Apple's iOS 27 Photos AI expands and reframes images with limits

At WWDC, Apple unveiled generative AI features for the Photos app in iOS 27, led by camera chief Jon McCormack. The new Extend tool lets users expand a photo by 25 percent, with AI filling in background pixels based on what it believes should be there. Spatial Reframe changes the perspective of an image, generating fake pixels for elements outside the original frame. The existing Clean Up tool also gets a boost from improved AI models, but remains restricted to removing unwanted background objects—not the primary subject. McCormack emphasized that these features are designed to solve compositional problems, not to enable wild fakery.

Apple is taking a cautious stance compared to rivals. The Extend tool works only once per image to prevent infinite expansion. Later this year, Apple will integrate Google DeepMind’s SynthID technology to embed invisible watermarks in AI-edited photos, allowing platforms to flag alterations. Product manager Della Huff said the team trained models to minimize hallucinations, ensuring the AI doesn't create objects that shouldn't be there. McCormack reiterated that a photograph should capture something that actually happened, and these tools aim to preserve the “sanctity of that moment” while giving users compositional superpowers without needing Photoshop.