Google is giving Gemini users the ability to hide the visible sparkle watermark on AI-generated content, though invisible safety markers will always remain embedded in the files. The new Media Watermark toggle is rolling out over the next few days across Gemini, Flow, and eventually Google Search, letting creators choose whether to display the telltale sparkle on images from Nano Banana, videos from Omni, and songs from Lyria.

What Exactly Is Changing With Gemini's Watermark Settings?

The Media Watermark control is straightforward to access. Users can open Settings on gemini.google.com or in the mobile app, tap Media Watermark, and toggle "Show watermark" on or off. The choice applies only to future AI-generated content and only within Gemini apps themselves; it does not strip watermarks from other Google products.

However, turning off the visible sparkle does not make the file anonymous or undetectable. Google is maintaining a dual-layer approach to content authenticity. Invisible SynthID watermarks stay embedded in the pixels or audio, and C2PA Content Credentials remain in the file metadata. Users can still ask Gemini directly whether a piece of content was AI-generated, and the forensic markers will persist regardless of the visual toggle.

"The company is striking a balance between creative control and safety," explained Josh Woodward, Google's VP for Gemini.

Josh Woodward, VP for Gemini at Google

Who Can Actually Use the Watermark Toggle?

The rollout is not universal. Google has imposed regional and account-type restrictions on the feature. Users in India, South Korea, or Vietnam can only access the watermark toggle if they have an AI Ultra plan; otherwise, the visible sparkle remains mandatory. Additionally, work and school accounts never receive the toggle option, meaning those users will always see visible watermarks regardless of their preference settings.

This tiered approach reflects Google's attempt to balance user control with regulatory and safety requirements in different markets. The company is also taking steps to ensure the broader ecosystem can verify AI content authenticity. Google is open-sourcing Credentio, a C++ library for reading C2PA credentials, so other applications and platforms can access the same provenance information.

How to Access and Manage Your Gemini Watermark Settings

• Check Your Settings: Open gemini.google.com or the Gemini mobile app and navigate to Settings to find the Media Watermark control. If the option is missing, you may be in an early rollout phase, on a work account, or in a region where visible watermarks are still required.
• Understand the Limits: The toggle only affects future AI-generated content within Gemini apps. It does not remove watermarks from images, videos, or songs created in other Google products, and it does not strip invisible forensic markers from any file.
• Verify Content Authenticity: Even with the visible sparkle hidden, you can ask Gemini whether a piece of content was AI-generated. The invisible SynthID and C2PA metadata will still be present in the file, allowing for verification by compatible tools and platforms.

Why This Matters in the Broader AI Watermarking Debate

Google's approach sits at an interesting intersection of the ongoing debate over AI content labeling. While some companies like Suno are embedding audio watermarks to help platforms detect AI-generated music, and Spotify is flagging AI persona artist profiles, Google is taking a middle path: allowing creators to hide the visible mark while keeping the forensic evidence intact.

This strategy reflects a broader tension in the AI industry between transparency, creative control, and safety. Visible watermarks serve as a clear signal to viewers that content is AI-generated, which can help prevent misinformation. Invisible watermarks, by contrast, allow creators more aesthetic freedom while preserving the ability to verify authenticity through technical means. Google's dual-layer approach attempts to satisfy both concerns, though it may not fully resolve the underlying debate about how much AI content labeling should be mandatory versus optional.

For now, users should check their Settings to see if the Media Watermark toggle is available in their region and account type. The rollout is expected to complete over the next few days, though the exact timeline may vary depending on location and account status.