Most coverage treated this as Google going soft on AI labels. The useful split is narrower: the logo is optional, while the less-visible identification remains.
Google now lets you hide its visible AI watermark. The invisible mark stays
Google now lets people turn off the visible watermark on images, videos, and music they create in Gemini Apps. Google’s August announcement also included Flow. The switch is Settings > Media Watermark. Turning it off does not disable SynthID, Google’s invisible watermark, or C2PA Content Credentials, the file-history metadata attached to the media.
We checked Google’s current Gemini watermark help page, its Flow help, Josh Woodward’s August 14 announcement, and same-day reporting from TechCrunch and The Verge. The current help page is the useful receipt because it spells out who gets the setting and what the switch does not change.
If you have been making a birthday card in Gemini and watching a sparkle sit in the corner like you did something wrong, this is the practical change. The picture can look clean. Google’s less-visible identification layers still remain.
This began as a company announcement on X, so it deserves company-announcement caution. Google documents what its own setting and detection tools are designed to do. That is not an independent test of whether SynthID survives every crop, screenshot, filter, or re-save.
What changed is the visible layer. In Gemini Apps, the setting covers newly generated images, a persistent overlay on generated video, and the visual player created with a music track. The current help page says people in India, South Korea, or Vietnam see the switch only with an AI Ultra subscription; otherwise the visible mark stays on. Work and school accounts do not get the switch.
What did not change is the identification underneath. Google says all AI-generated media created or edited in Gemini Apps includes invisible SynthID, and that Gemini Apps also adds C2PA Content Credentials. Gemini can check images, video, and audio for SynthID. Google says Search can also help check an image’s origin. These are useful signals, not a promise that every copied or altered file will always be identifiable.
The significance assessment, measured against this reader’s life: this is a cleaner-output setting, not a privacy revolution. If you create the media, you can decide whether the visible logo shows. If someone sends you a picture with no logo, that still does not prove a person made it.
I am mildly relieved for the people who only wanted a clean photo of a cake. I am not soothed by the idea that everyone will check the hidden evidence. Most people will not.
This is the situation as of September 12, 2026. Availability, subscription rules, and Google’s verification tools can change. Check Settings > Media Watermark in the product and account you actually use before assuming the switch is available.
If you make a holiday card in Gemini, look for Media Watermark before the next generation. If a missing sparkle is being offered as proof that an image is real, the honest answer is simpler: a visible logo can be switched off.
Hiding the stamp does not hide how the picture was made. It just stops shouting about it.
Limitation:
This is the situation as of September 12, 2026. Availability, subscription rules, and Google’s verification tools can change; readers should check Settings > Media Watermark in the product and account they use.
In conversation:
Google now lets eligible Gemini users hide the visible watermark. The invisible SynthID mark and Content Credentials remain, so a picture without a logo is not proof a person made it.
Questions people ask
If I turn the visible watermark off, can people still tell my picture was made with Google AI?
They may be able to. Google says media created or edited in Gemini Apps still contains invisible SynthID and C2PA Content Credentials, and Gemini can check for SynthID. Those are identification signals, not a guarantee that every screenshot or heavily altered copy will still be detected.
Does the setting remove a watermark from pictures I already made?
Do not assume it does. Google’s current help page describes the setting in terms of newly generated images and generated video or music players. Turn the setting off before creating a new file, then inspect the result.
Does the Gemini watermark switch work for every account and country?
No. Google says work and school accounts do not get the setting. In India, South Korea, and Vietnam, the switch appears only for AI Ultra subscribers; without that plan, Gemini Apps apply the visible watermark automatically.