Gemini Media Watermark setting: visible off, SynthID stays

In Gemini Apps, the Media Watermark setting controls only the visible watermark. Turning it off does not remove the invisible SynthID watermark or the Content Credentials (C2PA) metadata, which Google's help page says stay on for media made or edited with Gemini Apps. In India, South Korea and Vietnam the setting appears only with an AI Ultra subscription.

Gemini facts are from Gemini Apps Help, read 2026-10-01. Sume facts are from the image and video docs.

What does the Media Watermark setting change?

On gemini.google.com, the path is Settings, then Media Watermark. The page says it applies to visual media made in Gemini Apps, and to Gemini Apps only (web and mobile), not to AI features in other Google products.

Who can turn the visible watermark off?

Users in India, South Korea or Vietnam see the setting only with an AI Ultra subscription. Without it, visible watermarks are applied automatically to all visual media made in Gemini Apps. The page does not list other regional conditions.

What does a Sume route return instead?

Sume's docs do not describe any of these three layers for its outputs, so do not infer them. With model: "sume/auto" the docs say Sume does not disclose which family served a video request, and that the image router never discloses which one ran, so no Gemini setting can be assumed to apply. Inspect the file you receive. For a related walkthrough see Gemini Omni video watermark and SynthID.

What should I do if I need a visible mark of my own?

Add it yourself as an overlay. Timeline compose has an overlay layout where the plate keeps its aspect and width_ratio runs 0.05 to 1 of the width. That is your own visible label, separate from any invisible signal; see add a logo or watermark to video. Check each platform's disclosure rules from the platform itself; this post is not legal advice.