Agentic market $10.8B and climbing  ·  editor@gaasnews.com
Sections
HomeWhat is GaaS?PlatformsPricingGlossaryOpinionAboutContact
HomePolicy & RegulationGemini Watermarks
Policy & Regulation

Google now lets users hide Gemini's visible AI watermarks

Google VP Josh Woodward announced a watermark toggle for Gemini and Flow on Friday, letting users strip the visible AI badge from images, video, and music. Invisible SynthID signals and C2PA metadata remain in every file.

AJ
Andrew Jamerson
Founding Editor
Aug 16, 2026 · 2 min read
The spark badge on Gemini media is now a setting, not a certainty. // GaaS News

Google spent the week moving the opposite direction from the rest of the provenance conversation. On Friday the company began rolling out a Media Watermark setting that lets users switch off the visible watermark on AI-generated media, TechCrunch reported. The toggle covers images made with Nano Banana, video from Omni, and songs from Lyria, and applies in the Gemini app and the Flow video editor, with Google Search support coming next.

Josh Woodward, Google's VP for Gemini, announced the change on X and framed it as fixing a papercut for creators who did not want a badge on their output. The invisible layer is not optional. SynthID, the signal Google embeds in every generated file, stays in place regardless of the setting, as does C2PA metadata. Woodward said the company is "striking a balance here between creative control and safety." The toggle also will not appear in countries where visible labels on AI media are required by law.

The timing is hard to miss. In the same week, Anthropic switched on invisible watermarks for everything Claude produces, a follow-through on its EU AI Act transparency commitments. Both companies now satisfy the regulatory baseline, which asks for machine-readable, detectable output rather than badges humans can see. Where they split is the human layer. Anthropic added marks users cannot remove. Google removed the only mark most viewers ever noticed.

For agentic AI services the practical effect is immediate. Marketing agents, video pipelines, and content operations built on Gemini can now ship media with no visible tell, while the proof of origin migrates entirely into signals that require tooling to read. That raises the stakes for verification infrastructure, and Google seems to know it. Alongside the toggle, the company open-sourced Credentio, a C++ library that lets developers embed local content-credential validation in their own applications, per TechCrunch.

The bet embedded in Friday's change is that provenance is a plumbing problem, not a labeling problem. That may be right in the long run. In the short run, the average viewer cannot run a SynthID check, and the visible badge was the one disclosure that worked at a glance. Platforms, newsrooms, and procurement teams that relied on it now need a detector in the loop. The disclosure did not disappear. It just stopped being free.

AJ

Andrew Jamerson

Founding Editor, GaaS News

Andrew Jamerson is the founding editor of GaaS News, covering the economics of the agent era. He started the publication to cover Agentic AI as a Service as a dedicated beat and edits every article on the site.

Be on the list when the beat breaks

One email when a platform ships, a round closes, or the ground shifts under the software stack.