Apple Is Building a Tool to Prove Your iPhone Photo Is Real, Not AI-Generated

Key points

• Apple's iOS 27 beta 5 contains code for a feature called "Apple Reference Image" that embeds proof-of-origin data into iPhone photos at the moment of capture.
• The feature is off by default and requires users to actively opt in through the Camera app settings.
• Verification sends image data to Apple's cloud servers but Apple says it does not access the raw photo itself.
• Canon, Nikon, Sony and Google already support a similar industry standard called C2PA, which Apple has so far avoided.
• Authenticated photos can be viewed on iPhone, iPad or Mac.

Your phone takes a photo. But how does anyone know it is real? That question has become much harder to answer since AI tools started generating convincing fake images in seconds. Apple appears to be working on an answer.

Code spotted inside iOS 27 beta 5, the fifth test version of the iPhone software expected later this year, points to a system called Apple Reference Image. First reported by The Verge AI, the feature would embed a kind of digital fingerprint into a photo the moment you press the shutter, recording proof that a real iPhone camera captured it at a specific time.

How does it actually work?

The feature is not live yet, and Apple has said it will be switched off by default when it does arrive. To use it, you would go to Settings, then Camera, then Reference Image, turn on Reference Mode, and take photos using a new "Reference" option inside the Camera app. Only photos taken that way get the provenance stamp.

Getting that stamp verified is a separate step. You tap a "Reference" badge shown on the photo, and your iPhone sends the image along with its embedded data to Apple's Private Cloud Compute servers, the company's system for processing sensitive data without Apple staff seeing it. Those servers check the sensor signatures (a unique electronic fingerprint from your camera hardware), the capture time, and hardware identifiers tied to your specific device.

Apple returns a verified version of the photo with a unique ID attached. The company says it never sees the raw photo itself, though it does receive the sensor data. That matters because Apple could use that data to block authentication on photos from a device whose sensor has been tampered with, or pull back verification it already granted.

Why does this matter to ordinary people?

Fake images are genuinely hard to spot. A deepfake, meaning a fake photo or video created by AI that looks real, can now be made with a free app in under a minute. Apple Reference Image would give journalists, courts, employers or anyone else a way to check whether a photo you took is authentic, without trusting you to say so.

Instagram's head, Adam Mosseri, has said publicly that labelling real human-made content may actually be easier than hunting down every AI fake. Apple building this tool suggests the company agrees with that logic.

The industry already has a shared standard for this called C2PA, short for Coalition for Content Provenance and Authenticity, a system that records how and where an image was made and whether AI tools touched it. Camera makers including Canon, Nikon and Sony have added C2PA support to their hardware. Google's Pixel 10 phone supports it too. Apple has not adopted C2PA, and reliability concerns around it may explain why the company is building its own solution instead.

What should you do right now?

Nothing yet. The feature is not available to ordinary iPhone users. When it does arrive, turning it on would be a simple way to make your photos carry verifiable proof of authenticity, useful if you are a photographer, a journalist, or anyone who regularly needs to show that an image is genuine.

The honest takeaway: watch for this in the full iOS 27 release. If provenance matters to your work, it will be worth switching on.