Google will now allow users to remove visible watermarks from its AI generations, which include images, videos, and songs. The feature is rolling out in the coming days, letting users easily toggle the mark on or off.
To access this new control, users can navigate to Settings > Media Watermark. As of 2026, there was previously no official opt-out for the visible watermark on Gemini-generated images, meaning this update marks a significant shift in Google's policy.
Josh Woodward, the vice president of Google Labs, Gemini, and AI Studio, announced the update in a post on X. He stated that the new toggle supports outputs generated by the Nano Banana, Omni, and Lyria models.
Users will initially find the setting to turn off the visible mark inside Gemini, which was formerly known as Bard. It is also available in Google’s video editor, Flow, while support for turning off the watermark in Google Search's AI Mode is coming soon.
Invisible Tracking Remains
Although visible markers are now optional, AI-generated content will still have invisible SynthID watermarks and C2PA standard-related metadata embedded in the background. Woodward specified that this background tracking remains fully intact to maintain system transparency.
“We’re striking a balance here between creative control and safety: while the visible watermarks are now optional, invisible SynthID watermarks and C2PA metadata are still being used for transparency," Woodward wrote on X.
Because these invisible markers are retained in the files, official verification tools will continue to function normally. "So you can still use Gemini or Search to see if an image was AI-generated,” Woodward explained.
Creative Control and Industry Standards
Previously, Gemini typically applied a visible 48px watermark for standard images. It also applied a larger 96px watermark for high-res outputs to explicitly label the media.
For visual use cases—like building a slide deck, designing a website, or publishing an Instagram post—having a clean image is highly valuable. In fact, for 99% of users, removing the visible logo is the main priority when downloading AI creations.
While this official change might make it more difficult to quickly discern if an image or video is AI-generated, many users have already been editing out visible watermarks. Some users have relied on third-party tools when speed and batch processing mattered more than pixel-level control.
Google's shifting approach mirrors broader industry practices regarding AI transparency. Competitors like OpenAI also rely on invisible SynthID tracking and C2PA to signal whether something is AI-generated, rather than forcing visible logos.
Meanwhile, Meta has launched its own Content Seal standard to track origins across its platforms.