Google has opened a website that lets anyone check whether an image, video or audio clip was generated using AI. The service uses SynthID, the company’s watermarking technology, to look for identifying marks in uploaded media.
The launch makes a tool previously tested by selected journalists, media professionals and researchers available to the public. It offers a way to investigate a piece of media, though the source notes that watermarking tools do not always identify content made by their own creators’ models.
What the SynthID website checks
The website accepts a range of common media files. Images can be uploaded in JPG, JPEG, PNG, BMP, WEBP, AVIF, HEIC, HEIF, TIFF, TIF, and GIF formats. For video, it supports MP4, MOV, and WEBM.
Audio checks are available for WAV, MP3, OGG, FLAC, AAC, and M4A files. This broad format support lets users check different kinds of media through the same service, rather than relying on a separate verification process for each format.
The purpose is to determine whether a file contains a SynthID watermark. The result can offer a clue about how media was made, but it should be understood as a check for this particular watermarking system.
Where SynthID appears
Google introduced SynthID in 2023. The company says its Nano Banana, Veo and Lyria models use the technology to watermark generated media. It is also used by generation-capable tools including Gemini, Flow, ProducerAI and Vids.
Google has also built SynthID verification into the Gemini app and Google Chrome. The company says people currently make 1 million requests each day to verify content, suggesting that checking the origin of media has become a frequent use case for its tools.
The new website expands access beyond the groups who tested it. People can now use it directly to examine supported image, video and audio files.
Watermarks offer clues, not certainty
SynthID is one approach to identifying AI-generated media, but it is not the only one. Microsoft and Meta have their own standards for watermarking and verifying content. OpenAI and NVIDIA also support SynthID, while OpenAI maintains a separate site for checking content.
The source also says Apple is said to be adding support soon. That detail points to broader interest in tools that mark or check generated media, but different systems may use different standards and services.
There is an important limitation: these tools are not infallible. They often fail to identify content created by their makers’ own models. A missing watermark result therefore does not, by itself, establish that a file was made by a person or that it is authentic.
How to interpret a check
A SynthID check is most useful as one piece of information about a file. If the service identifies a watermark, that can help indicate that the media was generated with tools using SynthID. The tool’s stated limits mean that a result should not be treated as a universal verdict on a file’s origin.
For users, the practical change is straightforward: a public Google website now provides a place to check supported media for SynthID. The service adds access to existing options in Gemini and Chrome, while the limitations of watermark detection remain relevant when interpreting what a check says.