Google Allows Removal of Visible AI Watermarks

Josh Woodward, Google's Vice President for Gemini, announced on X that users can turn off visible watermarks by navigating to Settings > Media Watermark. The toggle currently applies to generations from Nano Banana, Omni, and Lyria AI models across Gemini and the Flow video editing application, with Google Search integration arriving shortly. To assist external software creators, Google is also open-sourcing Credentio, a library that allows third-party developers to embed local validation systems inside their own software applications.

Visible overlays frequently render synthetic content impractical for professional design and commercial production. However, maintaining provenance tracking remains critical for security. Google resolves this tension by leaving invisible SynthID watermarks—stealthy data markers embedded directly into files—and C2PA metadata intact. C2PA refers to open industry standards that certify the origin of digital content. Consequently, platforms and users can still query Google Search or Gemini to confirm if a file was artificially generated.

Google's updated watermarking policy offers a practical compromise for digital creators. By keeping hidden verification tech operational while removing visual distractions, the company aligns creative flexibility with responsible media attribution.