Google Gemini lets users turn off visible AI watermarks
Google has added a 'Media watermark' toggle to Gemini and its AI video tool Flow, letting users remove the visible 'sparkle' mark from AI-generated images, videos, and music while invisible SynthID and C2PA markers remain embedded.
👁 117 reads
Google has added an optional toggle to Gemini and its AI video generator Flow that removes the visible watermark from AI-generated images, videos, and music, the company announced on August 14, 2026. The setting, called “Media watermark,” is available in Gemini’s settings menu and, when switched off, eliminates the “sparkle” mark that previously appeared in the bottom-right corner of content produced by Google’s Nano Banana and Omni models.
- A new 'Media watermark' setting in Gemini and Flow lets users switch off the visible sparkle watermark on AI-generated content
- Invisible SynthID watermarks and C2PA metadata will still be embedded regardless of the toggle setting
- The toggle will not roll out in countries that legally require visible watermarks
Invisible protections remain in place regardless of how the toggle is set. According to Josh Woodward, vice president of Google Labs, Gemini, and AI Studio, AI-generated content will still carry invisible SynthID watermarks and C2PA metadata embedded in the background, meaning Gemini and Search can still identify whether a piece of content is AI-generated when asked.
Google framed the change partly as an acknowledgment of existing user behaviour: many people were already editing out visible watermarks manually. The company also pointed to competitive parity - OpenAI and Meta do not apply visible watermarks to their generated images either. OpenAI relies on SynthID and C2PA, while Meta has introduced its own Content Seal standard. In a separate development this week, Anthropic said it is applying invisible watermarks to both AI-generated text and images.
Google plans to extend the new setting to Search, though the toggle will not be available in countries where visible watermarks are legally required under local regulation.
Practitioner takeaway: For creators and developers distributing AI-generated media, the toggle means cleaner output without losing the provenance trail - SynthID and C2PA metadata still travel with every file. Content can still be verified as AI-generated through Gemini or Search, so the audit path survives even when the visual mark is gone. Anyone operating in jurisdictions with mandatory AI-labelling rules should confirm local requirements before switching the setting off.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →
More on this from QUANTSLANT
magazine·OpenAI Replaces CRO With Former Wiz Executive Dali RajicOpenAI has replaced chief revenue officer Denise Dresser after nine months, hiring Wiz president and COO Dali Rajic as the company continues a broad executive shake-up.