Google Photos has recently announced that it will add a digital SynthID watermark to photos edited using the generative AI feature, Magic Editor. According to Google, this new feature will be rolled out gradually within the week, aiming to help users quickly identify which images have been modified using Magic Editor's "Reimagine" tool.

SynthID is a watermark system developed by Google's DeepMind team that embeds digital metadata tags directly into images, videos, audio, and text to determine whether these contents were created or altered using AI tools. Currently, this watermark has already been applied to images fully generated by Google's Imagen text-to-image model. Other companies have also developed similar watermark systems, such as Adobe's Content Credentials for works created or edited in its Creative Cloud suite.

Previously, it was reported that users could easily edit photos with realistic details on their Google Pixel 9 phones using the "Reimagine" function by simply describing the content they wanted to add. Although AI editing tools are not inherently malicious, Magic Editor can generate images like crashed helicopters, drug paraphernalia, or even corpses, which are difficult to distinguish as AI-generated at first glance. Since at least October of last year, Google has been adding labels in the file descriptions of AI-edited images in Google Photos.

This watermark update aims to address these issues but still has certain limitations. First, SynthID does not visibly alter the image it is attached to; it can only be identified through Google's dedicated AI detection tool in the "About This Image" feature. Additionally, Google notes that some edits made with Magic Editor's "Reimagine" function "may be too subtle for SynthID to mark and detect."

While watermark technology can play a role when detecting modifications, some experts in the field believe that relying solely on watermarks is insufficient for effectively verifying the authenticity of AI-generated content on a large scale, requiring a combination of diverse methods working together.