Apple Overhauls Genmoji in iOS 27: New Editing and Privacy Concerns

Apple yesterday confirmed a complete overhaul of Genmoji at WWDC, replacing the feature's one-shot generation model with an iterative editing system that lets users refine results step by step. The iOS 27 Genmoji redesign is the biggest structural change to the feature since it launched. Separately, pre-release reporting from Bloomberg's Mark Gurman suggests Apple is also developing a proactive suggestion engine that would generate custom emoji based on typing habits and Photos library content, though Apple did not announce that system at WWDC and its final form remains unconfirmed.

That distinction matters. The iterative editing overhaul is confirmed news. Suggested Genmoji is reported behavior from a leak, with settings strings as supporting evidence. The two developments are connected, but they sit at different levels of certainty, and readers deserve to know which is which.

What was broken, and why it stayed broken

Genmoji debuted in iOS 18.2 as part of Apple Intelligence in December 2024. The premise was straightforward: describe an emoji, get a custom AI-generated result. The execution was not. Visual quality was "decidedly mixed" when the features launched, MobileSyrup reported last month, and Genmoji had "not exactly taken off" among users, Trusted Reviews noted. Specific problems included imprecise prompt handling, heavy battery consumption from the on-device model, and output that was often visually off, Heise reported.

iOS 26 addressed some of that. Apple added more expressive prompt controls, letting users describe emotions like "shocked" or "sleepy," specify physical traits including hairstyles, glasses, and accessories, and blend two emoji styles into a single result. Tapback reactions got Genmoji support too, Trusted Reviews and Notebookcheck reported earlier this year. The underlying model also improved.

None of it fixed the core problem. Every attempt at a Genmoji was independent. If the result was close but wrong, there was no way to nudge it in the right direction. Users could only start over. iOS 27 is built specifically around that gap.

How the new Genmoji creation experience works in iOS 27

The confirmed overhaul centers on iterative editing, and the mechanics are a meaningful shift from how Genmoji worked before.

The redesigned interface offers four starting points: a text description, an existing emoji, an image from the Photos library, or a specific person. From any of those, users generate a first result. What happens next is new. A "Describe a change" option lets users modify what's already been created, adjusting colors, swapping objects, or adding elements, without regenerating the entire Genmoji from scratch. Each revision builds on the previous version as a base, MacRumors reported. The difference between that and the old model is the difference between steering and gambling.

Stacking multiple requests can produce complex, multi-element results, which the original version had no reliable way to support. The interface also surfaces suggestions on what to try next, reducing the guesswork for users who aren't sure where to take a design, MacRumors noted. Additional emoji can be folded into the creation flow at any stage.

Output quality has reportedly improved as well. Genmoji now default to a 3D cartoonish style that looks more consistent with the standard emoji set, with sketch and drawing available as alternatives. Generation is reportedly faster and less battery-intensive than before, though no benchmark figures have been published to support those claims, according to MacRumors. Both Genmoji and Image Playground run on updated Apple Foundation Models, which Apple confirmed at WWDC.

The Image Playground overhaul runs parallel to the Genmoji changes. It now supports photorealistic image generation and selective AI editing within existing photos, a capability that didn't exist at launch.

Suggested Genmoji: what the reporting says, and what it doesn't

The more strategically ambitious piece of the iOS 27 Genmoji story isn't in the confirmed creation workflow. It's in what Bloomberg's Mark Gurman reported Apple is developing: a system-wide suggestion engine that would work like emoji autocomplete, except instead of pulling from a fixed library, it generates custom options on the fly.

According to reporting by Trusted Reviews and Notebookcheck, the system learns from how users type and what's in their Photos library. In practice, that could mean typing a strong reaction in Messages and having iOS surface a personalized Genmoji that matches the tone, or seeing one generated from a face or moment the system recognizes in the photo library, without the user initiating any request. The goal, as Trusted Reviews framed it, is to make Genmoji feel like something that appears when needed rather than a tool users have to go looking for.

Settings strings visible in iOS 27 pre-release builds add a layer of specificity. "Suggested Genmojis are created from your photos and frequently entered sentences," the text reads, Heise found. That's not a marketing statement; it's a functional description of what the system would access. The feature is reportedly optional, with a toggle to disable it, and may not be active immediately after launch, Heise reported.

The privacy picture has a significant gap: the original leak does not clarify whether the suggestion system would run entirely on-device, Notebookcheck noted. That's not a minor detail. Apple has positioned on-device processing as the foundation of Apple Intelligence's privacy model. Apple did confirm at WWDC that it collaborated with Google and the Gemini family of models to develop the next generation of Apple Foundation Models, TechCrunch reported. Whether that collaboration extends into how Genmoji suggestions are processed hasn't been addressed.

Two other longstanding problems also remain unresolved. When Genmoji's content filters block a harmless prompt, Apple currently offers no way for users to flag or dispute the error, unlike providers such as OpenAI, Heise noted. And certain generation flows require users to submit a photo of a recognizable person before proceeding, a step many users may be reluctant to take, Heise reported. Neither issue appears to have a fix on the horizon.

What beta testing needs to resolve

iOS 27 will support all iPhones from the iPhone 11 onward, which Apple described as making it available to more users than any previous iOS release, TechCrunch reported. That breadth raises the stakes on both sides of the Genmoji story.

The iterative editing model has a clear, testable premise. Does the "Describe a change" interface reliably hold state across multiple refinements, or do results degrade after a few rounds? Do the visual quality improvements hold up at scale across different types of prompts, or are they meaningful only in controlled demos? Those questions have answers that beta testing will surface quickly.

Suggested Genmoji is harder to evaluate without confirmed details. The most consequential open question isn't about user experience; it's about processing. If Apple confirms on-device handling for the suggestion system before public release, the privacy concern shrinks considerably. If that confirmation doesn't come, the feature's default status and opt-in framing become much more significant. A system that reads typing habits and photo libraries passively, without a clear explanation of where that processing happens, is a different proposition than one that keeps everything local.

The moderation dispute gap is the quieter unresolved issue. It doesn't generate the same attention as the privacy questions, but for users who get blocked on a legitimate creative request, it's a practical dead end. Whether Apple addresses that before iOS 27 ships publicly remains to be seen.

The iterative editing overhaul is defensible on its own terms, and the confirmed changes are substantial. Whether Suggested Genmoji earns the same verdict depends entirely on what Apple discloses, and when.