Editorial illustration for Google Lets Users Disable Visible Watermarks on AI Images
Google Lets Users Remove AI Image Watermarks
Google Lets Users Disable Visible Watermarks on AI Images
Google is giving users a way to strip the little "sparkle" icon off AI-generated images, videos and music. The company added a new "Media watermark" toggle to Gemini and Flow, its AI video generator, that lets people turn off the visible marker stamped in the bottom-right corner of anything made with Nano Banana or Omni. Until now, that watermark showed up automatically, a small but consistent flag that told viewers they were looking at machine-made content.
The change lands at a moment when visible watermarks are becoming the exception rather than the norm. OpenAI and Meta already skip them on their image generators, leaning instead on background metadata to do the disclosure work. Google's move follows that pattern, though it's not dropping disclosure entirely. Josh Woodward, vice president of Google Labs, Gemini and AI Studio, addressed what the company is keeping in place behind the scenes even as the visible marker becomes optional.
Though visible watermarks are now optional, AI-generated content will have invisible SynthID watermarks and C2PA metadata embedded in the background, according to Josh Woodward, the vice president of Google Labs, Gemini, and AI Studio. That means you can still verify if an image has one of these markers by asking Gemini or Search if it’s AI-generated.
Why this matters
For anyone building on top of Gemini or Flow, this is a small policy change with outsized implications for trust. Google is betting that SynthID and C2PA metadata can do the provenance work that a visible sparkle icon used to signal at a glance. That's a reasonable technical bet, but it shifts the burden of verification onto platforms, journalists, and ordinary users who won't know to check for invisible markers unless someone builds tooling to surface them.
Developers integrating Nano Banana or Omni into products should treat this as a reminder to expose SynthID checks in their own interfaces rather than assume end users will dig for them. Founders shipping AI content tools now have precedent to make visible watermarking optional, which will likely spread pressure across competitors like OpenAI and Midjourney to do the same. Researchers tracking misinformation should watch whether removing the visible cue increases the volume of AI content circulating without obvious disclosure, even if it's technically traceable.
The invisible layer is still there, but "traceable in theory" and "flagged in practice" are two different things.
Related Reading
Browse all Google news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic NVIDIA and Google Cloud let developers scale AI from prototype to production
Market Trends
Related Topic Google's FACTS benchmark shows 70% factuality ceiling across four tests
Research & Benchmarks
From Archives Claude Code now runs daily maintenance on Anthropic's software
From Archives Liquid AI's 3B Vision Model Shows Major Gains in Screen Reading, Object Grounding
From Archives Google's Gemini Flash 3.7 Cuts Price 50%, Boosts Coding Performance
From Archives Google's Gemini Slumps to 1.9% Market Share in July
Common Questions Answered
What visible watermark did Google previously use on AI-generated images in Gemini and Flow?
Google previously used a small "sparkle" icon watermark that appeared automatically in the bottom-right corner of all AI-generated images, videos, and music created with tools like Nano Banana or Omni. This visible marker was designed to clearly signal to viewers that the content was machine-made rather than created by humans.
Can users now completely remove all watermarks from Google AI-generated content?
While users can now disable the visible "sparkle" watermark through the new "Media watermark" toggle in Gemini and Flow, Google has not removed watermarking entirely. Instead, invisible SynthID watermarks and C2PA metadata are embedded in the background of all AI-generated content, allowing verification through Gemini or Search even when the visible watermark is turned off.
How does Google's invisible watermarking system work compared to the previous visible sparkle icon?
Google's invisible SynthID watermarks and C2PA metadata are embedded in the background of AI-generated content, shifting verification responsibility from a glance-visible marker to platforms, journalists, and users who must actively check using tools like Gemini or Search. This technical approach assumes that invisible markers can effectively prove provenance, but requires users to know to check for these hidden identifiers rather than relying on an obvious visual indicator.
What are the implications of making visible watermarks optional for content verification?
Making visible watermarks optional places a greater burden on platforms, journalists, and ordinary users to verify AI-generated content by checking for invisible markers, since most people won't know to look for SynthID or C2PA metadata without specialized tooling. This policy change represents a significant shift in how AI content authenticity is communicated, moving from transparent visual signals to hidden technical verification methods.
Further Reading
- Google will now allow users to remove visible watermark from its AI generations - TechCrunch
- Google's Gemini app adds a toggle to disable AI watermarks, with some exceptions - Digital Trends
- Google Gemini Finally Lets You Disable Visible AI Watermarks from AI Images and Videos - Android Headlines
- Manage watermark settings in Gemini Apps - Google Help
