Google opens its SynthID detector to everyone

Google DeepMind’s Pushmeet Kohli announced on 7 October 2026 that SynthID Detector is open to everyone, in We’re making it easier to identify AI-generated content globally. Search Engine Journal’s Matt G. Southern covered it in Google Opens Its SynthID Watermark Detector To The Public. The detector’s own site, synthid.com, renders its text with scripts we could not read, so the wording we quote from it is as reported by Search Engine Journal.

What does SynthID Detector check?#

Per Google’s post, anyone can now check whether an image, video or audio file was made with AI from Google or its partners, including OpenAI, NVIDIA and Kakao, and soon Apple. The tool is available globally in English from the day of the announcement. Southern says you upload a file at synthid.com, and that the site asks for the best-quality copy you have because that gives the most accurate result.

Google says it has watermarked over 180 billion images and videos, plus 240,000 years of audio, since it launched SynthID in 2023. Last year it introduced an early version of the detector for media professionals. Southern adds that journalists, media professionals and researchers could join a waitlist for it.

What does a result tell you?#

Southern describes a sample result in Google’s post: the panel says “SynthID was detected” and that the media was made or edited with Google AI, and it notes the file may have been edited further since. The site’s FAQ, he reports, says “no signal is completely robust” and advises collecting multiple points of data before deciding.

The site also says, per Southern, that this is not a general AI detector. It can only detect media from the companies that adopted SynthID. He draws two limits from that. A touched-up photo and one made from scratch can both read made or edited, and a file with no watermark may still be AI-generated by a company that does not use SynthID.

Is it a Google ranking signal?#

Neither Google’s post nor Search Engine Journal says so, and we found no source that does. Google presents the detector as a way for people to check media. The post adds that built-in verification features in Search, the Gemini app and Chrome now regularly handle over 1 million requests a day, which describes people checking media, not how pages rank.

The post names image, video and audio files, not text. For text, OpenAI’s own watermark is covered in OpenAI will watermark ChatGPT text in the EU, and Google’s published position on AI content is about quality, as in Google says fact-checking AI content is critical.

What should publishers do with it?#

Nothing about how you publish text changes. If you run images, video or audio from freelancers, stock libraries or AI tools, you can now check a file for a SynthID watermark before it goes live. Treat the result as one data point, as the FAQ advises. A positive result tells you a supported tool made or edited the file, not how much, and a clean result proves nothing about other tools.

Keep your own record of where each asset came from and which tool touched it. The AI search and content strategy hubs follow related changes.