Google will now allow users to remove visible watermark from its AI generations

Google has officially announced a shift in how it handles AI-generated media, confirming that users will soon have the ability to toggle off visible watermarks on content created by its models. This update applies to a wide range of media, including AI-generated images, songs, and videos.

Balancing Creative Freedom and Transparency

While the visible branding is becoming optional, the tech giant is doubling down on its commitment to platform safety. The company clarified that disabling the visible mark will have no impact on the underlying SynthID invisible watermarking or the C2PA metadata standards. These hidden identifiers ensure that the origin of the media remains verifiable by automated systems.

Josh Woodward, Google’s VP for Gemini, addressed the update on X, noting that the new toggle will be integrated into the Nano Banana, Omni, and Lyria models.

“We’re striking a balance here between creative control and safety: while the visible watermarks are now optional, invisible SynthID watermarks and C2PA metadata are still being used for transparency. So you can still use Gemini or Search to see if an image was AI-generated,” Woodward stated.

Implementation and Developer Tools

The feature is scheduled to roll out over the next few days. Once live, users can manage their preferences by navigating to Settings > Media Watermark. The setting will be available within the Gemini interface and the Flow video editor, with integration for Google Search expected to follow shortly.

Key Highlights of the Update:

• Scope: Applies to images, videos, and audio tracks.
• Availability: Accessible via Settings for Gemini and Flow users.
• Security: Invisible SynthID and C2PA standards remain active to ensure AI detection.
• Developer Support: Google is open-sourcing Credentio, a new library designed to help developers implement local validation mechanisms within their own applications.

This strategic pivot follows recent industry trends, such as Anthropic’s decision to implement watermarking in Claude to align with European Union regulations, highlighting the ongoing tension between creative utility and the necessity of AI identification.