Back to stories
product launchGenerated by an AI editor from the reporting and web sources listed on this page.

Google Lets Users Remove Visible Watermarks From AI-Generated Content

A new toggle in Gemini and Flow removes the sparkle watermark from images, videos, and music while preserving invisible provenance signals.

Published Updated The total reporting and web sources attached to this story.The AI editor’s assessment of how strongly the attached sources’ quality and agreement support this article.

What matters

  • Google now lets users disable the visible sparkle watermark on AI-generated images, videos, and music via a new Media watermark toggle in Gemini and Flow.
  • Disabling the visible marker does not remove the invisible benchmarks Google embeds to identify a file as AI-generated.
  • The change gives creators cleaner output while shifting provenance responsibility toward automated invisible-signal detection.
  • Platforms that relied on visible watermarks for manual content review will need automated detection to maintain provenance coverage.

Launch facts

Platforms:
Gemini, Flow

What happened

Google has introduced a new "Media watermark" toggle in Gemini and its AI video generator, Flow, that lets users turn off the visible watermark on AI-generated content. When disabled, Google removes the "sparkle" watermark that normally appears in the bottom-right corner of generated images, videos, and music. The setting gives creators cleaner-looking output for end use without requiring post-processing to strip the marker.

Crucially, turning off the visible watermark does not affect the invisible signals Google embeds in generated files. Those invisible benchmarks remain intact and can still be used to identify a file as AI-generated, preserving a machine-readable provenance layer even when the human-visible marker is gone.

Why it matters

The change reflects a tension at the heart of AI content provenance: visible watermarks are useful for transparency but can be aesthetically undesirable for creators who want polished output. By making the visible marker optional while keeping invisible signals in place, Google is betting that automated detection can carry the provenance burden without forcing a visible badge on every generated asset.

For platforms that rely on visible markers for manual content review, the update creates a gap. Staff who scan for watermarks to flag synthetic media will no longer be able to count on that cue. The invisible watermark becomes the primary automated mechanism for identifying Google-generated files, but only platforms that implement detection tooling will benefit from it.

The move also raises questions about how broadly invisible watermark detection is adopted across the ecosystem. If major social and publishing platforms cannot read Google's invisible signals, the practical effect of removing visible watermarks is that AI-generated content becomes harder to distinguish from human-created work for most viewers.

What to watch

Watch whether major platforms—social networks, news organizations, stock-image marketplaces—add or expand support for detecting Google's invisible watermark signals. Adoption there will determine whether the shift from visible to invisible provenance maintains meaningful transparency or simply removes the most accessible indicator of synthetic content.

Also watch whether other AI generation tools follow Google's lead in making visible watermarks optional, which could accelerate the need for cross-platform invisible-watermark standards.

What to do next

Developers

Check whether your app or platform can detect Google's invisible watermark signal and plan integration if provenance matters for your use case.

With visible watermarks now optional, invisible signals become the primary automated way to identify AI-generated files.

Founders

Evaluate whether your product should surface invisible watermark data to users as a trust or safety feature.

As visible markers disappear, platforms that detect and disclose AI-generated content can differentiate on trust.

PMs

Assess the impact on content moderation and labeling workflows now that visible watermarks may be absent from Google-generated media.

Teams relying on visible markers for manual review will need automated detection to maintain provenance coverage.

Investors

Monitor adoption of invisible watermark detection across major platforms as a signal for the broader AI-provenance market.

Google's move shifts provenance from visible to invisible signals, creating opportunity for detection and verification tooling.

Operators

Update internal content guidelines to account for AI-generated media that may no longer carry a visible watermark.

Staff who manually review or approve visual assets can no longer rely on visible markers to flag synthetic content.

How to test

  1. 1Open Gemini or Flow and navigate to settings.
  2. 2Locate the "Media watermark" toggle.
  3. 3Toggle the setting off.
  4. 4Generate an image, video, or music clip and download the output.
  5. 5Inspect the generated file for the absence of the sparkle watermark in the bottom-right corner.
  6. 6If possible, run the file through a tool that detects Google's invisible watermark signal to confirm it is still present.

Caveats

  • The sources do not specify which exact Gemini tiers or Flow access levels include the Media watermark toggle.
  • Detection of the invisible watermark may require specialized tooling not publicly detailed in the sources.