Google's Watermark Technology 'SynthID' for Identifying AI-Generated Photos | Release #441
Cover photo by 侑奈
AI technology is revolutionizing the world of photo editing. However, as AI editing advances, the line between 'authentic' and 'altered' becomes blurred. Google has introduced the technology of embedding invisible watermarks, 'SynthID', into AI-generated and edited content to address this issue.
This article delves into the workings of SynthID and its implications.
New 'Watermark' Technology for AI-Edited Images
SynthID is a digital watermark technology developed by Google's AI research arm, DeepMind. It is embedded at the pixel level in AI-generated or edited images and videos, allowing detection with specialized tools without affecting the visual appearance.
Photo by meg
Unlike traditional visible watermarks, SynthID is designed to remain intact even after cropping, color adjustments, and compression. This helps prevent the spread of misinformation on social media and news sites, making it easier for viewers to verify the source of images.
Enhancing Transparency in AI-Edited Photos
SynthID is an important step towards enhancing transparency, but it is not a complete solution. For instance, minor edits by Google's 'Magic Editor' may not be detectable by SynthID. Additionally, since it is a proprietary technology by Google, if other companies do not adopt similar watermarks, only some images will be identifiable.
Photo by Yuya
Moreover, there is a possibility that malicious editors might develop ways to circumvent SynthID. A unified standard across the industry is necessary.
Balancing AI Editing and Originality
While AI editing technology brings new creative possibilities, it also poses risks to the reliability of photos. When taking photos, it is important to value the original beauty while using AI as a supplementary tool.
The widespread adoption of technologies like SynthID will help maintain transparency in editing, allowing more people to enjoy photos with confidence.
