Google is gradually adding a way to turn off visible watermarks on some AI images, video, and music, while provenance signals such as SynthID and C2PA remain.
What changed
The option is rolling out to Gemini and Flow, with Google Search support planned later. Timing and eligibility vary by product and region, so it is not yet universal.
Visible watermarks and invisible provenance signals are separate. The user-facing toggle controls the on-screen mark, while SynthID and C2PA Content Credentials preserve or verify provenance.
What it means for creators
A visible mark can cover composition, text, or a product screen. Giving creators a display choice reduces visual interference, but it does not grant copyright protection or solve licensing questions.
Turning off the visible mark should not be read as deleting AI provenance. It separates a visual publishing choice from credentials that tools can read in the file.
What Credentio is for
Google also introduced Credentio, an open-source C++ library for C2PA Content Credentials. It is designed to validate content locally without sending media to the cloud.
Developers can integrate it into editors, media tools, and publishing workflows. It does not decide whether something infringes rights or prove who the true author is by itself.
What to watch next
Google’s feature is still rolling out, and eligibility and interfaces may vary. Turning off a visible mark does not mean all AI signals disappear, nor does a credential settle copyright or authorship.
