Apple’s Reference Image Feature Preserves Your Original Photos (Even After Editing).
Distinguishing the authenticity of online content is becoming increasingly difficult, and AI-powered watermarks are only partially effective in identifying counterfeits. But today, Apple unveiled a new standard that will at least help identify genuine images taken directly on an iPhone: Apple Reference Image. It won’t solve the problem of finding the truth online, but it could be a useful tool in that direction.
The new feature will be available on the first iPhone 18 Pro models. It works by using data from the camera sensor to create a reference image that cannot be edited, as well as a version that can be modified. The idea is to provide you with a version of the image that was verified at the time of capture, which you can compare with any modified versions later.
This metadata, confirming the image’s origin, can be stored via some supported sharing methods, which currently means that some other Apple devices can check the verification status. This doesn’t mean anyone can verify the image’s authenticity—especially if the published and distributed version is an edited version—but it does provide a tool for the person or organization that originally took the photo to provide proof of authenticity.
This tool came alongside the announcement that Apple will soon add support for the SynthID standard. This will potentially help users determine whether an image was created or modified using AI. Apple joins a long (though not entirely exhaustive) list of major tech companies supporting this standard, including OpenAI, Nvidia, and, of course, Google, the original creator of SynthID.
We still find ourselves in a fragmented image verification ecosystem. Apple’s “Reference Image” feature is a clever implementation that provides end users with a signed original for comparison, which is an improvement over C2PA metadata, which simply contains information about the image’s origin. It remains to be seen how widely these verification tools will be adopted and how easy it will be for users to trace an image’s origin and how exactly it was altered, but it’s certainly better than nothing.