Google lets users disable visible watermarks on AI media

Google has announced that users will soon be able to turn off visible watermarks on its AI-generated images, videos, and music. According to Josh Woodward, Google's vice president for Gemini, the new toggle will apply to content generated by the Nano Banana, Omni, and Lyria models. This setting will initially be accessible within Gemini and Google's video editing tool, Flow, with integration for Search slated to arrive in the near future.

To disable the watermark, users can navigate to Settings, then Media Watermark, once the feature rolls out in the coming days. While the visible overlay becomes optional, Google is maintaining its backend safety measures. The change will not affect invisible SynthID watermarks or C2PA standard-related metadata embedded in the files. Woodward noted that this approach strikes a balance between giving creators more control over their output and maintaining safety, as platforms like Gemini and Search can still detect if an image was AI-generated.

For creative professionals, this update resolves a common pain point where prominent visible watermarks make AI-generated assets difficult to use in commercial or polished portfolios. Alongside this update, Google is open-sourcing a new library called Credentio. This tool allows developers to integrate local validation mechanisms directly into their own applications, helping them verify AI-generated content programmatically.

This policy shift highlights the industry's ongoing struggle to balance transparency with user utility. It contrasts with recent moves by competitors like Anthropic, which recently introduced controversial watermarking for text and files generated by its Claude models to comply with European Union regulations. Google's dual-track approach of hiding the watermark while preserving metadata offers a potential compromise for the wider industry.