Can you prove a photo is real?
31-10-2023
Translation: Veronica Bielawski
A new process intends to prove the authenticity of photos with the help of encrypted metadata. This isn’t the first attempt of its kind. But it’s the most promising one so far.
The digital age has a problem with fakes. Fake images are getting more and more convincing – and easier to produce. AI-supported image generators such as Midjourney or Dall-E are just the tip of the iceberg. There’s an urgent need for a method that can prove the authenticity of photos.
Content Credentials aims to do just that. When you take a picture, the camera saves metadata such as the camera model, time and location. However, this metadata is typically unencrypted and can be easily changed. Content Credentials use an open source standard to store this metadata with encryption.
The Leica M11-P is the first camera to implement Content Credentials, with other brands being likely to follow. In fact, Nikon presented its own prototype based on the Z 9 at the Adobe Max 2023 trade fair. And Canon is a member of the Coalition for Content Provenance and Authenticity (C2PA), the entity behind Content Credentials. The same goes for Sony, although Sony has also developed its own method for the encrypted storage of metadata (linked press release in German). The feature is available for the Sony Alpha 7 IV, but must be purchased separately and is only intended for business customers.
Making edits known
Content Credentials also allow you to save post-processing work steps with encryption. Of course, your software must support this. Adobe Firefly, for example, does. A corresponding feature in Photoshop is currently in beta. This would make it known if a given photo was edited in any way. You can call up the image information right in your browser by clicking an info button.
In addition, there’s a website where you can upload and check an image. The report not only documents the individual editing steps, but also earlier versions of the photo. In the case of the Leica M11-P, the original image as snapped on the camera would also be visible. With that, it’s not possible to edit an image with Content Credentials and have it pass as the original. Of course, each image can be edited and saved again – but then it no longer has any credentials.
Content Credentials aren’t intended to automatically detect or prevent fakes. What they can do, however, is prove the authenticity of a controversial photo – for example, from a warzone – and show who the author is, as this information can also be encrypted and saved. Like this, the photographer can prove they took a given picture.
Much support
Content Credentials aren’t the first attempt at a method of proving authenticity. Camera manufacturers had already developed methods for encrypting metadata over ten years ago. However, both Canon’s image verification system and Nikon’s algorithm have been cracked.
The current attempt seems more promising; it’s an open standard that’s widely supported. Among those involved are Adobe and Microsoft as well as chip manufacturers Intel and Arm. Picture agencies and media companies are also on board.
David Lee
Senior Editor
David.Lee@digitecgalaxus.chMy interest in IT and writing landed me in tech journalism early on (2000). I want to know how we can use technology without being used. Outside of the office, I’m a keen musician who makes up for lacking talent with excessive enthusiasm.