Google is giving users the option to remove visible watermarks from AI-generated images, videos and music, changing how content created through some of its generative AI models will appear when published or used professionally.
The company is introducing a Media Watermark setting that will allow users to switch the visible marker on or off. The change applies to content generated using Google's Nano Banana, Omni and Lyria models and will be available through Gemini and its AI filmmaking tool Flow. Support for Google Search is expected to follow.
The update, announced by Josh Woodward, Vice President of Google Gemini, is rolling out over the coming days. Once available, users will be able to access the option through Settings and select Media Watermark to control whether the visible identifier appears on generated content.
Importantly, Google is not removing its underlying mechanisms for identifying AI-generated media. Content will continue to carry SynthID, Google's invisible watermarking technology, as well as C2PA-related metadata used to provide information about the origin and history of digital content.
SynthID embeds an imperceptible signal directly into AI-generated content. Google developed the technology to help identify material produced using generative AI while keeping the watermark largely invisible during normal viewing or listening. It has been designed for different formats, including images, audio, video and text.
The distinction means removing the visible watermark does not remove the underlying AI provenance signals. Users can still use supported Google services, including Gemini, to check whether an image was generated using Google's AI systems. Search is also expected to support this verification capability.
The move addresses a practical issue surrounding visible AI labels. While prominent watermarks provide an immediate indication that media has been generated by AI, they can also affect how the content looks when incorporated into advertising, design, social media, presentations and other professional creative work.
For marketers and creative teams, the change could make Google-generated assets easier to integrate into existing workflows without a visible AI symbol becoming part of the final creative. However, the continued use of SynthID and provenance metadata means Google is retaining a technical identification layer even when the visual marker is disabled.
The approach reflects an ongoing debate across the generative AI industry about how synthetic media should be identified as AI-generated tools become increasingly capable of producing realistic images, video and audio.
Other technology companies have also been developing approaches based on metadata, provenance standards and invisible identification systems. The challenge is balancing transparency around AI-generated material with the practical requirements of creators and businesses using generative tools in production environments.
Google is also open-sourcing Credentio, a new library intended to allow developers to embed local validation mechanisms into their applications.
The watermark update comes as technology companies face increasing scrutiny over the identification of synthetic content and the potential for AI-generated media to be presented without clear context.
Google's latest approach separates the visible branding of AI-generated media from its technical provenance. Users gain more control over how generated content appears, while the company continues to embed mechanisms intended to make its AI origins detectable.
For brands, agencies and content teams increasingly using generative AI, the change could reduce one visual barrier to deploying AI-generated assets while keeping questions around disclosure, provenance and responsible use firmly in focus.