Microsoft pushed an Insider Preview build to the Dev Channel on September 12, 2025, that quietly introduces Emoji 16.0 glyphs, a Copilot prompt box in Click to Do, and a sweep of Narrator accessibility improvements. Build 26220.6682 (KB5065782) arrives via the compact enablement-package model for Windows 11 25H2 and exemplifies the company’s current strategy: small, iterative updates that sharpen AI integration and polish everyday interactions. Unlike a feature-packed release, this flight focuses on making existing tools work better—particularly for users who rely on assistive tech or who frequently switch between gaming and productivity.
The update touches four user-facing areas: a curated set of seven Emoji 16.0 glyphs added to the Windows emoji panel, refinements to the Click to Do preview on Copilot+ PCs, targeted Narrator fixes for document reading and navigation, and a short-press/long-press remap for Xbox controllers that opens Task View. None of these changes reinvent Windows, but together they chip away at friction points that power users, gamers, and accessibility advocates have flagged for months. The controlled rollout model means early exposure requires the Dev Channel toggle, and even then, not every machine will see every feature at once.
Emoji 16.0 lands with seven new glyphs
The Windows emoji panel now includes a handpicked set of Emoji 16.0 characters: Face with Bags Under Eyes, Fingerprint, Root Vegetable, Leafless Tree, Harp, Shovel, and Splatter. The selection spans facial expressions, objects, nature, and symbols—Microsoft’s Fluent design team cherry-picked these to add expressive range without overwhelming the panel. Insiders on the Dev Channel with the 25H2 enablement package can summon them via the Win+. or Win+; shortcut, just like any existing emoji.
Behind the scenes, this signals continued alignment with the Unicode Consortium’s release cycle. Emoji 16.0 was finalized in 2024, and major platform vendors typically ship support within a year. Windows joins iOS and Android in offering the new glyphs, though visual style diverges sharply because Microsoft uses its own Fluent renderings. A splatter emoji on Windows looks distinct from its Apple or Google counterpart, and the leafless tree adopts a more stylized silhouette. Design consistency within the Windows ecosystem is high, but cross-platform messaging will still show platform-specific art.
Real-world compatibility is a patchwork. The emoji panel itself can display the new glyphs immediately after the update, but that doesn’t guarantee third-party apps will render them. Browsers like Edge and Chrome rely on their own font stacks, and until they bundle Segoe UI Emoji updates, users might see tofu (empty boxes) or fallback glyphs on web pages. Office apps fare slightly better because they pull from system fonts, but Outlook and Teams often cache older versions. Early Insider reports note mixed rendering—some messaging clients showed the new emoji perfectly, while others defaulted to generic placeholders. This is normal for a staged OS font update and not a bug with the build itself. For anyone planning to use the new emoji in broad communications, the advice is to test across recipient platforms first.
Click to Do gains a prompt box and on-device suggestions
Click to Do, Microsoft’s contextual Copilot overlay for Copilot+ PCs, gets three notable tweaks in this build. A text prompt box now appears in the Click to Do context menu, letting users type a custom command that is sent to Copilot alongside the selected screen content. Immediately below the box, suggested prompts appear—powered by an on-device language model rather than a cloud roundtrip. The suggestions aim to speed up common tasks like summarizing text, translating a phrase, or explaining a concept, and they work for English, Spanish, and French selections.
The distinction between on-device and cloud processing is the update’s most important privacy consideration. When generating suggested prompts, the model runs locally on the NPU, so no text leaves the device at the suggestion stage. However, once the user submits a prompt, Copilot may call cloud services to fulfill the request. For simple completions or knowledge retrieval, the response flows through Microsoft’s cloud infrastructure, and for image generation, it hits DALL·E. The exact data path depends on the action. Admins evaluating Click to Do for regulated environments should note that the initial suggestion generation avoids telemetry, but any subsequent cloud query could expose sensitive content. Enterprise Copilot governance controls—data residency, content exclusions, and audit logs—remain the recommended safety net.
UX polish is visible elsewhere. A new right-edge gesture animation provides clearer visual feedback when invoking Click to Do, addressing a long-standing complaint that the gesture felt ambiguous. Action tags like “Explain” or “Summarize” now surface popular Copilot actions, making the tool more discoverable. These are incremental improvements, but they build toward a Copilot experience that feels embedded in Windows rather than bolted on.
In a related experiment, the Start menu’s Recommended section may show Copilot prompt examples—gentle nudges like “Create an image of a mountain landscape” or “Summarize my last meeting.” The feature is lightweight and non-intrusive, but it hints at Microsoft’s broader plan to weave Copilot into every corner of the OS. The rollout is controlled and may not appear on all Dev Channel machines.
Narrator fixes tackle real-world reading pain points
The screen reader gets a meaningful refresh with fixes that touch continuous reading, table navigation, and voice consistency. Narrator users frequently report that continuous reading stops unexpectedly in long documents; this build corrects timing issues so that reading flows without interruption when using Natural Voices. Headings and grammar errors no longer trigger jarring pitch spikes, which had been a common source of fatigue. Footnote navigation now announces footnote numbers clearly and allows users to jump directly to the footnote text, and the focus tracking between document body and comments pane is more reliable.
Table and list handling sees multiple improvements. List items that wrap to multiple lines are read in full instead of cutting off, and list announcements are more consistent. In table scan mode, new keystrokes let users jump to the beginning or end of a row or column, with boundary announcements that reduce the risk of accidentally editing cell contents. These changes may sound technical, but for a screen-reader user editing a complex spreadsheet or a legal document with threaded footnotes, they slice away layers of friction. A Braille Viewer for refreshable Braille displays was also mentioned in adjacent Release Preview notes, offering additional output options for hardware users.
Accessibility advocates should treat this build as a milestone worth stress-testing. The Narrator improvements are not universally rolled out; they reach Dev Channel first and will likely stagger through Release Preview and stable over weeks or months. For organizations that depend on accessible document workflows, validating these fixes against real-world scenarios—multi-author Word files, dense PDFs, PowerPoint with complex tables—is essential. Microsoft’s Feedback Hub (Win+F) remains the primary channel for reporting gaps.
Xbox controller mapping: a long press for Task View
A subtle but immediately useful change remaps the Xbox button: a short press still opens Game Bar, but a long press now opens Task View instead of doing nothing. Pressing and holding the button still powers down the controller. For gamers who multitask—switching between a game, Discord, and a browser—the new mapping eliminates the need to reach for Alt+Tab or the keyboard at all. The change aligns controller ergonomics with Windows’ task-switching conventions and feels natural after a few minutes of use. It’s the kind of low-risk, high-convenience tweak that makes Dev Channel builds worth the occasional instability.
Controlled rollout and known rough edges
Every feature in KB5065782 rides on Controlled Feature Rollout (CFR). Users must enable the “Get the latest updates as soon as they’re available” toggle in Windows Update to receive early bits, and even then, server-side flags determine which features light up. Copilot+ PCs are required for Click to Do improvements. Some regional restrictions apply—certain Click to Do capabilities are gated in the European Economic Area and China due to regulatory considerations.
Microsoft’s release notes call out several known issues that Insiders should expect:
- The right-edge gesture for Click to Do may display swipe visuals on the wrong monitor in multi-display setups.
- Lock screen media controls can fail to appear.
- Windows Studio Effects may cause camera preview failures with some external webcams.
- Preview features are not fully localized, so users may encounter strings in English on non-English builds.
These are typical preview bugs and not showstoppers, but they underscore the value of testing in non-production environments.
Privacy, compliance, and the on-device shift
The local model driving Click to Do’s suggested prompts is part of a broader trend toward on-device AI in Windows. By running small language models on the NPU, Microsoft reduces latency and keeps sensitive text off the wire during the suggestion phase. However, the Copilot ecosystem still leans heavily on cloud services for full responses. IT administrators must assess whether the content users select—contract drafts, HR data, proprietary code—is appropriate to send to a cloud model. Microsoft provides tenant-level controls for Copilot, but the responsibility for classifying data and setting policies remains with the organization.
Emoji additions, while not a direct security concern, introduce a compatibility vector. Font updates can trigger rendering quirks in legacy apps that rely on custom text engines. Enterprises should test emoji rendering in critical line-of-business applications before deploying en masse.
Testing checklist for Insiders and IT
For those evaluating this build, a structured approach captures the most value:
1. Enable the Dev Channel toggle and install KB5065782.
2. Open the emoji panel (Win+.) and verify the new glyphs appear; then test rendering in Outlook, Word, Edge, Chrome, and Teams. Note any missing glyphs or fallback characters.
3. On a Copilot+ PC, select text, invoke the right-edge gesture, type a custom prompt, and assess suggested prompts in English, Spanish, and French. Confirm whether the suggestions remain local (no network activity) and whether cloud calls occur after submission.
4. Launch Narrator (Win+Ctrl+Enter) and open a complex Word document with footnotes, multi-level lists, and tables. Test continuous reading, footnote navigation, table scanning (Ctrl+Alt+arrow keys), and comment pane focus. Report any pauses or misreads.
5. Connect an Xbox controller, press the Xbox button briefly to open Game Bar, long-press to open Task View, and hold to power off. Verify behavior is consistent.
6. Log all observations in Feedback Hub, including repro steps and hardware specs.
What Microsoft got right—and where risk remains
The build excels in accessibility polish and pragmatic AI integration. The Narrator fixes are tangible improvements that address top user complaints, and the on-device suggestion model for Click to Do strikes a sensible balance between utility and privacy. The Xbox controller remap and Start menu prompts are low-risk, high-gain additions that won’t disrupt workflows.
On the risk side, the staggered rollout guarantees a fragmented experience. Not all Insiders will see the same features at the same time, making it difficult to gather consistent feedback. Cross-app emoji rendering will remain inconsistent for weeks or months, depending on how quickly third-party apps adopt updated system fonts. The privacy boundary between on-device suggestions and cloud-powered responses needs clearer communication from Microsoft—users may not realize that a simple “Explain this” prompt sends selected content to the cloud.
Recommendations and the road ahead
For individual Insiders, the build is worth installing on a secondary device to explore the new Copilot interactions and Narrator improvements. Back up critical data and accept that preview software carries minor instability. Accessibility testers and caregivers should prioritize Narrator validation in real productivity scenarios—Microsoft’s fixes are promising, but real-world confirmation is what will drive further polish.
IT administrators tasked with managing Copilot deployments should use this build as a sandbox for evaluating data flow risks. Test Click to Do with representative sensitive data and verify that tenant-level controls intercept or block unwanted cloud calls before the features reach a wider audience.
Looking forward, the feature set here is a bellwether for Windows 11’s 25H2 release. Emoji 16.0 support, refined on-device AI, and deeper accessibility tooling point to a mature OS that evolves through quiet enablement packages rather than disruptive overhauls. The community should watch for broader rollout to Beta and stable channels, application adoption of new emoji fonts, and the next wave of Copilot governance tooling. For anyone tracking Windows 11’s AI and accessibility trajectory, KB5065782 is a small but significant waypoint.