How to Use Google’s SynthID Detector to Check AI Content

How to Use Google’s SynthID Detector to Check AI Content

Google’s SynthID Detector is now open to everyone. On 7 October 2026 the company made the tool at synthid.com available worldwide in English so anyone can upload an image, video, or audio file and check for an invisible SynthID watermark. TIME later named SynthID one of the Best Inventions of 2026. The watermark is added at creation by Google’s models and by partner systems from OpenAI, NVIDIA, and Kakao (Apple support is expected soon). It survives many common edits such as cropping, filters, and compression.

A positive result is strong evidence that supported AI tools helped create or edit the file. A negative result does not prove the file is human-made. The detector only looks for SynthID; it is not a general AI detector.

What SynthID Detector Actually Checks

SynthID embeds an imperceptible digital watermark into AI-generated images, audio, video, and (in some cases) text at the moment of creation. Google states that more than 180 billion images and videos and 240,000 years of audio have already been watermarked.

The public detector portal accepts:

• Images: JPG, JPEG, PNG, BMP, WEBP, AVIF, HEIC, HEIF, TIFF, TIF, GIF
• Video: MP4, MOV, WEBM
• Audio: WAV, MP3, OGG, FLAC, AAC, M4A

Text is not supported. File size is limited to 100 MB. Videos must be shorter than 90 seconds and audio shorter than one hour. You must sign in with a Google, Apple, or OpenAI account. Google applies a daily quota of approximately 10 checks per user (images, video, and audio combined) to reduce the risk of people using the tool to develop watermark-removal methods.

Files are processed and then deleted; the service does not keep the original media.

Step-by-Step: Checking a File

1. Open a browser and go to https://synthid.com.
2. Sign in with a Google, Apple, or OpenAI account when prompted.
3. Accept the terms and privacy notice.
4. Drag the highest-quality original file you have into the upload area, or click Select file. Avoid screenshots or heavily re-compressed social-media downloads when possible; the watermark is designed to survive many changes, but the original file gives the clearest result.
5. Wait a few seconds while the portal scans the file.
6. Read the result. If a watermark is found, the page indicates that SynthID was detected and, for images, highlights regions most likely to contain the mark. For video and audio it can show specific segments.

You can also check Google-origin content inside the Gemini app by uploading the file and asking whether it was created or edited with Google AI (or using the @synthid shortcut). The dedicated portal covers partner watermarks as well.

Understanding the Results

• Watermark detected: The file (or part of it) carries a SynthID mark from Google or a supported partner. Treat it as AI-generated or AI-edited content from those systems.
• Watermark not detected: No SynthID signal was found at a confident level. The file may still be AI-generated by a tool that does not use SynthID, or it may have been so heavily altered that the watermark is no longer recoverable. It may also simply be a real photograph or recording.

The detector does not distinguish between full generation and light editing, nor does it judge truthfulness. It only reports the presence or absence of the watermark.

Practical Tips for Better Results

Use the original file whenever you can. Social apps often re-encode media; downloading the source version improves detection chances. For videos and audio, pay attention to the segment view—the watermark may cover only part of the file (for example, an AI voice-over on otherwise real footage).

Cross-check important items with other signals. Look for C2PA content credentials when available, examine metadata, and apply ordinary critical thinking. The same verification capability is also built into Google Search, the Gemini app, and Chrome, so you may already see labels in those products.

Respect the daily quota. Automated bulk checking or multiple accounts to bypass limits violates the terms and can result in access being suspended.

If you work with Android and Gemini features regularly, the same underlying technology appears in several Google products. For related Android AI updates see our guide on Android Halo and Gemini Spark status bar features.

Limitations You Should Know

SynthID Detector only finds watermarks from Google and the partners that have adopted the system. Content from other models, open-weight tools, or older files created before watermarking was enabled will return “not detected.” Extreme edits, screenshots of AI images, or aggressive compression can also reduce reliability, even though the system is designed to be robust against everyday changes.

The interface is currently English-only. Results are binary (detected / not detected) rather than a confidence percentage in the public portal.

Conclusion

The public SynthID Detector gives anyone a straightforward way to check whether an image, video, or audio file carries the invisible watermark used by Google and major partners. Sign in at synthid.com, upload the highest-quality file you have, and read the result carefully. A positive detection is useful evidence of supported AI involvement. A negative result only means the specific watermark was not found.

Combine the tool with ordinary source checking and other provenance signals. As more companies adopt SynthID, the practical value of a quick upload check will continue to grow.

Sources

• Google Blog: SynthID named to TIME’s Best Inventions of 2026 (15 October 2026)
• Official SynthID Detector portal: https://synthid.com
• Ars Technica and other contemporaneous reporting on the 7 October 2026 public rollout, supported formats, partner coverage, and daily quota
• Google DeepMind SynthID documentation on watermarking scope and volume