You can now turn off Google Gemini’s visible watermarks
TL;DR
Google is making the visible watermark on AI-generated content optional. A new Media watermark setting in Gemini and in Google's video generator Flow lets you switch off the sparkle mark that Nano Banana and the Omni models place in the bottom-right corner. The labelling does not disappear entirely: according to Josh Woodward, VP of Google Labs, Gemini and AI Studio, SynthID watermarks and C2PA metadata stay embedded in the file.
Nauti's Take
For anyone using AI images in production this is a clear advantage, since the sparkle mark simply got in the way in thumbnails, ads and slide decks. The risk lands on the audience, because without a visible cue nobody spots generated content at a glance, and almost nobody checks SynthID or C2PA in daily use.
Teams delivering client work should set their own disclosure rules rather than leaving it to the tool.
Briefingshow
Sichtbare Kennzeichnung war bisher der einfachste Weg, generierte Bilder im Alltag zu erkennen. Google verlagert diese Aufgabe jetzt auf technische Verfahren, die im Feed, im Chat oder in einer Präsentation praktisch niemand prüft. Damit wird Kennzeichnung zu einer redaktionellen Entscheidung, die jedes Team für sich treffen muss.
Decide where you will label AI images yourself, for example in captions or in the credits of client work. Also check whether your approval process quietly relied on the visible watermark as a control.
Relevant for newsrooms, agencies, marketing teams and anyone publishing generated images.
Watch whether platforms or regulators mandate visible labelling while providers make it optional.
Solid: two independent outlets reporting, plus an on-the-record statement from a Google executive on SynthID and C2PA.
Sources