Google is making visible watermarks optional for AI-generated images, videos and audio, giving users more control over how their generated media appears while keeping its underlying AI-identification systems intact.
Google VP for Gemini Josh Woodward announced the change on Friday, August 14. The option applies to content generated using Nano Banana, Omni and Lyria, which cover Google’s image, video and music-generation tools.
Visible Watermarks Can Be Turned Off
Once the feature reaches their accounts, users will be able to open Settings > Media Watermark and choose whether Google’s visible AI mark appears on generated content.
The setting is rolling out to Gemini and Flow, Google’s AI video creation and editing platform. Google plans to add the same control to Search later. The rollout is taking place over the coming days.
Google says the change is intended to give creators more flexibility, particularly when visible branding makes generated material less suitable for professional or creative projects.
Invisible AI Tracking Remains
Turning off the visible watermark does not remove Google’s less obvious methods for identifying AI-generated content.
Generated media will continue to contain SynthID, Google’s imperceptible watermarking technology. Google will also continue using C2PA Content Credentials, which provide machine-readable information about how digital media was created or modified.
This means users may be able to remove the watermark that appears directly on an image or video while the content can still carry information indicating that Google’s AI tools were involved.
Google also allows users to check supported media through Gemini to determine whether SynthID is present, with similar verification capabilities expanding across Search and other Google products.
Google Open Sources Credentio
Alongside its wider push for AI transparency, Google has also open-sourced Credentio, a C++ library for validating C2PA Content Credentials.
Credentio performs verification locally rather than requiring media to be uploaded to external servers. Google says the underlying technology already powers C2PA functionality across nearly 40 of its products and has been used at a scale of tens of billions of generated assets.
The change comes as AI companies continue experimenting with different ways to identify synthetic content while balancing transparency with users’ ability to use generated material without prominent branding.
The post Google Will Let You Remove Gemini’s Watermark from Photos and Videos appeared first on ProPakistani.