Announced August 14, 2026, a new Settings toggle lets you hide the visible corner watermark on Nano Banana images, Omni video, and Lyria music. Invisible SynthID watermarks and C2PA provenance metadata stay embedded either way, and the control carries regional and account exceptions.
2026-08-14 · by Moe Ameen
On August 14, 2026, Google said it would let people turn off the visible watermark on media made with the Gemini app. Josh Woodward, the VP who runs the Gemini app, framed it as striking a balance between creative control and safety: users get to decide whether their creations show the visible mark, while Google keeps the invisible provenance signals in place. The control lives in the app's settings — a new Media watermark section with a simple on/off toggle — and Google said it is rolling out over the next few days.
The toggle covers Google's own generative models across media types: Nano Banana for images, Omni for video, and Lyria for music. Google said the change applies in the Gemini app and its Flow video editor, with support in Search coming later. The one thing the toggle does not touch is the invisible layer. Every image, video, and song Gemini generates still carries an invisible SynthID watermark and C2PA provenance metadata embedded in the file, on or off — so even with the corner mark hidden, the content remains machine-detectable as AI-generated.
There are exceptions, and they matter. According to reporting on the rollout, the option will not appear in countries whose laws require visible AI labeling. In India, South Korea, and Vietnam, the ability to hide the visible mark is limited to paid Google AI Ultra subscribers; everyone else in those markets keeps the visible watermark automatically. Work and school (Google Workspace) accounts do not get the setting at all — the visible watermark stays mandatory there. Google has said removing the visible Nano Banana watermark was among the most-requested changes after the Gemini team asked users for their biggest frustrations, which is the backdrop to shipping it now. Treat plan names, market lists, and rollout timing as a launch-day snapshot and confirm against Google's own settings and help pages.
The useful way to read this is: Google just removed the one cosmetic reason you were re-editing Gemini images before posting them. A clean Nano Banana plate, an Omni clip, or a Lyria track is now export-ready — as a raw asset. It still is not a post, and that is exactly the gap [Kompozy](/) fills. Drop a watermark-free Gemini image into Kompozy as source media and it becomes a brand-exact [Carousel Post](/glossary/output-buckets) or quote card composited through [HyperFrames](/glossary/hyperframes), a scene inside a [Persona Frames](/glossary/persona-frames) video layer, or the still behind a captioned short — reframed to 9:16, 1:1, and 16:9 and pushed across the eight social platforms plus blog and email on [Autopilot](/glossary/autopilot). Kompozy is a full generation-and-publishing engine, so the clean Gemini plate is one input among the [18 formats](/glossary/output-buckets) it produces and schedules, not the finished product.
The provenance angle is where a disciplined creator wins. Because SynthID and C2PA ride along whether the corner mark shows or not, the honest move is not to hide the AI — it is to disclose it cleanly and keep the work on-brand enough that reach never depends on the tell. Kompozy's per-post review gate is where you bake a consistent AI-disclosure line into everything before it ships, and its [Persona Brief](/glossary/persona-brief) keeps voice and look coherent across the whole set. Google gave you a cleaner asset; Kompozy is what turns that asset into a week of disclosed, on-brand, published content.
On August 14, 2026, Google added a Media watermark toggle in the Gemini app's settings that lets you hide the visible corner watermark on generative media — Nano Banana images, Omni video, and Lyria music — in the Gemini app and the Flow video editor, with Search support coming later.
No. The toggle only hides the visible mark. Google keeps an invisible SynthID watermark and C2PA provenance metadata embedded in every generated file whether the visible watermark is on or off, so the content stays detectable as AI-generated.
Per reporting on the rollout, the option is unavailable in countries whose laws require visible AI labeling, limited to paid Google AI Ultra subscribers in India, South Korea, and Vietnam, and not offered on Google Workspace (work or school) accounts, where the visible watermark stays mandatory.
Bring the watermark-free asset into a content engine like Kompozy. Kompozy composites it into a brand-exact carousel, quote card, or video layer via HyperFrames, captions and reframes it per platform, and schedules and publishes it across the eight social platforms plus blog and email — with a review gate where you add a consistent AI-disclosure line.