You can now turn off Google Gemini’s visible watermarks
Google has updated its Gemini platform and AI video generator Flow to give users control over visible watermarking on AI-generated content. A new "Media watermark" setting, when toggled off, removes the "sparkle" icon that previously appeared in the bottom-right corner of images, videos, and music created with Google's Nano Banana and Omni models.
The move makes visible watermarks optional rather than mandatory - a notable shift for a company that has, until now, applied them by default as a way of signaling AI-generated origin to viewers. Josh Woodward, vice president of Google Labs, Gemini, and AI Studio, confirmed that turning off the visible mark does not strip all provenance information from the content.
Behind the scenes, Google will continue to embed invisible SynthID watermarks alongside C2PA metadata in all generated media, regardless of whether the user-facing sparkle is shown. SynthID is Google DeepMind's technology for encoding imperceptible signals into AI-generated content, while C2PA - the Coalition for Content Provenance and Authenticity - is an industry-wide standard for attaching verifiable origin data to media files. Both methods are designed to allow platforms, researchers, and fact-checkers to identify AI-generated content even when no visible label is present.
The practical effect is that creators who want cleaner outputs for professional or commercial use can now do so without losing the underlying provenance trail. It also reflects a broader industry conversation about where the responsibility for labeling AI content should sit - with the tool, the platform distributing it, or the person who created it. By keeping invisible watermarks in place, Google is hedging: accommodating user preference on aesthetics while maintaining a technical record that content did not originate with a human.
