally the simplest possible change could make the difference.
To top it all off, since the entire thing goes through Copilot, users are basically forced to send their images to Microsoft’s servers. Sure, importing an image only attaches it to Copilot and doesn’t send it automatically, but that makes no difference, because you have to send it if you want anything to get done with it.
In the end, I’m not using this feature anyway, so it doesn’t make too much of a difference. But it does make a difference for users who actually need it. And for me, I can’t help but notice that the more you look at individual features in Windows 11, especially the AI-powered ones, the more inconsistencies you see.
source
