SynthID Detector opens to everyone, checking media from Google and partner AI tools

Google DeepMind opened SynthID Detector, its tool for recognizing AI-generated content, to everyone on October 7, 2026. The tool is available worldwide at synthid.com, in English for now. According to Google's announcement, anyone can now check whether an image, video or audio file was made with Google's AI tools or with tools from partners such as OpenAI, NVIDIA and Kakao. The announcement also names Apple as a partner coming soon.

An early version of the tool was introduced last year for journalists and other media professionals. With this release, information about how a piece of online content was made becomes available to ordinary users as well. Google says the Detector sits alongside the built-in verification features in Search, the Gemini app and Chrome.

SynthID is an invisible watermark embedded in content

SynthID is a digital watermarking method developed by Google DeepMind. When an AI model generates an image, video, audio clip or text, a signal that people cannot see or hear is placed inside the content. According to Google DeepMind's page, the signal is woven into the pixels of images and video and into the frequency components of audio. Labels and metadata attached to a file can be lost when it is shared; SynthID is part of the content itself, so it travels with the content as it is shared and reshared. The page says the watermark is designed to stay detectable after common changes such as adding filters, changing colors and saving with lossy compression.

The Detector scans images, video and audio

Users upload a file to synthid.com and the tool checks it for a SynthID watermark. According to the site's FAQ, the tool currently supports three content types: for images it reports whether a watermark is present, and for video and audio it shows the segments that carry the watermark. SynthID can also watermark text; the types the Detector supports today are images, video and audio. When a watermark is found, the tool reports that the content was likely generated or edited by a SynthID-enabled AI model. Google recommends uploading the highest quality version of a file for the best results.

Covered companies and Apple's timing

The Detector recognizes content made with tools from companies that have added SynthID to their products. The site lists them as Google, OpenAI, NVIDIA and Kakao. In an announcement on May 19, 2026, OpenAI said it was adding SynthID watermarks, through a partnership with Google, to images created with ChatGPT, Codex and the OpenAI API. According to a July 31 update on the same page, supported audio content created with these tools now carries SynthID watermarks too.

The coming soon label for Apple depends on Apple's own update schedule. According to Apple's announcement of September 14, 2026, SynthID support will arrive in a software update later this year. With it, images created in Image Playground can be identified, and most images edited with Apple Intelligence will carry the watermark, depending on the edits applied. Google DeepMind has not given a date for when Apple will be added to the Detector.

Access, terms and privacy

The tool is available in all countries, in English. Google's announcement does not mention an age limit or pricing. The service is governed by Google's Terms of Service; the tool's additional terms refer to signing in with an account and to daily usage limits. They prohibit attempts to break or bypass the watermark and automated bulk queries that bypass the interface. According to the privacy notice, uploaded files are processed in real time and deleted right after the results are returned; to prevent abuse, a digital signature of the file is stored with the user's identifier for 24 hours.

According to Google DeepMind's page, the Detector recognizes only content that carries a SynthID watermark; content made with tools that do not use SynthID, or created before a company adopted it, cannot be identified with this tool. The same page says that a watermark not detected result does not fully rule out AI generation, that heavy editing and compression can weaken the signal, and that in very rare cases the tool can give a false result on content without a watermark. Google recommends gathering additional evidence before reaching a conclusion. For a fuller picture, follow evaluations from independent researchers as well.

Sources: Google: Making it easier to identify AI-generated content globally, SynthID Detector, Google DeepMind: SynthID, Google DeepMind on X, OpenAI: Advancing content provenance, Apple Newsroom: Major updates for Apple software platforms.