Apple pivots to generative AI photo editing, adopts Google’s SynthID watermarking
The tech giant unveils advanced image editing features for iOS, including spatial reframing and generative expansion, while embedding synthetic content identifiers to address deepfake concerns.
Apple has announced a significant shift in its approach to photographic integrity at WWDC 2026, introducing a suite of new AI-powered photo editing tools for iOS. The update includes an enhanced Clean Up tool, a generative 'Extend' feature for expanding image dimensions, and 'Spatial Reframing' for perspective adjustment. Additionally, the Image Playground app now supports photorealistic generation, a capability previously restricted to cartoon-like styles to mitigate deepfake risks. To address concerns regarding synthetic content and reality distortion, Apple will embed Google’s SynthID watermark in all images manipulated using these tools.
The announcement marks a departure from the company’s previous stance, articulated two years ago by then-software chief Craig Federighi, which emphasised the importance of “purveying accurate information, not fantasy.” At that time, Apple launched 'Clean Up', an AI object removal tool, while expressing hesitation to provide more extensive editing capabilities. Competitors Google and Samsung had already introduced extensive AI editing suites allowing users to add various elements to photographs via text prompts, including potentially harmful inclusions.
The new features allow for unprecedented manipulation of images Apple still refers to as “photos.” The updated Clean Up tool offers better quality and more realistic infill for removing distractions in complex scenes. The 'Extend' tool enables users to expand images beyond their original dimensions, such as turning a portrait into a landscape, using generative AI to fill in blank spaces. This functionality mirrors Adobe’s Generative Expand feature in Photoshop.
A key component of the update is 'Spatial Reframing', which uses spatial models similar to those in the Vision Pro headset. This allows users to drag parts of a photograph to change the perspective, generating new content only where the perspective has been adjusted. Apple states this ensures the reframed photo stays consistent with the original scene, although consistency does not equate to authenticity.
To manage the risks associated with these capabilities, Apple is adopting Google’s SynthID watermarking system. Any photos adjusted with Apple Intelligence, including those edited with Clean Up, Extend, and Spatial Reframing, will be embedded with SynthID. This near-invisible watermark helps identify synthetic content, aligning with industry moves by competitors such as OpenAI, which recently adopted the same system.
Users can check for these watermarks using Google’s Gemini or AI-powered Search, although online platforms are still in the early stages of automatically labelling content carrying SynthID data. Apple’s pivot suggests a belief that robust watermarking can prevent users from being misled, allowing for more expansive generative AI editing features while attempting to preserve trust in digital imagery.
