Google Gemini introduces optional visible AI watermarks

Google has introduced a new setting for its Gemini AI service that allows users to toggle visible watermarks on and off for generated images, videos, and music. This update applies to content created using models such as Nano Banana, Omni, and Lyria, as well as the video editing tool Flow.

While users can opt to remove the visible sparkle-style icon to provide a cleaner output for creative projects, Google emphasized that the content remains technically identifiable. The company will continue to embed invisible SynthID watermarks into pixels and audio tracks, alongside C2PA metadata, to ensure machine-readable provenance and transparency.

The rollout is being implemented across Gemini web and mobile platforms, with integration into Google Search expected in the future. However, the ability to disable visible watermarks may be restricted in jurisdictions where local laws mandate visible AI labeling. Additionally, users with work or school accounts currently do not have access to this customization.