#synthiddetector

Google Launches SynthID Detector Portal to Combat AI Misinformation

New tool SynthID Detector helps identify AI-generated content.

SynthID Detector can detect SynthID watermarks in Google AI-generated images.

SynthID Detector, a new Google online service, will help identify AI-generated material. Generative AI allows people to create new text, sounds, images, and videos, raising context, authenticity, and verification concerns.

SynthID Detector verifies Google AI-generated content quickly and effectively. It promotes transparency in the fast-changing world of generative media by teaching detection across media types. The portal can highlight which content parts are most likely to include SynthID watermarks.

This application builds on Google's groundbreaking SynthID. SynthID initially offered a variety of technology solutions to identify AI-generated content and eliminate disinformation and misattribution by integrating invisible watermarks.

The SynthID watermark can be identified after content has been copied or altered due to its durability. After focussing on AI-generated graphics, SynthID now generates text, audio, and video. Veo, Lyria, Imagen, and Gemini are Google models. Google claims SynthID has watermarked over 10 billion documents.

SynthID Detector works how?

The SynthID Detector webpage is easy to use:

Upload content: Google's AI creates images, audio tracks, videos, and texts.

Check for watermarks: Next, the portal checks uploaded media for SynthID watermarks.

View outcome: The site shows scan results. If a SynthID watermark is identified, it will flag likely watermarked content. The site finds watermarked audio portions and suggests photo watermarks.

Text and video watermark detection will be added in the following weeks. The product is currently in development, thus specs may change.

Environment Rollout and Growth

Early testers can now use the SynthID Detector from Google. Future expansion is planned by the company. Researchers, media professionals, and journalists can join the portal waitlist.

Google is working hard to build a solid environment around SynthID and the gateway. Any developer can integrate their open-sourced SynthID text watermarking technology into their models.

Google is also forming commercial partnerships. They worked with NVIDIA to watermark Cosmos preview NIM microservice movies on build.nvidia.com this year. This partnership extends SynthID embedding beyond Google-generated content.

Google also partnered with GetReal Security, a leading content verification tool, to let third parties discover SynthID watermarks, recognising their importance. Google offers a form for interested companies to enquire about SynthID collaboration.

Google acknowledges that content transparency remains tough. To educate and empower users of AI-generated content, the company believes transparency tools and AI community collaboration are crucial.

Google launched the SynthID Detector along with new generative media models and tools, Veo-powered AI filmmaking, Google AI Ultra membership, AI Mode for shopping, AI in Search updates, and more AI insights from its CEO.