For more than a year, Google’s Gemini-generated images and videos have carried a small Gemini sparkle in the corner, making it easy to tell that the content was created using Google’s AI tools.
While that can be useful for transparency, the visible mark has also been a frustration for businesses, designers and creators who want to use AI-generated visuals in presentations, campaigns and other professional work without a Google logo sitting on the final image or video.
That is now changing. On August 14, Google announced that users will soon be able to turn off the visible watermark through a new Media Watermark setting in Gemini. The option will also cover AI-generated images, videos and music, with the rollout taking place over the coming days.
What Exactly Is Google Changing?
Google is not removing AI watermarking altogether. It is giving users control over the visible watermark that appears on AI-generated content.
As Josh Woodward, VP at Google, shared on X, users can now toggle visible watermarks on or off in Gemini and Flow, with Search coming next.
✅ Papercut fixed: You can now toggle visible watermarks on or off in Gemini and Flow, with Search coming next.
This applies to watermarks on all images (Nano Banana), videos (Omni), and songs (Lyria) except in countries where it’s required by law to keep them. pic.twitter.com/utHN0yDmD3
— Josh Woodward (@joshwoodward) August 14, 2026
The change applies to watermarks on images created with Nano Banana, videos created with Omni, and songs created with Lyria. There is one exception: the option will not be available in countries where the visible watermark is required by law.
Woodward said Google is trying to balance user choice with transparency:
“We’re striking a balance here between creative control and safety.”
He added that while visible watermarks are now optional, invisible SynthID watermarks and C2PA metadata will continue to be used. This means users can still use Gemini or Search to check whether an image was generated by AI.
This distinction is important because Google is actually using three different layers of AI identification.
1. Visible Watermark
This is the small Gemini sparkle or logo that appears on AI-generated images, videos and other media. It is the part users will now be able to turn off.
For businesses, creators and designers, this is the change that matters most. They can use Gemini-generated content without having the visible Google AI mark appear on the final creative.
2. SynthID
SynthID is Google’s invisible watermarking technology. Instead of placing a visible logo on the content, it embeds an imperceptible signal into AI-generated images, videos, audio and text.
The idea is that the signal can remain detectable even after common changes to the content, making it useful for identifying AI-generated material without affecting how it looks or sounds.
Google says SynthID has now been used on more than 10 billion pieces of content across images, video, audio and text.
So turning off the Gemini sparkle does not mean Google is removing its invisible AI watermark.
3. C2PA and Content Credentials
The third layer is C2PA, short for the Coalition for Content Provenance and Authenticity. Unlike SynthID, C2PA is not Google’s invisible watermarking system. It is a provenance standard that can attach machine-readable information about where content came from and how it was created or modified.
Google says C2PA metadata will continue to be used even when users turn off the visible watermark.
This means the final image may look completely clean to a person viewing it, while still carrying information that compatible tools can use to understand its AI origin.
The difference between these three layers is important. The visible Gemini watermark is being made optional. SynthID and C2PA are not being removed. Google is essentially moving AI transparency away from something everyone can see toward systems that can be checked by software when needed.
The Internet Had Already Built a “Gemini Watermark Remover” Industry
The internet has a simple habit when people face a problem: someone builds a tool to solve it. The Gemini watermark was no different.
Our research found multiple products specifically built to remove Google’s Gemini, Nano Banana and Veo watermarks before Google officially gave users the option to turn them off. From Chrome extensions and browser-based services to open-source projects, developers had already found ways to remove the visible mark from AI-generated content.
AI Toolbox
One of the more established examples is AI Toolbox. In a June 2026 report, the company said its Chrome extension could automatically detect Gemini’s full-resolution image downloads, remove the visible sparkle and save the cleaned image.
According to the company, the extension had:
- 35,000+ users
- A 4.5/5 rating on the Chrome Web Store
- Free and paid plans
- Automatic watermark removal
- Local processing
- A reverse-alpha-blending technique
The technical approach is also interesting. Rather than simply using generative AI to paint over the watermark, AI Toolbox says it uses calibrated masks and reverse alpha blending to reconstruct the pixels underneath the semi-transparent watermark.
In simple terms, the tool attempts to work backward from the visible watermark to recover what was underneath it.
Gemini Watermark Remover
Another Chrome extension called Gemini Watermark Remover had already reached 50,000 users by July 2026, according to its Chrome Web Store listing.
The extension works directly with Gemini-generated images and offers tools to preview, copy and download the processed image. More importantly, its developers make a clear distinction between the visible Gemini logo and Google’s invisible provenance technology.
The extension specifically states that it does not remove SynthID.
That 50,000-user figure is significant because it shows this was not just a niche technical experiment. A large number of Gemini users were actively looking for a way to remove the visible watermark.
PureGemini
Another extension, PureGemini, had crossed 5,000 users and offered one-click removal for Gemini images as well as videos generated using Omni.
Its documentation also makes the same distinction: removing the visible watermark does not mean removing Google’s invisible SynthID watermark.
That distinction matters because most users probably think of the small Gemini sparkle as “the watermark.” Technically, Google’s content-identification system goes much deeper than that visible mark.
GeminiWatermark.io
Developers also built dedicated web services around the same demand.
GeminiWatermark.io markets itself specifically as a browser-based solution for removing Gemini and Veo watermarks from images and videos. The service says processing takes place locally in the browser, meaning the content does not need to be uploaded to a remote server for processing.
The fact that there are dedicated websites for this purpose shows how quickly a single product limitation can turn into its own small software category.
iLoveWatermark
Even broader watermark-removal platforms joined in.
iLoveWatermark was publishing a dedicated guide for removing the Gemini watermark as early as March 2026, giving users instructions specifically for Google’s AI-generated images.
This is an important detail because it shows the demand was not limited to developers building Gemini-specific extensions. Existing image-editing and watermark-removal companies were also starting to target Gemini users.
Open-source projects
The trend went beyond commercial tools.
An open-source Gemini watermark-removal project on GitHub had accumulated around 5,100 stars and 731 forks when we checked it. The project uses client-side processing and reverse alpha blending to remove Gemini’s visible image and video watermarks.
There was also a separate open-source project specifically designed to remove Veo video watermarks, again using a mathematically precise reverse-alpha-blending approach.
Google is not inventing the capability. It is productizing it.
Google has not suddenly discovered that users want watermark-free AI content. The market had already demonstrated that demand. Developers had built extensions, websites and open-source tools specifically because Gemini users wanted a clean version of their generated content.
Google is now bringing that capability inside its own products.
The difference is that Google’s official approach also keeps SynthID and C2PA in place. So unlike many of the third-party tools that emerged around the visible watermark, Google’s solution is designed to give users a clean visual output while retaining the underlying systems that can help identify the content as AI-generated.
In other words, Google has effectively productized a capability that a small ecosystem of third-party developers had already productized.
