Google's SynthID checks in Gemini, Search and Chrome: how to use them

Google says you can upload an image to Gemini, or ask in Search or Chrome, and it will check for a SynthID watermark. The Gemini help page sets file and quota limits, and says Gemini can currently only recognize content from Google's own AI tools. A not-detected result does not show who made an image.

How do I check an image for SynthID in Gemini?

Google's Gemini help page on verifying AI-generated images, videos and audio gives these steps for a computer.

1. Sign in to Gemini. The page says you need a personal account or a qualified Workspace account, and the latest version of the app if you use the app.
2. Go to gemini.google.com and click "Add files". You can upload from your device, from Google Photos or from Google Drive.
3. Choose one file. Then ask a question such as "Was this image/video/audio created or edited by Google AI?" or "Is this AI-generated?"

Google's SynthID page says the same in one sentence: upload the file to your chat and ask whether Google AI created or altered it. Your file goes to Google.

What are the limits on a Gemini check?

The same help page lists these limits as of October 30, 2026. Google says the quotas are approximate, in any 24-hour rolling window.

• One file at a time, of 100 MB or less.
• Images: about 10 checks.
• Video: about 10 checks, up to 5 minutes in total. Each video must be less than 90 seconds.
• Audio: about 10 checks, up to 3 hours in total. Each file must be less than 1 hour.
• Screenshots: Google says to crop tight around the image, not to upload a collage.

Quotas change, so read the page first.

How do I check in Search or in Chrome?

Google's post of May 19, 2026 says SynthID verification for image, video and audio was added to the Gemini app first. It says Google was expanding it to Search that day, and to Chrome over the coming weeks. It names Lens, AI Mode, Circle to Search and Gemini in Chrome, and suggests asking whether the image is made with AI.

Google's post of October 7, 2026, updated October 29, says the built-in checks in Search, the Gemini app and Chrome now regularly handle over 1 million requests daily. We found no quota or file limit for Search or Chrome on the pages we opened, so the numbers above are for the Gemini app only.

What do the answers mean?

The Gemini help page describes three outcomes for SynthID.

• If a watermark is detected, Google says it means all or part of the image or video was created or edited by Google's AI models. For video, Gemini states which parts carry it.
• If a watermark is not detected, Google says it means the file was not created or edited by Google AI, but another AI system could have made it.
• If it is not clear, Google lists two reasons: "Not enough details to watermark", such as very simple or abstract content, and "Likely too small an edit".

The SynthID FAQ on the same help page adds a limit. It says the watermark will usually still exist after an image is re-scaled, re-colored or compressed, but there is a chance that after many alterations it will not be detected. So a not-detected answer describes the file you uploaded, in the state you uploaded it.

Which images can these checks recognize?

The Gemini help page says that while other companies have started to adopt SynthID watermarks, Gemini can currently only recognize content created by Google AI tools. The separate SynthID Detector is wider. Google's October post says it checks whether a file was made with AI from Google or its partners, including OpenAI, NVIDIA, Kakao and Apple. The page at synthid.com says it is not a general AI detector and can only find media from the companies it lists.

On October 30, 2026 that page showed a terms and privacy dialog. It says uploaded files are deleted once results return, and that a digital signature of your media, tied to your user ID, is kept for 24 hours to enforce daily quotas. We did not accept the dialog or upload anything.

What did we measure about this route?

Our checker does not read SynthID, so we cannot test the Google checks with our files. We can report two things.

First, on October 30, 2026 we read the subject of every certificate in the copy of the C2PA Trust List that our checker uses (32 signer anchors, from a list commit dated October 17, 2026). Google LLC holds 6 of the 32, the largest block of any organization. The next largest is 4, held by DigiCert. A list shows who may sign, not what any Google product adds to a file.

Second, our test set holds 5 real files filed in Wikimedia Commons categories named "Images generated by" Gemini (2 files), Nano Banana (1) and Imagen (2). None carries a Content Credentials record. Four read "No supported provenance signals found". One Gemini PNG reads "Metadata declares AI use", because unsigned XMP in the file carries an IPTC source type for trained algorithmic media. That is a label anyone can add or remove. Third parties uploaded all five, so we cannot know how each was handled. We did not run them through Google's checks, so the files give the Gemini route nothing to confirm or contradict. The counts are on our accuracy page.

What this means when you check an image

Open the checker, choose the file, and read the headline. For 4 of our 5 files filed under Gemini, Nano Banana or Imagen, the headline is "No supported provenance signals found". It means we found no signed record, no source label or tool name in the metadata and no Stable Diffusion watermark pattern. It does not mean the image is real or human-made.

That headline says nothing about SynthID. Every result carries a line that begins "Not checked here: Google's SynthID watermark", with a link to Google's detector. If you want Google's answer, use one of the routes above, and keep the original file.

The checker's limits line applies: "These checks cannot determine whether an image is AI-generated." Treat neither tool's answer as a verdict.

Guides on this topic

• What is SynthID? is the main page on this topic.
• Gemini and Nano Banana: SynthID and Content Credentials covers what Google says its image models add.
• Does OpenAI use SynthID? covers the partner that Google names first.
• SynthID for video and audio covers the other file types.

Frequently asked questions

How many images can I check a day?

Google says about 10 image checks in any 24-hour rolling window in the Gemini app. We found no stated number for Search or Chrome.

Will Gemini find SynthID from OpenAI or other partners?

Google's Gemini help page says Gemini can currently only recognize content created by Google AI tools. Google's October post says its separate SynthID Detector checks files made with AI from Google or its partners.

Does a not-detected answer mean a person made the image?

No. Google says another AI system could have made it, and that heavy alteration can stop a watermark being found.

Does New Watermark run this check for me?

No. Every result says we did not check SynthID.