Google Photos to Automatically Tag AI-Edited Images with Undetectable Watermark

Google Photos is adding a new tool that will let users recognize photographs that have been modified with artificial intelligence (AI). This update will apply to photos that have been updated using Magic Editor's Reimagine tool, which uses Google's Imagen 3 AI model to add new aspects to images. The tech giant has used SynthID, an advanced watermarking technology, to recognize such AI-enhanced photos.

SynthID, created by Google DeepMind, embeds an invisible and imperceptible tag within the image's pixels, making it difficult to remove by cropping, adding filters, or duplicating the image. This watermark can be viewed in Google Photos' "About this image" section or detected with specialized programs. Unlike traditional watermarks, SynthID is incorporated directly into the structure of media, ensuring authenticity across a wide range of formats such as photos, videos, text, and audio.

This move intends to combat misinformation by decreasing the dissemination of AI-generated deepfakes, which are distorted media that mislead audiences. While most AI-edited photos will have this watermark, tiny adjustments, such as slight color changes to background objects, may not be tagged with SynthID. Users can check the metadata of an image to see if the watermark was applied.

By deploying SynthID, Google demonstrates its commitment to openness in AI-generated material, allowing consumers to discern between original and AI-enhanced photos while reducing the danger of digital deception.

Information sourced from Gadgets 360