Apple Is Working on a Way to Prove a Photo Was Really Taken on an iPhone

Apple is building a feature for iPhones that can verify a photo was actually taken by the phone's camera and not created or altered by artificial intelligence. The feature, called Apple Reference Image, was discovered in a test version of iOS 27 by technology news sites 9to5Mac and MacRumors on August 10, 2026, with The Verge confirming the findings on August 11, 2026 (9to5Mac, The Verge).

The feature is not yet available. According to a privacy notice in the test software, it will be turned off by default. When released, users will be able to switch it on through Settings > Camera > Reference Image > Reference Mode. Only photos taken using a new "Reference" option in the iPhone Camera app will include the information needed for verification (The Verge).

Checking a photo is not automatic. A person viewing the photo must tap a Reference badge on the image to start verification. When tapped, the photo and its embedded data — including details about the camera sensor, the time the photo was taken, and unique hardware identifiers — are sent to Apple's servers for checking (MacRumors). After checking, Apple returns an authenticated version of the photo with a unique ID. Authenticated photos can be viewed on an iPhone, iPad, or Mac (The Verge).

Apple says it does not view the actual photo during the verification process. However, the company may receive sensor data that lets it block images from cameras that have been tampered with, or to cancel authentication on photos previously linked to those cameras (MacRumors).

Apple Reference Image is similar to an existing open standard called C2PA, or Content Credentials, which can trace how and where an image was made and whether AI tools were used. Camera makers including Canon, Nikon, Sony, FujiFilm, and Leica have been adding C2PA to their cameras. Google's Pixel 10 phone also supports it. Apple has not adopted this standard directly, and Reference Image appears to be its own separate approach to the same problem (The Verge).

MacObserver reported that Apple Reference Image could verify photos were genuinely taken on an iPhone using hardware-backed sensor data, helping people trust photos at a time when AI-generated images are becoming more common (MacObserver).

The broader context here is the growing difficulty of trusting that a photograph is real. AI tools can now create highly realistic images, and the systems meant to verify photo authenticity are still fragmented and not widely understood. Apple's decision to build its own system instead of using the existing C2PA standard raises a practical question: if Apple's verification does not work with the C2PA system already built into cameras from Sony, Canon, Nikon, FujiFilm, and Leica as well as the Pixel 10, then a photo verified on an iPhone may not be recognized as authentic outside of Apple's own ecosystem.

The way Apple has designed the system is also worth noting. By linking verification to specific camera hardware and routing the checking process through Apple's own servers rather than handling it entirely on the phone, Apple is making itself the only authority that can confirm whether an iPhone photo is authentic. The ability to cancel authentication after the fact, if a camera sensor turns out to have been compromised, suggests Apple keeps a record of sensor states over time. This gives Apple a way to fight tampering, but it also means trust is concentrated in Apple's hands rather than spread across an open standard that anyone can use.

The fact that the feature is optional fits Apple's approach to privacy, but it also limits how useful the system becomes. A verification system works best when lots of photos carry the data that makes checking possible. If users have to deliberately switch on a separate camera mode, the portion of iPhone photos carrying that data may stay small, especially at first. Photos taken in the standard Camera app will not carry the data at all.

There is also the question of what happens when an authenticated photo leaves Apple's world. Apple says authenticated photos can be viewed on Apple devices, but the available information does not say whether the authentication badge, the assigned ID, or the underlying data survives when a photo is shared to other platforms, compressed by social media apps, or captured as a screenshot. These are exactly the situations where verification data tends to get stripped away today, and they are the specific problem that the open C2PA standard was designed to solve.

For anyone following this closely, the key facts are straightforward: the feature exists in a test version of iOS 27, it is optional and off by default, it requires a separate camera mode, verification requires a tap on a badge, and the checking happens on Apple's servers. The C2PA standard is the right point of comparison, but whether Apple's system will eventually work alongside it or remain separate is something we do not yet know. The details that would answer that — the format of the data, how it behaves when exported, and whether outside developers will get access — have not been disclosed.