Microsoft has rolled out a new Windows 11 Canary channel build that embeds AI-powered image editing and visual search directly into File Explorer’s right-click menu, alongside new privacy controls and a long-requested seconds display in the notification clock. These changes, while individually small, signal a strategic shift toward platform-level AI integration that could reshape how users interact with their files and manage generative capabilities.

AI Actions Arrive in File Explorer

The most eye-catching addition in this Canary flight is the new “AI actions” submenu, which appears when you right-click an image file (JPG, JPEG, or PNG) in File Explorer. This menu offers four distinct workflows:

  • Bing Visual Search: Sends the image to Bing’s image search engine to find similar pictures, shopping results, landmarks, or extract text—without requiring a manual upload.
  • Blur Background: Opens the Photos app with the image loaded and automatically detects the subject, allowing you to apply a customizable background blur or refine the effect with a brush.
  • Erase Objects: Launches a generative erase tool in Photos to remove unwanted elements from the picture.
  • Remove Background: Uses Paint’s one-click background removal to create an instant cutout of the primary subject.

These actions are designed as shortcuts to existing capabilities in Windows’ built-in apps rather than full editors inside File Explorer. When you click an option, the relevant app (Photos or Paint) opens with the edit already staged, or a quick model-driven preview appears before saving. At launch, only common raster formats are supported; RAW, PSD, TIFF, and other professional file types are not yet compatible, which may limit utility for photographers and designers.

One major caveat is ambiguity around where the AI inference actually happens. Microsoft has indicated that some operations can run locally on-device—especially on Copilot+ PCs with a dedicated NPU—while others may rely on cloud processing or require specific Microsoft 365 and Copilot licenses for advanced features. However, the exact execution path often depends on hardware, licensing, and server-side toggles that Microsoft controls, leaving users uncertain whether their image data is processed locally or sent to the cloud.

A Privacy Dashboard for Generative AI

To accompany the new AI features, Microsoft has added a “Text and image generation” settings page (under Privacy & security) that aims to give users visibility into which apps are using Windows’ on-device generative models. The page lists recent activity from the past seven days, showing each app that requested generative AI access, and provides per-app toggles to allow or block that access.

Enterprise admins can also manage these permissions via Group Policy or MDM, which is a critical step toward governance. However, the dashboard has notable limitations: it logs that an app requested generative capabilities but does not reveal what data was sent, where inference occurred, or what was retained. Moreover, disabling the toggle only blocks apps from using Windows-provided local models—it does nothing to stop apps from using their own cloud-based AI services. For regulated data, IT teams will need to treat this as one layer of control among many.

The Seconds Clock Makes a Comeback

In a smaller but welcome change, Windows 11’s Notification Center can now display a larger clock that includes seconds. The option is found at Settings > Time & language > Date & time > “Show time in the Notification Center.” When enabled, the flyout shows a full HH:MM:SS readout above the calendar, restoring a feature many power users missed since Windows 10. This is particularly handy for scripts, troubleshooting, and any task requiring second-level precision.

Strategic Implications: AI as a Platform Service

These updates collectively reflect Microsoft’s ambition to weave generative AI into the very fabric of the operating system. Rather than confining AI to standalone applications like Paint or Photos, the company is testing ways to surface it as a contextual, one-click utility where users already work—in File Explorer, the Settings app, and the notification area. This approach reduces friction for common tasks and positions Windows AI models as platform services that any app can invoke.

The introduction of the privacy dashboard is equally significant; it acknowledges that platform-level AI demands platform-level transparency. While still rudimentary, the dashboard provides a foundation for future auditing tools, telemetry exports, and retention policies that enterprises will demand before broad deployment.

What Should Windows Insiders and IT Pros Do?

For enthusiasts testing this Canary build on non-production hardware, the new features are worth exploring with non-sensitive images to understand the workflows. Navigate to the “Text and image generation” privacy page after using an AI action to verify that logs appear. If you need the seconds clock, the toggle is easy to enable.

For IT administrators and security teams, the picture is more complex:
- Pilot before rolling out: These are experimental Canary features subject to change; run them on isolated machines first.
- Validate execution locality: Before processing any regulated data, confirm with Microsoft documentation whether inference is local and non-retained. If guarantees are absent, block the capability via policy.
- Monitor app activity: Regularly check the new privacy page’s Recent activity logs to spot unauthorized usage of generative models.
- Leverage policy controls: Use the Group Policy or MDM settings to deny access to system AI models until you’ve established clear governance.
- Educate users: Ensure staff understand which data is safe to process with one-click AI edits and which apps are approved.

The Road Ahead

Microsoft will likely expand these experiments in coming months. Expect File Explorer to gain actions for document summarization, list generation, and deeper Copilot integration—tied increasingly to Microsoft 365 subscriptions and Copilot+ hardware. More granular audit capabilities in the privacy dashboard and extended file format support are also probable, though timelines remain fluid given the Canary channel’s nature.

The ultimate success of this strategy hinges on Microsoft’s ability to pair convenience with ironclad transparency. Clear, verifiable statements about where inference runs, what data is transmitted, and how long artifacts are retained will determine whether IT departments embrace these features or lock them down entirely.

For now, the Canary build offers a tantalizing preview of a Windows desktop where AI assistance is never more than a right-click away—but with enough unanswered questions to warrant cautious optimism rather than wholesale adoption.