Google Opens SynthID Detector to Everyone for Spotting AI-Made Images, Video and Audio

Google has made its SynthID Detector available to the public, giving anyone a way to check whether an image, video or audio file was made with AI - though the tool comes with real limits.

SynthID, introduced in 2023, embeds an imperceptible watermark in content created by Google's Gemini model. The detector scans an uploaded file for that mark and returns a simple yes or no. Users visit SynthID.com, upload a file and wait a few seconds for a result. A limited version launched in 2025 for select researchers and journalists before this wider release.

The tool also recognizes watermarks from Google's SynthID partners, meaning it can flag content made with AI models from OpenAI, NVIDIA and Kakao alongside Gemini. Apple participates in SynthID, but detector support for its creations is still in the works.

Because it checks for a watermark rather than analyzing media for AI patterns, false positives should be eliminated - but the same simplicity creates blind spots. It cannot detect content from AI models that don't use SynthID, such as Meta's tools, and it only reports whether a label is present, not which parts of a file were AI-generated.

Google says SynthID signatures persist even after AI generations are edited. To keep people from probing the detector for ways to strip the watermark, users are reportedly limited to about 10 checks per day and must be logged in to a Google, OpenAI or Apple account.

Rival detectors exist, but each looks for its own marker. OpenAI's tool picks out AI-confirming metadata and SynthID watermarks, while Meta's Content Seal detector was found in a 2026 Reuters report to sometimes miss its own watermarks. Anthropic and Microsoft have begun watermarking AI media but offer no consumer detection tools. No reliable detector spots generative AI across the board, and openly available general-purpose tools are prone to flagging human-made work as AI.