Google has introduced a “Media Watermark” setting in Gemini that allows users to toggle off visible watermarks on AI-generated images, videos, and songs. According to Josh Woodward, Google’s VP leading Gemini, AI Studio, and Google Labs, the update responds to high user demand for more creative control, though invisible SynthID watermarks and C2PA metadata remain embedded for transparency.
How to Disable Visible Watermarks in Gemini and Flow
Users can now remove the visible “sparkle” icon—typically located in the bottom-left corner of media—through a new menu.
Josh Woodward confirmed via X (formerly Twitter) that this toggle applies to several Google AI models: Nano Banana for images, Omni for videos, and Lyria for songs. The feature is rolling out across Gemini and Google’s AI video editor, Flow, with a similar option planned for Google Search.
Pro Tip: If you can’t find the Media Watermark toggle in your settings yet, check back in a few days. Google is rolling the feature out incrementally.
Invisible Tracking: SynthID and C2PA Metadata
Removing the visible icon does not make the content “unmarked.” Google continues to use SynthID, an invisible watermarking technology, to ensure AI-generated content can still be identified. Woodward stated that the company is balancing “creative control and security” by keeping these hidden markers and C2PA metadata intact.
This dual-layer approach means that while a professional creator can remove a distracting icon for a presentation or project, the file itself still carries a digital signature. According to Woodward, users can still use Gemini or Google Search to verify if a piece of media was produced by AI.
User Demand Driving AI Interface Changes
The decision to make watermarks optional follows direct feedback from the community.
This shift highlights a tension in AI development: the need for clear labeling to prevent misinformation versus the needs of professional users who find visible stamps disruptive to the aesthetic of their work.
Did you know? Certain countries may still require visible watermarks by law. In those regions, the option to toggle the watermark “Off” will not be available.
Comparison of AI Watermarking Methods
| Marker Type | Visibility | User Control | Purpose |
|---|---|---|---|
| Sparkle Icon | Visible | Optional (Toggle) | Immediate Identification |
| SynthID | Invisible | Permanent | Technical Verification |
| C2PA Metadata | Embedded | Permanent | Provenance Tracking |
Frequently Asked Questions
Can I remove all watermarks from Gemini AI?
No. While you can remove the visible icon, Google’s invisible SynthID watermarks and C2PA metadata remain embedded in the file.
Which Google AI models support this feature?
The toggle applies to Nano Banana (images), Omni (videos), and Lyria (songs).
Where is the watermark toggle located?
It is found under Settings > Media Watermark in the Gemini interface.
Join the Conversation: Do you think AI-generated content should always have a visible label, or should the choice be left to the creator? Let us know in the comments below.