The Pulse
Google Opens SynthID Detector to Everyone Worldwide
Google opened its SynthID Detector to anyone worldwide on October 7, 2026, with the tool available in English. It checks images, video and audio for SynthID watermarks from Google and named partners, including OpenAI, NVIDIA and Kakao.
AI.info Team · Written with AI ·
Google makes SynthID checks public
Google opened its SynthID Detector to everyone worldwide on October 7, 2026, replacing a limited-access phase aimed at journalists and other media professionals. The web tool is available in English and checks uploaded images, video and audio for watermarks associated with Google AI and participating partners. Google DeepMind vice president Pushmeet Kohli announced the change in a company blog post.
“Agora, estamos a expandir, a partir de hoje, o acesso a todos, com a ferramenta disponível globalmente em inglês,” Kohli writes. The announcement names OpenAI, NVIDIA and Kakao among the partners whose AI-generated media the detector can check, and says Apple is expected to join soon. Google does not give a date for Apple’s inclusion.
Watermarks, not a universal AI verdict
SynthID embeds signals that people cannot ordinarily see or hear into images, video and audio. The detector looks for those signals in submitted media; it does not establish that every item made with AI will be identified. Its reach depends on a SynthID watermark being present, so a scan that finds none cannot, by itself, show that a file was made by a person.
Google first introduced the detector in May 2025 as a verification portal for content made with Google AI. At that launch, the company said the portal could identify likely watermarked portions of a file, rather than only returning an overall result. The public release now extends access to anyone, while the new partner checks broaden the sources the portal says it can cover.
Google reports 180 billion watermarked images and videos
Google says it has added SynthID watermarks to more than 180 billion images and videos, as well as audio totaling 240,000 years. Those figures describe the company’s watermarking activity, not the number of files the public detector has scanned. Google’s announcement does not provide detection-accuracy rates or separate totals for each partner.
The detector also joins verification features already built into Google Search, Gemini and Chrome. The company says those features now process more than one million requests a day. It did not specify how many of those requests relate to SynthID or how many involve the newly opened portal.
The result depends on who adopted SynthID
Opening the portal globally makes Google’s watermark check available beyond its original professional testing group, but the named partner list also defines its limits. The tool checks for SynthID signals; it is not a general-purpose judge of whether any image, clip or recording came from AI. For media without a supported watermark, the announcement offers no broader detection guarantee.
That distinction matters as Google expands a system first launched for its own AI output into a partner-supported check. Users can now submit media through the SynthID Detector, but the result answers a specific question: whether the tool finds a supported SynthID watermark in the submitted file.