A new Gemini watermark setting lets you hide the visible sparkle on AI images, videos, and songs. Google’s official Gemini account said the toggle is rolling out over the next few days in Gemini and Flow, with Search next. Invisible SynthID watermarks and C2PA metadata will always stay embedded in the background.
Where the switch lives, and what it does not touch
Google’s help page names the control Media Watermark. Specifically, open Settings on gemini.google.com or in the mobile app, tap Media Watermark, and turn Show watermark on or off. The choice applies to future images from Nano Banana, videos from Omni, and songs from Lyria. Also, it only covers Gemini Apps. It does not strip marks from other Google products.
However, turning the sparkle off does not make the file anonymous. SynthID stays in the pixels or audio. C2PA Content Credentials stay in the metadata. You can still ask Gemini if a clip was AI-generated. Josh Woodward, Google’s VP for Gemini, put it this way: the company is striking a balance between creative control and safety.
Who does not get the toggle
Meanwhile, Google keeps visible marks on in some places. If you are in India, South Korea, or Vietnam, you only see the setting with an AI Ultra plan. Otherwise the sparkle stays on. Work and school accounts never get the switch. As a result, those users keep visible watermarks no matter what they pick in Settings.
This sits next to other AI-label fights we already covered. Suno is embedding audio watermarks so platforms can spot AI songs. Also, Spotify now flags AI persona artist profiles. Google is doing the opposite of a full wipe: hide the stamp, keep the forensic mark.
Google is also open-sourcing Credentio, a C++ library for C2PA credentials, so other apps can read the same provenance. Until the rollout finishes, check Settings. If Media Watermark is missing, you are either early, on a work account, or in a country that still requires the visible mark.
