Apple adds on-device generative AI to iOS Photos with non-destructive editing stack

Design · 5 min read

In the new iOS developer beta announced today, Apple introduced an on-device generative AI editor for Photos that creates context-aware variations and fills image regions while preserving the original file. Apple emphasizes privacy by performing inference locally on recent A-series and M-series chips, and by exposing user controls to opt out or delete generated artifacts.

Designers will find the update significant because edits are stored in Photos’ non-destructive layered history, allowing creators to revert or tweak AI-generated changes just like any other adjustment. The UI integrates directly into the existing edit workflow, with a small contextual panel for style presets and tool masks.

Developers can extend the feature through new APIs in the Photos framework, enabling third-party apps to invoke the on-device generator and import or export edit stacks. Apple says the APIs are sandboxed, and developers must request a new permission scope for invoking local generation on user photos.

Early feedback from design communities highlights the potential to speed mockup creation while raising questions about version provenance and asset attribution. Apple plans a fuller rollout in the next public iOS release and will share more developer documentation at the upcoming WWDC follow-up sessions.