Microsoft has pushed generative AI features into every corner of Windows 11, but most of them aren’t actually useful to most people, and some of them are actively distracting or harmful. Thankfully, the latest addition is actually great: a ‘Describe image’ feature.
Microsoft is rolling out Windows 11 Insider Preview Build 26200.5702 (KB5062653) to the Dev Channel forInsider testers, with a new ‘Describe image’ option in the Click to Do menu. When you select the menu option, Windows will create a text description of the selected image, with a one-click button to copy it to your clipboard. This works a lot like Google Lens and other object recognition software, but it uses a local AI model running on your device, so the images are not sent to any external server during processing.
Describe image is in the sameClick to Do menuthat already has options for blurring image backgrounds, erasing objects, and other actions. You can open the menu on a compatible Windows 11 PC by pressing the Q and Windows keyboard keys together, or clicking the mouse while holding down the Windows key.
So, why is this useful? Well, if you have limited vision, this feature can describe any image or visual content as spoken text. This could also be a great way to make image descriptions for documents, articles, presentations, art, or anything else with visual elements. For example, every image in every article I write here atHow-To Geekneeds analternative text description, so screen readers and search engines can understand them. Most other publications and websites do the same thing for images.
I builtmy own browser extensiona while back for quickly writing image descriptions, using OpenAI’s GPT cloud model, and it looks like this Windows feature creates similar results. The main catch is that even after it rolls out to stable Windows 11, it will require aCopilot+ PC. This feature can definitely work on any PC with a moderately-powerful GPU and enough RAM. For example, you can get similar results withLM Studioby dragging an image into the chat window, after downloading vision-compatible model likeGoogle Gemma 3 4B.
Microsoft said in the blog post, “The new describe image action in Click to Do is rolling out to Windows Insiders on Snapdragon-powered Copilot+ PCs, with support for AMD and Intel-powered Copilot+ PCs coming soon.”
The Insider build has some other useful changes, like an updated design for privacy dialogs, an ‘Administrator protection’ feature for performing admin actions, the deprecation of PowerShell 2.0, and more. Many of these updates already showed up in other preview builds in the Insider channels.
