Google rolled out a new option on Friday that lets users hide the visible watermark that appears on AI‑generated content. The toggle works for images, video clips and audio tracks produced by the company’s Gemini family of models, including Nano Banana, Omni and Lyria.
While the visible mark can be turned off, the company said the invisible SynthID watermark and the C2PA‑standard metadata will remain embedded in every file. Those hidden identifiers let Google and third‑party tools verify whether a piece of media was created by artificial intelligence.
"We’re striking a balance here between creative control and safety," said Josh Woodward, Google’s vice president for Gemini, in a post on X. "The visible watermarks are now optional, but invisible SynthID watermarks and C2PA metadata are still being used for transparency. So you can still use Gemini or Search to see if an image was AI‑generated."
Google said the new setting will appear in the Gemini interface and in its video‑editing app Flow. Support for the option inside Search is slated to follow shortly. Users will find the control under Settings → Media Watermark once the rollout is complete, which the company expects to happen in the coming days.
The move reflects an evolving approach to labeling AI‑generated media. Visible watermarks have drawn criticism from creators who say they hinder professional use, yet regulators and advocacy groups continue to push for clear disclosure of synthetic content.
In parallel with the toggle, Google announced it will open‑source a library called Credentio. Developers can embed Credentio in their applications to perform local validation of the invisible watermarks, giving them a way to confirm authenticity without relying on external services.
Google’s decision arrives shortly after Anthropic introduced a mandatory watermark on text and files generated by its Claude model to comply with European Union rules. By keeping the invisible marks while making the visible ones optional, Google aims to give creators flexibility without abandoning the traceability that regulators and platforms increasingly demand.
Industry observers note that the shift could make Google’s generative tools more appealing to advertisers, filmmakers and musicians who need clean assets for commercial projects. At the same time, the continued presence of hidden identifiers means that the company can still flag AI‑generated content when required by policy or law.
Google has not disclosed a specific timeline for the Search integration, but the rollout schedule suggests the feature will be widely available before the end of August.
Questo articolo è stato scritto con l'assistenza dell'IA.
News Factory APP - notizie agentiche per potenziare il tuo SEO e AEO.