Google introduces SynthID Detector to spot AI-generated content: What is it

Google's SynthID Detector aims to improve transparency around AI-generated content by helping users trace its origin-supporting the responsible use of AI

Google has launched a verification portal called SynthID Detector to help identify content created using Google AI. This tool is designed to detect whether text, images, audio, or video were generated using AI models. The tool will work by scanning content for hidden watermarks added by SynthID, Google’s watermarking technology.

SynthID Detector: Details

SynthID Detector will highlight which parts of the content are more likely to have been watermarked with SynthID. These watermarks are invisible to the human eye and do not affect the quality of the content. This new tool is built to support different types of content. It will work with all AI-generated media made by Google developed tools like Gemini, Imagen, Lyria, and Veo.

 

Google said that over 10 billion pieces of content have already been watermarked using SynthID. With the launch of SynthID, it aims to provide a suite of novel technical solutions to help minimise misinformation and misattribution.

 

SynthID Detector will be available through a web portal where users can upload their files. Once a file is uploaded, the tool will check for a SynthID watermark. If it finds one, it shows which parts of the content are most likely marked.

 

For example, in audio files, it will highlight certain parts of the track where the watermark is found. In images, it will show specific areas that are likely to contain the watermark. It will work the same way for video and text.

The SynthID Detector is rolled out for early testers, it will be available more widely in coming weeks.

How SynthID Detector works

Upload content: Upload an image, audio track, video or piece of text created using Google's AI tools.

Scan for watermarks: The portal scans the uploaded media and detects whether the uploaded content, or specific portions of it, contain a SynthID watermark.

View results: The portal will show the results and if a SynthID watermark is detected, it will highlight which parts of the content are more likely to have been watermarked.

Expanding the SynthID ecosystem

To help grow this ecosystem, Google has already open sourced SynthID text watermarking, so any developer can build with this technology and incorporate it into their own models.

Google said “Content transparency remains a complex challenge. To continue to inform and empower people engaging with AI-generated content, we believe it’s vital to continue collaborating with the AI community and broaden access to transparency tools.”