tech

Google Removes Visible AI Watermarks — But Still Tracks Every Pixel

The 'sparkle' icon on AI-generated media can now be toggled off, a major concession to creators demanding cleaner outputs. Yet Google's invisible SynthID technology ensures no generated content will ever truly be anonymous, a classic case of giving with one hand and holding tight with the other.

SignalEdge·August 15, 2026·3 min read
A digital artist scrutinizing the corner of a high-quality art print in a studio, symbolizing the removal of AI watermarks.

Key Takeaways

  • Google is adding a user-controlled toggle to remove visible watermarks from content generated by its AI tools.
  • The setting applies to tools like the Gemini chatbot and the Flow video generator, removing the 'sparkle' icon.
  • Despite the visible change, Google's invisible SynthID watermark will remain embedded in all generated files for traceability.
  • The move aligns Google's tools with competitors that offer less intrusive watermarking, targeting professional creators.

Google will now allow users to remove the visible watermarks from images, videos, and audio created with its AI tools. The update, detailed by The Verge, introduces a new 'Media watermark' toggle in services like Gemini and the AI video generator, Flow. This allows creators to disable the 'sparkle' icon that previously appeared in the corner of all generated content, a move that appears to prioritize user freedom and professional use cases.

The concession, however, is purely cosmetic. While the visible mark is now optional, Google is not relinquishing its ability to identify AI-generated files.

The Appearance of Choice

For creators and designers, forced watermarks are a persistent annoyance. They can disrupt the aesthetic of an image or video and signal to a client that a piece of media was, at least in part, machine-generated. By making the visible watermark optional, Google makes its tools more competitive against rivals like Midjourney, which have long offered cleaner outputs for professional work. According to The Verge, the new toggle will control the 'sparkle' watermark that appears in the bottom-right corner of generated media, giving users a simple way to produce unblemished content.

This is a direct response to a clear market demand. As generative AI tools move from novelty to professional workflow integration, the expectation for seamless, unbranded output grows. Google's decision is a pragmatic one, aimed at capturing and retaining users who require a polished final product without a distracting digital stamp.

SynthID's Invisible Grip

The real story is not what Google is letting users remove, but what it is keeping. Both TechCrunch and Engadget confirm that while the visible watermark is now optional, the company’s underlying invisible watermark, SynthID, will remain embedded in all generated content. SynthID works by making imperceptible changes to the pixels of an image or the data of an audio file, creating a signature that is robust enough to survive compression, cropping, and other edits.

This creates a two-tiered system for identification. The public gets cleaner images, free of distracting icons, which blurs the lines between human and machine creation for the casual observer. Meanwhile, Google and anyone with the right tools retain the ability to forensically prove a file’s origin. This suggests the industry is coalescing around a standard where overt disclosure is optional, but covert traceability is mandatory. It is a strategic compromise, allowing Google to court creators while maintaining a powerful mechanism for combating misinformation and tracking content provenance behind the scenes. The company gets a public relations win for flexibility without ceding any meaningful control.

Sources & References

Daily Newsletter

Stay ahead of the curve

Get the most important stories in tech, business, and finance delivered to your inbox every morning.

You might also like