AI in photography is no longer a feature hidden in a settings menu. It decides how your phone handles a dark restaurant, removes the stranger from a holiday photo, and can build a whole scene from a typed sentence. The camera still matters, and so does the person holding it, but much of what used to take skill, time or a tripod now happens in software.
The camera now does some of the seeing
A phone camera has a small lens and sensor, so it collects far less light than a big camera. In dim light the result is noise: random speckles of brightness and colour across the picture. Google's answer, described in its own research blog, was to take many pictures and combine them. In 2014 it introduced HDR+, which captures a burst of frames, lines them up and merges them. Merging also cuts noise, because the speckles differ from frame to frame while the scene stays the same.
Night Sight took the same idea further. Google introduced it with the Pixel 3 in November 2018, for light between 3 lux and 0.3 lux — below the range where a merged burst copes on its own. A single phone photo starts to struggle at 30 lux. Night Sight was built for that darker band, with no flash and no tripod.
The trick is to keep collecting light after you press the shutter. Night Sight captures a burst: up to 15 frames when the phone is held by hand, each as long as 333 milliseconds, or six frames of one second each on a tripod. It measures movement and shortens each frame when things move, so faces stay sharp. The frames are then aligned and merged, and a learning-based method fixes the white balance, which decides whether a photo looks yellow under street lamps or the way your eyes remember it.
The result is a photo of a scene your eyes could barely see, in colour. The idea has since become a normal part of phone cameras.
Editing that took an evening now takes seconds
The second wave arrived after the shot was taken. Google's Magic Eraser first appeared on the Pixel 6, and in February 2023 Google opened it to Google One members on Android and iOS. It spots a distraction, removes it and fills in the background with what the software guesses should be there.
The Pixel 8 and Pixel 8 Pro, launched in October 2023, went further, with four new features in Google Photos: Best Take, Magic Editor, Audio Magic Eraser and Zoom Enhance. Best Take blends a series of similar group photos into one image, so you can pick the version where everyone has their eyes open. Magic Editor lets you tap or circle an object, drag it elsewhere or pinch it bigger, and it can change the sky or the lighting. Google said the tool was in its early stages and that there would be times when the result was not exactly what you imagined.
Apple followed a year later. Clean Up arrived with iOS 18.1, on iPhone 15 Pro and iPhone 16 models. You tap, brush or circle what you want gone, and the tool fills in the space. Apple notes that some items are highlighted for you automatically, and that brushing over a person's face may blur it with a pixelated effect. It works on any photo in your library.
In May 2024 Google opened Magic Editor to all Google Photos users on Android and iOS, with 10 free saves a month. Tools that once needed a desktop app and training now sit in the app where most people already keep their photos. The edit that used to take an evening now takes seconds.
Whole scenes can be built from a sentence
The most far-reaching change is not removing things but making them. Adobe launched Firefly, its family of image-making models, in March 2023, and on 23 May 2023 it put Firefly inside Photoshop as Generative Fill. You select part of a photo, type what you want — a tree, a dog, a wider sky — and the tool paints it in, matching the light, perspective and style of the rest of the image. It can also stretch a photo beyond its edges, filling in the parts the camera never saw.
Adobe said Firefly's first model was trained on Adobe Stock images, openly licensed content and other public domain content, and was designed for commercial use. By the time Generative Fill was unveiled, beta users had made over 100 million assets with Firefly. It reached general availability in the second half of 2023.
The same release carried a second idea that matters as much as the editing. Generative Fill supports Content Credentials, which Adobe describes as being like nutrition labels for digital content. The label stays attached to the file and records whether it was made by a person, made by AI, or edited by AI, so a viewer can check the history of an image instead of guessing from the pixels. The system was developed by the Content Authenticity Initiative, which Adobe founded.
The standard behind those labels is open. The Coalition for Content Provenance and Authenticity, or C2PA, records the origin and the edits of digital content in an open technical standard. The label is a signed record inside the file, so changing the file after the label was written breaks the signature. It is not a perfect answer — a screenshot or a re-saved copy can strip the label away.
The contest that showed the trust problem
The question of what counts as a photograph came to a head in 2023. Boris Eldagsen, a German artist and photographer, entered an image called "The Electrician" in the Sony World Photography Awards. Made in 2022 with the text-to-image tool DALL-E 2, using prompts plus inpainting and outpainting, it won the Creative category of the Open competition, with the winners announced on 14 March 2023.
Eldagsen then refused the award. He said he had entered to test whether photography competitions could tell AI images from photographs. "AI images and photography should not compete with each other in an award like this. They are different entities. AI is not photography. Therefore I will not accept the award," he wrote. The organisers removed him from the competition and took the image down from the exhibition in London.
The episode showed that the problem is not only technical. Judges, viewers and buyers now have to decide what they are looking at, and the tools have moved faster than the rules for judging them. Competitions have had to write new rules about what AI is allowed, and newsrooms have had to decide when an AI-edited photo can still be published as a record of something that happened. The label on the file is often the only thing that tells you.
Where AI in photography falls short
The tools are impressive, but they fail in ways worth knowing before you trust them. The first failure is accuracy. A merged night photo can smooth away fine detail, and a generative edit can invent the wrong thing: extra fingers, warped text on a sign, a window that does not line up with the wall around it. Google itself warned that Magic Editor would sometimes produce a result that was not what the person imagined.
The second failure is access. The newest tools need the newest phones or a paid plan. Apple's Clean Up needs an iPhone 15 Pro or an iPhone 16, or a recent iPad or Mac. Magic Editor is free for 10 saves a month, and after that it needs a Premium Google One plan or a Pixel phone. Adobe's Generative Fill sits inside Photoshop, which is a subscription.
The third failure is honesty. Content Credentials only work if they stay attached, and a screenshot or a re-saved copy can strip them away. An image with no label might be a straight photograph, or one that was heavily edited and then cleaned of its history.
There is also the question of who benefits. Firefly was trained on licensed stock photos, so the photographers whose work taught it were paid through the stock library. Adobe states what Firefly was trained on; a photographer choosing a tool can check the same for any other tool.
Frequently asked questions
Can AI tell if a photo is real?
Not from the pixels alone. The reliable method is a Content Credential, the signed label that records how a file was made and edited. If the label is present and valid, it tells you whether AI was involved. If the label is missing, no tool can prove a photo is real.
Does AI editing make a photo fake?
It depends on the edit. Removing a stranger from the background or brightening a dark room leaves the moment intact. Replacing a grey sky or adding something that was never there changes what the photo shows. The honest rule is simple: say what you changed, and keep the original.
Do I need a new camera to use AI photo tools?
No. Most of the editing tools live in phone apps, not in cameras. Magic Editor opened to all Google Photos users in May 2024, and Apple's Clean Up works on any photo in your library, including older ones. A newer phone adds the newest features, but the basic tools work on photos you already have.
AI in photography has changed both ends of the process. The camera now takes many frames and computes a clean picture from them, and the editor can reshape that picture with a tap or a sentence. The craft has not disappeared: choosing what to photograph, when to press the shutter, and what a picture should honestly show still belongs to the person behind the camera.
The useful habit is to ask two questions about every striking photo you see: what was actually there, and what did the software add? If you make photos for your business or your community, those two questions protect your work and your readers. If you need clear, honest content that readers can trust, see our services, or get in touch through our contact page.
What to watch next
Watch whether Content Credentials spread from Adobe's tools onto every phone camera, so the label is written when a photo is taken. Watch whether competitions and newsrooms settle on shared rules for AI edits, after the confusion of the 2023 Sony award. And watch the price: as the newest features arrive on cheaper phones, the gap between photographers should narrow.
Sources
- Google AI Blog — Night Sight: Seeing in the Dark on Pixel Phones — date checked: 2026-11-07 — https://ai.googleblog.com/2018/11/night-sight-seeing-in-dark-on-pixel.html
- Google — 4 new Google Photos features on Pixel 8 and Pixel 8 Pro — date checked: 2026-11-07 — https://blog.google/products-and-platforms/products/photos/google-photos-features-pixel-8-pro/
- XDA Developers — Google is bringing Magic Eraser to all Pixel owners and Google One members — date checked: 2026-11-07 — https://www.xda-developers.com/google-one-magic-eraser-announced/
- Droid Life — Best Google Photos Feature Coming to All Users, Even iPhone — date checked: 2026-11-07 — https://www.droid-life.com/2024/04/10/google-opening-magic-editor-to-all-google-photos-users-including-iphone/
- Adobe — Adobe Unveils Future of Creative Cloud With Generative AI as a Creative Co-Pilot in Photoshop — date checked: 2026-11-07 — https://news.adobe.com/assets/downloads/pdfs/2023/5/052323photoshopfirefly.pdf
- Apple Support — Use Apple Intelligence in Photos on iPhone — date checked: 2026-11-07 — https://support.apple.com/guide/iphone/use-apple-intelligence-in-photos-iphf7de217f0/18.0/ios/18.0
- Hyperallergic — German Photographer Refuses Award for His AI "Photo" — date checked: 2026-11-07 — https://hyperallergic.com/german-photographer-refuses-award-for-his-ai-photo/
- Uncrop — What Is C2PA? Content Credentials and How to Verify Them — date checked: 2026-11-07 — https://uncrop.si/guides/content-credentials-c2pa/
Eri-Tech Global Ltd
Eri-Tech Global Ltd is a digital solutions company working with small businesses and growing brands. Learn more at /about.
