A new feature in Apple’s iOS system could allow users to confirm the authenticity of photographs taken with the iPhone. The company is working on a tool that aims to combat the proliferation of images manipulated by artificial intelligence, popularly known as “deepfakes”, through a validation system at the time of capture. The information indicates that the functionality is present in the beta 5 version of iOS 27.
The system, called “Apple Reference Image”, is not yet publicly available. A privacy note in the iOS beta states that the feature will be turned off by default. When launched, users will be able to activate it through the device’s settings, navigating to Settings > Camera > Reference Image > Reference Mode. Only photos captured using the new “Reference” option in the iPhone Camera app will include the provenance data required for authentication.
On the same topic: Classmate creates sexual deepfakes of teenagers at school with AI and shocks victims
The verification process does not happen automatically, requiring the user to tap a specific badge in the photograph. After this action, the original image and embedded provenance data that includes sensor signatures, capture period, and unique hardware identifiers are sent to Apple’s Private Cloud Compute servers for analysis. An authenticated version of the photo is then returned, accompanied by a unique ID.
During the authentication process, Apple does not access the original photo itself, but may receive sensor data. This measure would allow the company to prevent the validation of images associated with compromised sensors or even revoke previous authentications of photos linked to these sensors.