Gemini · Google · SynthID
Gemini’s watermark was born at DeepMind. It has been in the product for years.
SynthID is Google’s. They have been stamping Gemini pictures, audio and video for years, and they open-sourced the text version. Gemini is not waiting on a future model. The mark is already in the stack.
Reviewed by Can Balkaya, CTO · Published
or type below, or drop a file
Inspected in your browser. A plan is required to finish.
Choose the surface, then the tool
| Surface | Gifi action | What can be verified |
|---|---|---|
| Ordinary words | Rewrite the prose | The words changed; no private detector certificate |
| C2PA, EXIF, XMP, properties | Strip named metadata | Action list plus an independent reinspection |
| Pixels, audio, or video | Not supported | No removal claim |
They built the stamp the others licensed
Verification lives in the Gemini app, Search and Chrome. That is Google’s detector. A Gemini hit says nothing about ChatGPT or Claude. Each lab keeps its own key.
A mark in the pixels, the soundtrack or the video is out of scope. C2PA on a Gemini file can be stripped. Soft binding can still re-link it.
The words use the same dice
SynthID-Text tilts the next-word choice. Light edits usually keep the trail. A rewrite picks new words. We cannot see Google’s key.
What this cannot promise
- A rewrite does not prove the text was never AI-assisted.
- No one outside the lab can certify their detector. The key is private.
- Pixel watermarks, audio and video are out of scope.
Questions this usually raises
- Does Gemini watermark text now?
- Yes. Google has applied SynthID across text and media, and open-sourced the text method.
- Can you remove a Gemini image watermark?
- Not from the pixels. C2PA in the file, yes.
- Does a Gemini hit mean ChatGPT wrote it?
- No. One key, one lab.
Primary sources
Gemini
Statistical signal? Rewrite the prose.
A Gemini mark, if it is in the words, needs different words. Hidden characters and C2PA need the file cleaner.