Gemini Watermark Toggle: Turn Off Google's Visible AI Mark

Google rolled out a Gemini watermark toggle on August 14, 2026, called Media Watermark, that lets users turn off the visible mark on AI images, video, and music made with Nano Banana, Omni, and Lyria. The invisible SynthID watermark and C2PA metadata still get embedded no matter what you choose.

Gemini VP Josh Woodward announced the change, calling the visible watermark one of users’ most common complaints. It’s rolling out first in the Gemini app and Flow, Google’s AI video editor, with Search support coming soon. Not everyone gets it on day one: the EU has it switched off entirely, and a few other countries need a paid subscription just to see it.

How Do You Turn Off the Gemini Watermark?

Flipping the switch takes under a minute once it reaches your account. Open the Gemini app, tap Settings, find Media Watermark, and turn it off. The change only applies going forward, not to anything you already made.

1. Open the Gemini app (or gemini.google.com on desktop)
2. Tap Settings in the bottom-left corner
3. Select Media Watermark
4. Toggle the setting off

The rollout covers three model families, each with its own watermark behavior:

Lyria is the newest of the three, and our coverage of Lyria’s launch has more on how the music model works.

The Watermark You Can’t Turn Off

Turning off the visible mark does not touch the layer Google actually relies on for provenance. Woodward was direct about this, saying that “invisible SynthID watermarks and C2PA metadata are still being used for transparency” no matter how the visible setting is configured.

SynthID embeds an invisible signal into the pixels or audio of a generated file, detectable even after edits, compression, or cropping. Content Credentials, the C2PA standard, attaches metadata documenting that a file was AI-generated. Neither one shows up to a casual viewer, and neither one is optional. Google paired the announcement with an open-source library, Credentio, for checking these markers, which makes the invisible layer easier for third parties to verify now that the visible one is optional.

Who Doesn’t Get the Gemini Watermark Toggle?

Most countries get the toggle on every personal Gemini account. India, South Korea, and Vietnam only show it to AI Ultra subscribers, the European Union doesn’t get it at all, and work or school accounts never see it regardless of region or plan.

The EU exclusion traces to Article 50 of the EU AI Act, which took effect on August 2, 2026, and requires AI-generated content to carry clear, visible labeling. Google treats the visible watermark as part of meeting that requirement, so the toggle simply doesn’t appear for EU accounts. India, South Korea, and Vietnam sit in a different category: Google’s support page states plainly that in those three countries, “you’ll only see this setting if you have an AI Ultra subscription.”

The End of DIY Watermark Removal

Before this update, a clean Gemini image meant going outside Google entirely: browser extensions, crop-and-clone tools, or GitHub scripts built to strip the corner mark off Nano Banana outputs. Android Authority had already spotted evidence of an official removal setting in a Gemini APK teardown back in July, weeks before Google confirmed it. None of the third-party workarounds were sanctioned, and running an unfamiliar script against your own files carried its own risk on top of sitting in a gray area against Google’s terms.

An official toggle closes that gap. There’s no third-party code to trust and no ambiguity about whether removing the mark violates anything, since Google is the one offering the switch.

The bigger effect lands on commercial use. Marketers and video editors dropping AI-generated clips into finished work now have one less visual tell that a shot started as a machine output. Flow already competes for that audience alongside the tools in our best AI video generators comparison, and a persistent on-screen watermark has been one of the clearer signals separating a finished edit from a rough AI draft. The invisible SynthID and C2PA layers mean nothing about a file’s origin actually disappears; it just stops being visible to whoever’s watching.

FAQ

---

Sources

1. TechCrunch: Google will now allow users to remove visible watermark from its AI generations
2. Google: Manage watermark settings in Gemini Apps
3. Android Authority: Gemini watermark removal setting
4. Business Standard: Google Gemini lets you remove visible watermark from AI images, videos