Google Opens Synthid Detector To Everyone
Google’s free website checks images, video and audio for SynthID watermarks, but it cannot identify every piece of AI-generated media.
Google has opened SynthID Detector, a website for checking whether images, videos and audio clips carry a watermark associated with AI-generated or AI-edited media. The tool is free, though Engadget reports that users must sign in with a Google, OpenAI or Apple account. Its wider release follows a period of testing with journalists and other media professionals.
Table of Contents
What the website checks
Users can submit media to the site for a SynthID check. According to TechCrunch, Google introduced the watermarking technology in 2023 and uses it across its Nano Banana, Veo and Lyria models, as well as generation tools including Gemini, Flow, ProducerAI and Vids. OpenAI, NVIDIA and Kakao also support SynthID, the reports say.
TechCrunch reports that the website accepts common image formats including JPG, PNG, WEBP, HEIC and GIF; video formats MP4, MOV and WEBM; and audio formats including WAV, MP3, FLAC and M4A. The outlet says Google first offered the detector to a limited group of journalists, media professionals and researchers for testing at Google I/O last year.
Apple support and other ways to verify
Support for images created or edited with Apple Intelligence is expected soon, according to Engadget. That includes images made with Apple’s Image Playground. The planned support does not mean the website can identify those images yet.
The dedicated site gives people another way to check media without starting in one of Google’s existing products. Engadget says Google also offers verification through Circle to Search and the Gemini app, while TechCrunch reports that verification is built into Gemini and Chrome. According to TechCrunch, Google says people currently make 1 million content-verification requests a day.
Why a result is not definitive
A check that finds no SynthID watermark does not establish that a file was made without AI. Engadget notes that companies use different watermarking standards and that watermarks can sometimes be removed. It also says the website does not distinguish between media generated entirely by AI and media that was merely edited with AI, or identify which parts are synthetic.
Other companies have separate approaches. TechCrunch says Microsoft and Meta use their own watermarking and verification standards, while Engadget describes a Meta tool that can identify images from its latest Muse Image models but not images from other companies or Meta’s earlier models. Engadget also reports that ElevenLabs has committed to adding SynthID support. For now, the new website makes one watermark easier to check, rather than providing a universal test for AI-generated content.
Sources
This story was compiled by AI from the reports below. Read the originals for the full details.