Apple's new AI photo editing tools in iOS 27: mostly effective, but raise concerns
The iOS 27 developer beta introduces three AI-powered photo editing features—Clean Up, Extend, and Spatial Reframing—that largely work well, yet the author highlights potential issues with authenticity and the uncanny valley effect.
New AI capabilities in iOS 27 beta
Apple has integrated serious AI photo editing into the world’s most popular camera for the first time. The iOS 27 developer beta includes three tools: Clean Up, Extend, and Spatial Reframing. While not as aggressive as Google’s Pixel offerings, they mark a turning point for what iPhone users can do natively.
Clean Up gets a major upgrade
Clean Up removes unwanted objects from photo backgrounds. The previous on-device-only version was poor, leaving artifacts. Now Apple uses cloud-based models, similar to Google’s approach, resulting in convincing edits. The author notes this is the least ethically troubling AI tool and demonstrates it by removing a stranger from a background and a smudge from a child’s face.
Extend: reverse cropping
Extend expands the edges of a photo, adding AI-generated plausible filler. It has limitations: it avoids editing people, sometimes only allows expansion in one direction, and adds only a small amount. In one example, the tool added a side mirror to a rally car that was originally out of frame. However, it can also invent objects, like a potted plant on a side table—convincing but not real.
Spatial Reframing: 3D perspective shift
This tool adds a third dimension, allowing users to adjust perspective as if moving the camera. The range is limited to about arm’s length. When applied to a photo of Apple executives, the AI generated a fictional person next to Craig Federighi. Closer subjects produce more distorted, uncanny results, especially on faces. The author finds the concept appealing but the execution uncomfortable.
Labels and trust issues
Images edited with these tools receive Synth ID labels. Instagram surfaces this via an “AI Info” menu, but only if tapped. The author warns that even small doubts about a photo’s authenticity can accumulate, eroding trust in photography over time.
