Google will help distinguish real images from AI fakes

Google intends to make publicly available a solution that will help determine whether an image was taken with a camera, edited using software like Photoshop, or created using a generative artificial intelligence model. In the coming months, Google’s search results will feature an updated “About This Image” feature to help users understand whether AI was used to create or edit an image.

Image source: c2pa.org

The technology Google uses is based on a metadata standard promoted by the Coalition for Content Provenance and Authenticity (C2PA), one of several organizations seeking to address the problem of AI-generated images. C2PA authentication includes information about the origin of images and is used by hardware and software systems to create a digital footprint. It is supported by Amazon, Microsoft, Adobe, Arm, OpenAI, Intel, Truepic and Google, but the deployment of the solution is slow. Integration into Google search results will be the first major test of this initiative.

Google participated in the development of the latest technical standard, C2PA 2.1, and the new search feature will be used in conjunction with the C2PA Trusted List. “For example, if data indicates that an image was taken with a specific model of camera, a trusted list can help confirm the accuracy of that information,” Google explained. The company also plans to integrate C2PA into its advertising systems and is considering introducing the standard to YouTube.

C2PA is already supported by Leica and Sony cameras, and Nikon and Canon have promised to add it to their products. It is expected that Apple and Google, the developers of the largest mobile operating systems, will take a similar step. C2PA metadata is added to Adobe Photoshop and Lightroom, but is not yet supported in Affinity Photo and Gimp. Most online platforms have not yet integrated this solution either, but perhaps they will follow the example of Google search engine.