Google Allows Disabling Watermarks in Gemini: What Changes for Users and How AI Content Verification Now Works

On August 17, 2026, Google announced the launch of a new feature in its Gemini artificial intelligence ecosystem. Users can now disable visible watermarks on images, videos, and music tracks created using AI models. This decision affects content generation via the Gemini and Flow platforms and extends to the Nano Banana, Omni, and Lyria models.

How the new setting works

The toggle for managing visible labeling is located in the settings section: Settings > Media watermark. Users can decide for themselves whether to display the Gemini logo on generated content. However, it is important to note that in countries where legislation requires mandatory visible labeling of AI content, this option will be unavailable. This means Google's decision accounts for regional legal norms and does not violate local transparency requirements for digital content.

Hidden identification mechanisms remain active

Removing the visible watermark does not mean abandoning the identification of AI content. Invisible SynthID watermarks and C2PA metadata continue to be saved in files. SynthID is Google's system for detecting content created or edited by its AI models. It operates at the pixel and audio data level, ensuring the technical ability to identify AI origin even in the absence of visual labeling.

C2PA Content Credentials contain signed information about the file's origin and its change history. These metadata allow tracking the chain of content creation and editing, which is particularly important for journalism, rights holders, and moderation platforms.

Contradictory data

Although Google claims that Gemini can check images, videos, and audio for the presence of SynthID, the absence of a detected watermark does not prove that the content was created without AI. This creates a certain uncertainty: users might assume that the lack of a visible sign means the content is "clean," whereas technically it could still be AI-generated. On the other hand, experts note that this policy aligns with real-world practice: many authors have already removed watermarks themselves, and now Google provides an official tool for this.

Consequences for creators and platforms

For creators, disabling the logo allows for cleaner visual material without an element that might interfere with the final composition or professional layout. This is particularly relevant for designers, videographers, and musicians using AI as a creative tool. However, for content verification systems, this means greater reliance on detection tools and metadata. Platforms such as social networks or news aggregators will have to implement support for SynthID and C2PA to effectively identify AI content.

Implementation plan and future steps

The feature will be activated gradually over several days. The ability to manage visible labeling in Google Search is planned to be added later. This indicates that Google views the issue of AI content labeling as a multi-level system where visual and technical identification will develop in parallel. In the future, integration of these mechanisms into other company products, including YouTube, Google Photos, and Android, can be expected.