Google is giving users more control over how its AI-generated media is labelled. The company is introducing an option to switch off the visible watermark that appears on content created using some of its generative AI models, while retaining less obvious systems designed to establish where that content came from.<\/p>\n
The change applies to AI-generated images, videos and songs, and is available through Gemini and Google\u2019s AI video creation tool, Flow. Search support is also planned.<\/p>\n
What is changing?<\/p>\n
Until now, AI-generated media from Google\u2019s tools could carry a visible \u201csparkle\u201d watermark, typically placed in the bottom-right corner. Google is now making that mark optional.<\/p>\n
Users will be able to go to Settings > Media Watermark and choose whether the visible label appears. The option applies to images created with Nano Banana, videos produced with Omni and songs generated using Lyria. Google says the feature is rolling out over the coming days.<\/p>\n
There are exceptions: in countries where the law requires a visible watermark, users will not be able to remove it.<\/p>\n
Why is Google making the watermark optional?<\/p>\n
Visible labels are useful for transparency, but they can also get in the way when AI-generated material is being used creatively or professionally. A prominent mark can interfere with the appearance of an image, video or other piece of media, particularly when the content is intended for presentation or further editing.<\/p>\n
Google is therefore separating the visible label from the underlying technology used to establish the provenance of AI-generated material.<\/p>\n
Even when users switch off the visible watermark, Google says the media will continue to carry SynthID, its invisible watermarking technology, along with C2PA Content Credentials in the metadata.<\/p>\n
That means removing the visible mark does not mean removing every indication that AI was involved.<\/p>\n
How can AI-generated media still be identified?<\/p>\n
Google says users will still be able to check whether content contains these markers through Gemini, with Search also set to support the feature.<\/p>\n
The distinction is important. A visible watermark is designed for people to notice immediately, while an invisible watermark and provenance metadata can be used by verification systems without changing the appearance of the media.<\/p>\n
This approach gives creators more flexibility while retaining a way to establish the origin of AI-generated content.<\/p>\n
Google introduces Credentio<\/p>\n
Alongside the watermark change, Google is open-sourcing Credentio, a C++ library designed to help developers work with C2PA Content Credentials.<\/p>\n
The library is intended for local validation, meaning applications can check the provenance of media without sending the files to an external server. Google says this can reduce bandwidth requirements and latency while keeping the media within the developer\u2019s own environment.<\/p>\n
Credentio is also designed to handle large files efficiently, including high-resolution images and video files running into multiple gigabytes. It can inspect manifests, assertions, digital signatures and other information contained within C2PA credentials, while allowing developers to use configurable trust lists.<\/p>\n
Google says the same underlying code has already been used across nearly 40 conformant C2PA-enabled products and has handled tens of billions of generated assets, including images, video, audio and documents.<\/p>\n
The company plans to expand Credentio beyond validation in the future, including capabilities for creating and embedding Content Credentials.<\/p>\n
A changing approach to AI labelling<\/p>\n
Google\u2019s move comes as technology companies face growing pressure to make AI-generated content identifiable without unnecessarily restricting how creators use it.<\/p>\n
The company is effectively separating visibility from traceability: the obvious watermark can be removed where regulations permit, but invisible provenance markers remain.<\/p>\n
For users, that means an AI-generated image or video may no longer visibly announce its origin, yet the information needed to verify its provenance can still remain embedded in the content.<\/p>\n","protected":false},"excerpt":{"rendered":"Google is giving users more control over how its AI-generated media is labelled. The company is introducing an…\n","protected":false},"author":2,"featured_media":140922,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[9],"tags":[24,67194,26984,3234,132,1429,26986],"class_list":["post-140921","post","type-post","status-publish","format-standard","has-post-thumbnail","category-google","tag-ai","tag-ai-watermark","tag-c2pa","tag-gemini-ai","tag-google","tag-google-ai","tag-synthid"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/140921","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=140921"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/140921\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/140922"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=140921"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=140921"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=140921"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}