How to Verify AI Media with Google SynthID Detector

AI-generated images, videos, and audio now appear everywhere. Distinguishing them from human-created media has grown harder. Google expanded its SynthID Detector on 7 October 2026 so anyone can check supported files for invisible watermarks.

The tool lives at synthid.com. It scans for SynthID marks added by Google and partners. This guide covers what the detector does, which files it accepts, and the exact steps to run a check.

What SynthID Detector checks

SynthID embeds an imperceptible watermark into AI outputs at creation time. The mark survives common edits such as cropping, compression, and filters. Detection does not rely on visual artifacts or metadata that can be stripped.

The public detector looks for these marks in images, video, and audio. It does not analyze text. A positive result means the file or part of it carries a watermark from a supported system. A negative result only means no supported watermark was found. It does not prove the content is human-made.

Google has applied SynthID to more than 180 billion images and videos and 240,000 years of audio since 2023. Built-in checks in Search, the Gemini app, and Chrome already handle more than one million verification requests daily.

Supported providers and file limits

The detector currently recognizes watermarks from Google, OpenAI, NVIDIA, and Kakao. Apple support is listed as coming soon. Each provider uses its own implementation of the SynthID standard.

File limits are clear. Uploads must stay under 100 MB. Video clips should be shorter than 90 seconds. Audio tracks should be under one hour. Supported formats include JPEG, PNG, BMP, and WebP for images; MP4, MOV, and WebM for video; and MP3, WAV, OGG, AAC, M4A, and FLAC for audio.

You sign in with a Google, Apple, or OpenAI account before scanning. Daily limits apply, typically around ten checks per media type. The interface is currently available in English.

Step-by-step: run a check

1. Open a browser and go to synthid.com.
2. Sign in with a Google, Apple, or OpenAI account.
3. Select the media type: Image, Video, or Audio.
4. Upload a file from your device or drag it into the upload area.
5. Wait for the scan to finish. Results appear within seconds for most files.
6. Review the output. If a watermark is detected, the tool names the provider and highlights affected segments on video or audio timelines.

The result screen shows whether the entire file or only portions contain the mark. It also notes that the media may have been edited further after watermarking. You can play video or audio directly in the results view.

If the option is unavailable in your region or the upload fails, confirm the file meets the size and format rules. Refresh the page or try a different browser if the interface does not load.

Limitations to keep in mind

SynthID Detector is not a general-purpose AI detector. It only finds watermarks from the listed partners. Content generated by other tools, or content that had the watermark removed or never applied, will return a negative result.

Screenshots of watermarked media often lose the mark. Heavily edited files or those processed through multiple tools may also fail detection even if they started as AI-generated. The tool does not evaluate text or code.

These limits are intentional. Google designed the system as one signal among others rather than a definitive verdict on every piece of media.

Tips for practical use

• Start with original files rather than screenshots or heavily compressed copies.
• Combine the detector with other checks. Upload the same file to the Gemini app and ask whether it was created or edited with Google AI.
• For OpenAI images, also try the provider’s own verification tool if available.
• Note the daily limit and prioritize files that matter most.
• Share the method with colleagues who review user-generated content or news media.

For Android users who also want stronger protections against scam texts, see the guide on enabling Enhanced Protection in Google Messages.

Conclusion

Google’s SynthID Detector gives a practical way to check supported AI media for watermarks. The process takes under a minute once you are signed in. Use it as one layer of verification alongside your own judgment and other provenance tools.

The global English rollout on 7 October 2026 makes the check available without a waitlist. As more partners adopt the standard, the detector covers a larger share of commonly encountered AI content.

Sources

• Google Blog: Google expands SynthID Detector for AI content (7 October 2026)
• Google DeepMind SynthID overview
• Official SynthID Detector site at synthid.com
• Coverage from Android Authority and The Verge on the public rollout