Google has unveiled a new tool called SynthID in an effort to combat the spread of misinformation through computer-generated images. This tool utilizes an invisible watermark that is embedded directly into images created by Google’s text-to-image generator, Imagen.
The purpose of SynthID is to identify and flag computer-generated images, making it easier for users to determine whether an image is authentic or artificially created. The watermark remains on the image even if it undergoes modifications, providing a reliable tool for detecting manipulated content.
One of the key features of SynthID is its ability to scan incoming images and determine whether they were generated by Imagen. It categorizes images into three levels of certainty: detected, not detected, and possibly detected. While Google acknowledges that the technology is not perfect, it claims that it is accurate against many common image manipulations.
This move by Google comes as the issue of deepfake and edited images continues to grow. These manipulated images have become increasingly realistic, making it difficult for users to discern between what is real and what is not. As a result, tech companies are actively seeking solutions to identify and flag manipulated content.
Google’s announcement of SynthID places them among a group of startups and tech companies working on similar solutions. The Coalition for Content Provenance and Authenticity (C2PA), backed by Adobe, has been leading efforts in digital watermarking. However, Google has chosen to develop its own approach.
In addition to SynthID, Google launched a tool called About this image in May. This tool allows users to find information about when images were indexed, where they were first seen, and where else they can be found online. While these technical solutions are a step in the right direction, it remains uncertain whether they can fully address the problem, especially as AI technology accelerates at a pace faster than human oversight.
SynthID is currently available in beta to select customers of Vertex AI, Google’s managed machine learning platform. It will be interesting to see how effective this tool becomes in the fight against manipulated content and misinformation, as well as how other companies contribute to the ongoing efforts to tackle this issue.
“Zombie enthusiast. Subtly charming travel practitioner. Webaholic. Internet expert.”