Google is flipping a switch that will let users scrub the visible watermark from AI-generated content across its ecosystem. The toggle, announced Friday, applies to images, videos, and songs produced by the Nano Banana, Omni, and Lyria models, according to Josh Woodward, Google’s VP for Gemini. Users will find the setting in Gemini and the Flow video editor, with Search support arriving later.
The move doesn’t touch Google’s invisible safeguards. SynthID watermarks and C2PA metadata remain embedded regardless of what users choose, which means anyone can still verify whether a piece of media came from a Google model. Woodward framed it as a balancing act: “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.”
For creators who’ve been exporting AI assets with a branded stamp in the corner, this removes a real friction point. Visible watermarks have been the source of steady complaints from professionals who want clean outputs for client work or content pipelines. Google is betting that invisible attribution is enough to satisfy transparency demands while giving users the polished files they actually need.
Google is also open sourcing Credentio, a library that lets developers build local validation directly into their apps — a signal that the company wants verification to become infrastructure rather than a UI feature. The timing is hard to ignore. Anthropic recently caught heat for watermarking Claude’s text and files to comply with EU regulations, and Google’s move positions it as the more creator-friendly option. Whether regulators see it that way is another question. The EU’s appetite for visible labeling hasn’t exactly softened, and a toggle that hides watermarks could draw scrutiny even if the invisible markers remain intact.
