Apple Delves into the World of AI-Driven Photo Editing Fantasy
Apple has shifted its stance on generative AI in photo editing, moving from skepticism to embracing tools that alter our perception of reality. At WWDC 2026, the company unveiled a suite of AI-driven photo editing features that allow users to easily manipulate images, which Apple still calls “photos.”
Two years back, Apple introduced Clean Up, an AI tool for removing unwanted objects in photos, akin to Google Photos’ Magic Eraser. At that time, Apple’s software chief emphasized the importance of providing “accurate information, not fantasy.” The company appeared cautious about rolling out more advanced AI editing capabilities, while competitors like Google and Samsung forged ahead with tools that let users add nearly anything to images through simple descriptions, including potentially harmful elements.
Now, Apple is stepping into the arena with its own image manipulation tools that utilize prompt descriptions. The revamped Image Playground app now features the ability to create photorealistic images. Apple claims this offers users powerful new ways to express their creativity.
Image Playground enables users to modify images by describing intricate changes in natural language or by interacting directly with specific objects to move or resize them. During a keynote demo, we saw the app generate an image of a woman holding a birthday cake, using a real photo as a reference. The result not only added the cake but also completely changed the background. Previously, Apple had focused on cartoonish styles that avoided realistic deepfakes. So, what prompted this change?
The answer appears to be SynthID, a watermarking system that tags content generated by AI. Apple states that any photos edited with its AI tools will carry SynthID, making it easier to identify AI-altered images. While Apple had already been labeling metadata for images edited with Clean Up or created through Image Playground, it now employs its own “forensics” feature, which is unique among major tech platforms.
SynthID watermarks will be applied to images edited with Clean Up, Extend, and Spatial Reframing—three AI-powered tools in Apple’s Photos app. The updated Clean Up tool has received a significant upgrade, allowing for the removal of distractions with improved quality and more realistic fills, even in complex scenes.
The new Extend tool lets users expand images beyond their original dimensions, using generative AI to fill in the gaps—similar to Adobe’s Generative Expand feature in Photoshop. This allows for transforming a portrait into a landscape, albeit with the understanding that the new background isn’t real.
Spatial Reframing enables users to adjust the perspective of images as if they were 3D scenes. You can select parts of a photo and drag them around to create the illusion of a different angle. Apple claims this feature builds on spatial models developed for the Vision Pro headset, generating new content only where the perspective has been altered, ensuring consistency with the original scene.
However, consistency doesn’t equate to authenticity. Any image edited with Apple’s tools will be marked with AI watermarks, raising the question of whether these images truly reflect reality. This topic has been widely debated, and Apple has previously expressed concerns about AI’s potential to distort how people perceive photographic content.
The AI labels aim to help users distinguish between genuine photographs and misleading AI alterations. Support for SynthID is growing across the industry, recently adopted by other tech entities. Users can check for SynthID data by uploading images to AI-powered tools and inquiring about the watermark. While this process isn’t particularly user-friendly, it does provide some means of verifying image authenticity. Online platforms are also working to automatically label content with SynthID data for quicker identification of AI-manipulated images.
These initiatives are still in their infancy, and much of the deepfake and synthetically generated content online remains unlabeled. Nonetheless, it’s significant that Apple is placing its faith in SynthID, despite its earlier concerns about AI’s capacity to distort reality. If SynthID proves effective, Apple may feel confident enough to expand its generative AI editing features.
Historically, Apple has emphasized the importance of photography as a reliable means of capturing real memories. However, this focus seems to be shifting. The company is now encouraging users to manipulate their personal photos in groundbreaking ways, raising the question: what is the ultimate goal? A more “perfect” image than reality? While Apple may not want to contribute to the flood of manipulated content online, it is banking on SynthID to mitigate potential issues. This marks a significant departure from its previous stance that photography should represent “a personal celebration of something that really, actually happened.”
