Google Makes Visible Watermarks Optional for AI-Generated Media

As AI-generated content becomes more prevalent, Google has introduced a significant update to its watermarking approach. The company now allows users to decide whether visible watermarks appear on AI-created images, videos, and audio tracks.

How the Watermark Control Works

The new feature operates through a straightforward toggle. According to Google Gemini vice president Josh Woodward, users will find a "media watermark" setting within the Gemini platform and the Flow video editor, giving them direct control over visible watermark display.

This control applies to content produced by several of Google's AI models, including their latest media generation systems. Support for watermark management in search results is expected to follow soon.

What Protection Layers Remain Active?

It's important to note that this change only affects human-visible watermarks. Google specifically highlighted two core protection systems that remain unchanged:

  • SynthID Invisible Watermarks: Digital fingerprints embedded in pixels or audio, designed for machine detection
  • C2PA Metadata Standards: Industry specifications that track content origin and edit history

These invisible verification layers continue operating in the background, supporting content provenance and authenticity checks.

New Developer Tools for Verification

To support broader adoption of content verification technology, Google has open-sourced a software library called Credentio. This toolkit enables third-party developers to integrate local content verification capabilities directly into their applications, without relying on cloud services.

Industry observers see this move as an important step toward making content authentication more accessible, particularly for platforms handling large volumes of user-generated media.

The feature rollout is expected to complete within the coming days. Google stated the adjustment aims to balance user experience with content transparency while preserving core authentication mechanisms.