Google Opens SynthID Detector to Everyone

Google has put its AI-content detection tool in everyone's hands. On Oct. 7, the company made its SynthID Detector available to the public worldwide in English, opening up a tool that had previously been limited to members of the media.

The website lets anyone upload a photo, video or audio file to check whether it carries a SynthID watermark indicating AI generation. Supported formats include JPG for images, MP4 for video, and WAV and MP3 for audio. The interface is simple by design: upload a file, and the detector reports whether it finds the watermark.

What it detects

SynthID is Google's watermarking technology for AI-generated media, and the detector checks for watermarks from an expanding list of companies. It can identify content created with SynthID technology from Google, OpenAI, Nvidia and Kakao. Apple plans to support SynthID later this year, which would bring the technology's reach to the images produced by one of the world's most popular camera platforms.

The same detection technology is also being built into Google's own products, including Search, Gemini and Chrome, meaning users may start encountering SynthID checks in places they already browse and search.

The limits

Google is careful about what the tool cannot do. The detector cannot tell the difference between media that was created by AI and media that was merely edited with AI assistance. It is not a general-purpose AI detector; it only recognizes content carrying a SynthID watermark, and anything generated by a system that does not use SynthID will slip past it.

The tool also requires users to sign in, which will limit its reach among people who do not want a Google account tied to their uploads. And because SynthID adoption is still growing, the detector's coverage of the AI content circulating online today is incomplete.

Why it matters

The public launch arrives weeks before the U.S. midterm elections, in a year when synthetic media has moved from curiosity to campaign reality. Deepfakes and AI-altered videos have already featured in elections around the world, and the United States has no comprehensive federal framework for labeling AI-generated content.

A public, free detector does not solve that problem. Watermarks can be removed, and most AI systems still do not watermark at all. But the launch represents the mainstreaming of a countermeasure that, until now, lived behind closed doors. For journalists, researchers and ordinary voters trying to verify what they see, a watermark check is now one more tool in the kit.

The harder work, building watermarking into every major AI system and making detection reliable across formats, is still ahead. Google's public detector is a start, not a finish.