For the better part of a decade, Apple's camera marketing rested on a simple proposition: what you see is what you get. Deep Fusion merged multiple exposures. Photonic Engine applied machine learning. But the pixels in your photo came from photons that hit the sensor. The image was real. That was the point.
iOS 27 changes this. Apple's Photos app will use generative AI to add pixels that never existed, extending images beyond their original frame, recovering detail in blown-out areas and enhancing compositions after the fact.
Jon McCormack, Apple's camera engineering chief, defended the shift in a Wired profile, saying AI can give users "superpowers" and insisting Apple is not using the technology "for the sake of AI.
The word "superpowers" is doing a lot of work in that sentence. The power in question is the ability to create a photograph of something that did not happen.
Philosophical line Apple is crossing
Computational photography, as Apple practised it, was aggressive but honest. The phone took dozens of frames, selected the best data from each and composited a result that was optically impossible from a single exposure but derived entirely from captured light. The output was enhanced. It was not invented.
Generative AI operates differently. When the model extends an image beyond its original frame, the new pixels are not recovered from sensor data. They are synthesised by a neural network trained on millions of other images. The model predicts what should be there based on patterns, not on what was there.
The result may look convincing. It is not a photograph in any traditional sense. It is a photograph with AI-generated margins, and the user may not be able to tell where the real image ends and the synthetic one begins.
The (not) "for the sake of AI" defence
McCormack's insistence that Apple is not chasing AI trends for their own sake is a pre-emptive response to the obvious criticism: that the company is doing what every other phone maker has done, adding AI features because the market demands them, not because users need them.
Samsung and Google have both shipped generative photo tools. Samsung's object removal and Google's Magic Eraser and Best Take features create or delete content in images. Apple has watched from the sidelines for two years. iOS 27 is the capitulation.
The question Wired posed, "what even is a photograph these days?", is the right one. The answer is becoming: a photograph is a starting point that the software finishes. The user presses the shutter. The model decides what the image should look like.
Superpowers or brain rot?
McCormack's framing is optimistic. AI tools that recover shadow detail or correct lens distortion are genuinely useful. Tools that extend an image beyond what the camera captured are something else. They are tools that teach users to trust images that are partly fictional.
In a world already struggling with synthetic media, deepfakes and AI-generated misinformation, training two billion iPhone users to accept that their photos contain invented content is a choice with consequences that extend well beyond the camera app.
Apple is betting that users will embrace the convenience without questioning the authenticity. They are probably right. Whether that is a superpower or the beginning of something less benign depends on how much you think the line between real and synthetic still matters.
Apple's camera chief thinks it is a superpower. The rest of us get to decide whether we agree, one AI-extended sunset at a time.
