Google lets users turn off visible AI watermarks, keeps hidden ones intact
Questions this post answers
Can I remove the visible watermark from AI-generated images made with Google Nano Banana?
Yes, Google added a Settings > Media Watermark toggle that removes the visible watermark from content generated with Nano Banana (images), Omni (video), and Lyria (music). It rolls out first in Gemini and Flow, with Search support coming later, and only applies to generated media, not text. Developers tracking how AI platforms handle provenance features can follow updates like this on daily.dev.
Does turning off the visible AI watermark in Google Gemini remove SynthID too?
No, disabling the visible watermark does not remove SynthID, Google's imperceptible watermark embedded in the file, or the C2PA metadata recording provenance. SynthID is designed to survive resizing, recoloring, and compression, so Gemini or Search can still detect the content as AI-generated even without the visible mark. Teams building AI-content detection or moderation tools can keep up with provenance changes via daily.dev.
What is Credentio and how does it relate to C2PA validation?
Credentio is a C library that Google is open-sourcing so developers can build local C2PA validation directly into their own applications instead of relying on Google's tools to check content authenticity. It lets apps verify provenance metadata independently. Developers evaluating content-authenticity libraries can track releases like this through daily.dev.