You can now turn off Google Gemini’s visible watermarks
Source Entity
Emma Roth
Google is introducing a toggle to remove visible 'sparkle' watermarks from AI-generated media created by Gemini, Flow, and other models. Despite this change, Google will maintain underlying security through invisible SynthID watermarks and C2PA metadata standards.
Evolution of AI Content Labeling
Google has officially announced a significant update to its generative AI ecosystem, granting users the ability to disable visible watermarks on content produced by its models. This policy shift applies to images, videos, and audio generated through platforms like Gemini and the video editor Flow. By introducing a 'Media watermark' toggle, Google is shifting the default aesthetic experience for users who find the existing 'sparkle' icon in the bottom-right corner disruptive to their professional or creative workflows.
The Dual-Layer Security Approach
While the visible indicator is becoming optional, Google remains committed to transparency through robust, invisible back-end protocols. The company confirmed that disabling the visible watermark does not strip away essential identification markers. Specifically, Google’s proprietary SynthID technology and the industry-standard C2PA metadata will continue to be embedded within the files. This ensures that while the final output may look cleaner for the end user, the provenance of the content remains verifiable by systems designed to detect AI-generated material.
Impact on Creative Workflows
For creators, designers, and multimedia professionals, the presence of a mandatory watermark has often been a point of friction, limiting the immediate utility of AI-generated assets in commercial or high-end artistic contexts. By allowing users to toggle these off, Google acknowledges that professional-grade content requires a level of aesthetic freedom that a permanent, branded watermark previously inhibited. This change is being rolled out across specific model architectures, including Nano Banana, Omni, and Lyria.
Implementation and Future Roadmap
Josh Woodward, Google’s VP for Gemini, confirmed that the toggle feature is being integrated directly into Gemini and the Flow video editor. While the rollout is currently focused on these creative suites, the company has indicated that support for Search-based generations is on the horizon. This phased approach suggests that Google is carefully monitoring the balance between user autonomy and the broader societal need for AI transparency.
Broader Implications for AI Ethics
This development highlights a growing trend in the tech industry: moving away from 'obtrusive' labeling in favor of 'intrinsic' verification. By prioritizing C2PA standards and SynthID over visible icons, Google is aligning itself with global efforts to standardize how AI-generated content is tracked. As the line between human-made and machine-made content continues to blur, these invisible cryptographic signatures will become the primary method for maintaining digital trust without compromising the user experience.