Google is flipping a long-running creative complaint. Starting Friday, Gemini users can turn off the visible watermark that the company stamps on AI-made images, videos, and songs. Josh Woodward, VP for Gemini, said the toggle covers Nano Banana, Omni, and Lyria, and will live in Gemini and the Flow video editor first, with Search support coming soon. The setting path is simple: Settings, then Media Watermark.
The hidden layer does not go away. Google is keeping its invisible SynthID watermark and C2PA provenance metadata, so you can still ask Gemini or Search whether a file was generated. Woodward framed it as a balance: visible marks often wreck a graphic that a designer, marketer, or musician actually wants to ship, but the company still wants a way to check origin later. That is a quieter safety story than “no labels at all.”
For everyday users, the change is practical. A poster, a product mock, or a short clip no longer has to carry a corner badge that screams AI. For newsrooms and platforms, the real test is whether the invisible marks survive screenshots, recompression, and re-uploads. Google says the feature is rolling out over the coming days. If you make media with Gemini, check the new toggle before you assume every export is still stamped.
