Google Makes Its Visible AI Watermark Optional - SynthID Stays On Either Way
Users can now turn off the visible mark on Gemini-generated images, videos, and songs. The invisible SynthID layer and C2PA metadata are unaffected.
Google's Gemini watermark is now optional. Starting August 14, users can turn off the visible mark on AI-generated images, videos, and songs - a setting that covers Nano Banana, Omni, and Lyria music generations. Josh Woodward, Google's VP for Gemini, posted the news on X.
Nothing invisible changes. Removing the Google Gemini watermark only affects the cosmetic mark - SynthID and C2PA metadata stay on regardless, and those are what AI detection tools and Gemini Search actually use to flag generated content. A visible watermark was always the cosmetic layer; detection never depended on users seeing it. Woodward put it plainly: "You can still use Gemini or Search to see if an image was AI-generated."
- Settings → Media Watermark → toggle visible watermark off
Gemini and Google Flow get the toggle first, with Search support coming soon. Nano Banana, Google's AI image generator from Made by Google 2026, Omni, and Lyria are the three covered models. Google also open-sourced Credentio alongside this, a C2PA validation library developers can embed in their own apps for local AI content detection. The feature rolls out over the coming days.