You can now turn off Google Gemini’s visible watermarks
Google has made its visible “sparkle” watermark optional for AI-generated images, videos and music created through Gemini and its Flow video tool. The change may make AI content harder to identify at a glance, although Google says underlying provenance markers will remain available for verification.
Users can disable the new “Media watermark” setting in Gemini, removing the bottom-right symbol from output made with Google’s Nano Banana and Omni models. Content will still carry invisible SynthID watermarks and C2PA metadata, which Gemini or Google Search can help detect; Google plans to extend the option to Search, except in countries that require visible watermarking.
In short
- Google now lets users hide Gemini’s visible AI watermark.
- Invisible SynthID and C2PA markers will remain embedded.
- The feature will not launch where visible labels are legally required.
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Users should retain control over their content's appearance, particularly as AI tools become mainstream and legitimate. The invisible technical markers—SynthID watermarks and C2PA metadata—preserve provenance information for verification, meaning transparency mechanisms remain intact rather than eliminated. For professional creators using AI responsibly, the option to remove visual branding allows cleaner presentation without sacrificing accountability for those who genuinely need to verify authenticity.
The case against
Visible watermarks serve a critical public interest function by immediately signalling AI-generated content to all viewers, regardless of technical literacy or effort. Most people will not actively check for invisible metadata, making watermark removal practically equivalent to removing transparency altogether. This change facilitates potential deception and misinformation by allowing AI content to appear indistinguishable from human-created material at first glance. Responsible AI adoption requires clear, visible disclosure that ordinary people can understand without specialised verification tools.
Explained simply
Google’s Gemini is an AI service that can create images, videos and music from instructions. Its Flow tool is designed for making videos, while Nano Banana and Omni are names for Google’s AI models.
AI-generated media can be difficult to distinguish from work made by people. Google uses visible symbols and less noticeable digital markers to indicate that media was made with AI; the latter can help verify its origin even when no symbol is shown.
These markers matter as AI content spreads online, where people may want to know how something was made. Rules on labelling AI-generated media differ between countries, and some require a visible notice.
AI Entertainment Music Software Technology