Google AI watermark removal is being introduced across Gemini and Flow, allowing users to switch off the visible label on generated images, videos and music while Google’s hidden provenance tools remain active.
The change applies to Nano Banana images, Omni videos and Lyria songs. It will reach users through the Media Watermark option in Settings over the coming days, with support for Search planned later.
Josh Woodward, Google’s Vice President for Gemini, said the company was responding to demand for cleaner media.
“We’re striking a balance here between creative control and safety,” he wrote in an announcement on X.
The switch removes the visible sparkle placed on generated media. Google says the option will remain unavailable in countries where visible identification is required by law.
Hidden Markers Will Remain
Turning off the label will not remove Google’s invisible SynthID watermark or metadata created under the C2PA standard. Those markers preserve information about a file’s origin without placing a logo over the content.
Users will still be able to ask Gemini or Search whether an image was generated by artificial intelligence. That matters because visible labels can be cropped or edited out, while hidden signals provide another way to examine a file.
Google is also releasing Credentio, an open source C++ library that allows developers to validate C2PA Content Credentials within their applications.
The library can inspect signatures, manifests and other credential information locally without requiring media files to be sent to a remote server.
The change comes as technology companies test different approaches to identifying synthetic content. Anthropic recently moved in another direction by introducing watermarks for material produced through Claude.
Google’s decision is narrower than it first sounds. Creators can remove the branding, but the technical trail is meant to remain.
READ ALSO: Apple Warns Users in 110 Countries Over Suspected Spyware Attacks
