Remove AI Tag from Gemini Images: What Works

You can remove the C2PA metadata from a Gemini image, but not its SynthID watermark, and not a visible watermark. Google's Gemini Apps Help says all AI media created or edited with Gemini Apps includes an invisible SynthID watermark, that Gemini Apps add C2PA Content Credentials, and that new images can also get a visible watermark. ScrubPhoto removes only the metadata, so a scrubbed Gemini image still carries Google's watermark.

Check a file for Instagram, TikTok and Facebook label triggers or scrub it now. Both run in your browser and nothing is uploaded.

What Gemini writes into an image

Google's Gemini Apps Help on watermark settings describes three layers. The first is a visible watermark on newly generated images, controlled by a setting called Media Watermark. In India, South Korea and Vietnam, Google says only AI Ultra subscribers see that setting, and everyone else there gets the visible mark automatically.

The other two layers sit outside that switch. The same page says all AI-generated media created or edited with Gemini Apps includes an invisible SynthID watermark, and that Gemini Apps include C2PA metadata about the content's origin. Google notes that the Media Watermark setting does not affect either one. In November 2025, Google also said images from Nano Banana Pro in the Gemini app would carry embedded C2PA metadata.

Why SynthID cannot be removed by a metadata tool

SynthID lives in the pixels, not in the file's metadata. Google's Gemini help says the watermark usually survives re-scaling, re-coloring, compression and other changes, though many alterations can stop it being detected. Upload an image to Gemini, Search or Chrome and, Google DeepMind says, they will look for it. ScrubPhoto does not decode or change pixels at all, so the watermark passes through intact. The full answer is in our guide can you remove SynthID.

Which platforms react to a Gemini image

A visible Gemini watermark, where there is one, sits in the picture for everyone to see. The C2PA data speaks to software instead. Meta's apps, Instagram, Facebook and Threads, watch for C2PA and IPTC AI indicators. Instagram's resulting label reads AI info, has also read AI content, and started out as Made with AI. TikTok, per its May 2024 newsroom post, applies its label on its own to content with Content Credentials. Google's own products look for SynthID. Anywhere else, like Pinterest or LinkedIn, the platform's help center is the place to check.

Remove the AI tag from a Gemini image: what scrubbing does

Run the Gemini file through the tool and the C2PA manifest goes, along with XMP, EXIF and other metadata blocks, with no loss for JPEG, PNG or WebP. That changes what the file declares to software. It does not change what the image is, and it leaves both watermarks exactly as they were.

Google's rules matter here too. Its Generative AI Prohibited Use Policy bars misrepresenting the provenance of generated content by claiming a human made it alone, in order to deceive. So the honest uses are narrow: keeping control of what a file carries when you disclose AI yourself, or seeing what the checker finds before you decide. Meta, for its part, makes disclosure mandatory for photorealistic video and realistic audio that was made or changed digitally.

A real photo edited in Gemini

If you edit one of your own photos in Gemini, Google's help says media edited with Gemini Apps carries SynthID too. A real photo with a Gemini edit is an AI-edited image, and a label saying so is accurate. If you want a picture with no AI marks at all, post the original from your camera, which never passed through Gemini. If that original came from a phone, our photo metadata page covers the location and device data it may hold.

Questions about Gemini images and AI tags

Can you remove SynthID from a Gemini image?

No. SynthID is built into the pixels, and Google says it usually survives common edits. ScrubPhoto never alters those pixels and has no SynthID detector, so the watermark stays put.

How do I turn off the visible watermark in Gemini?

Google's help describes a Media Watermark setting in Gemini Apps that turns the visible mark on or off. In India, South Korea and Vietnam, only AI Ultra subscribers see it. The setting does not affect SynthID or Content Credentials.

Will Instagram still label a scrubbed Gemini image?

The scrub removes the C2PA signal Meta reads, so the file no longer declares AI in its metadata. Any visible watermark, a self-disclosure and the platform's other checks are outside its reach, so a label is still possible.

Does ScrubPhoto detect SynthID?

No. ScrubPhoto reads and removes metadata only. For a SynthID check, Google points people to Gemini, Search and Chrome.

Sources

Official pages this page was checked against in October 2026: