Google is changing the way it labels content created with artificial intelligence. The company will soon let users remove the visible watermark that appears on some AI generated images, videos and songs, giving creators more control over how their work looks.
However, there is an important detail that users should understand. Turning off the visible watermark does not remove Google’s hidden AI watermark or the other information that can help identify how the content was created. Google’s SynthID system and C2PA content credentials can still provide information about the origin of AI generated content.
This means Google is not simply removing AI labels. Instead, it is separating the obvious label that people can see from the hidden technology used to verify AI content. The change could be important for designers, video creators, musicians, businesses and anyone who wants to use AI generated content professionally.
What Is Google Changing?
Google is introducing a setting that allows users to turn off the visible watermark on certain AI generated content.
The setting will be available for Google’s Nano Banana, Omni and Lyria models. Users will be able to control the setting through Gemini and Google’s Flow video editor, while support for Google Search is expected later. The feature is rolling out gradually.
When the option becomes available, users will be able to go to Settings > Media Watermark and choose whether they want the visible mark to appear.
This is useful because visible watermarks can sometimes get in the way of creative work. A large or obvious mark on an image or video can make the content less suitable for a presentation, advertisement, design project or other professional use.
Google is therefore giving creators more control over the final appearance of their work.
What Is a Visible AI Watermark?
A visible watermark is a mark that people can see on an image, video or other piece of content.
You may have seen AI generated images with a small Google logo or other mark showing that the content was created using AI. The purpose is to make it obvious that the media is not a normal photograph, video or recording.
This can be useful for transparency because someone looking at the content does not have to guess whether AI was involved.
However, visible watermarks also have a downside. Imagine a designer creates an AI generated background for a business presentation or a video creator makes an AI generated scene for a project. A visible watermark may make the final work look less professional.
That is one reason Google is giving users the choice to remove it.
Removing the Visible Watermark Does Not Remove SynthID
This is the most important part of Google’s announcement.
When you turn off the visible watermark, Google’s hidden SynthID watermark remains.
SynthID is Google’s technology for placing an invisible signal inside AI generated content. A normal person looking at the image, video or listening to the audio will not notice it, but compatible systems can check for the signal. Google says SynthID has already been used to watermark more than 100 billion images and videos, as well as around 60,000 years of audio.
In simple terms, you can remove the label people see, but you cannot simply switch off the hidden signal that helps identify the content as AI generated.
This is how Google says it can balance creative freedom with transparency.
What Is SynthID?
SynthID is Google’s digital watermarking system for AI generated content.
Unlike a normal watermark, it is designed to be invisible. It can be added to different types of content, including images, videos, audio and text.
Google has been developing SynthID for several years. The company introduced it in 2023 and has since expanded it across many of its AI products.
The technology is useful because AI generated content can look very realistic. A person may look at an image or video and have no idea that it was created by artificial intelligence.
SynthID gives Google another way to check the origin of content.
Google has also added SynthID verification to Gemini, allowing users to upload an image or video and ask whether it was created using Google AI. The system checks for the hidden watermark and can tell users whether it detected SynthID.
What Is C2PA?
Google is also using another technology called C2PA.
C2PA stands for Coalition for Content Provenance and Authenticity. It is an industry standard designed to provide information about where digital content came from and how it has been changed.
Think of it as a digital record attached to a piece of content.
For example, C2PA information can help show whether an image was taken with a camera, edited with software or created using generative AI. Google says it is expanding its use of C2PA Content Credentials across its products.
This is different from the visible watermark. C2PA is mainly about providing information about the history and origin of a piece of digital content.
Why Is Google Making Visible Watermarks Optional?
The simple answer is creative freedom.
AI generated content is now being used for much more than experiments and funny pictures. Businesses use AI images for marketing, designers use them for ideas, video creators use AI generated scenes, and musicians can use AI tools to create sounds and songs.
In many of these situations, creators may not want a visible watermark covering the final product.
A company creating an advertisement, for example, may want the image to look clean and professional. A filmmaker may want to use an AI generated scene without a large label appearing in the frame.
Google appears to recognise this problem.
Instead of forcing users to choose between visible labeling and professional looking content, the company is allowing users to hide the visible mark while keeping its hidden transparency systems active.
Does This Mean Google Is Hiding AI Content?
Not exactly.
This is where the change can easily be misunderstood.
Someone may see an AI generated image without a visible watermark and assume Google has stopped identifying AI content. That is not what is happening.
The visible label can be switched off, but SynthID remains. C2PA information can also remain attached to supported content.
Google says this approach is meant to balance creative control and safety. Users get more control over how their creations look, while people and systems can still have ways to check whether AI was involved.
Can People Still Detect Google’s AI Content?
Potentially, yes.
Google has been building tools that allow users to check AI generated content through Gemini and Search.
For example, Google says users can ask Gemini whether an image was generated with Google AI. The system can look for the SynthID signal and provide information about what it finds.
Google has also been expanding these verification tools to Search and Chrome.
This means that the absence of a visible watermark does not necessarily mean the content cannot be identified as AI generated.
That distinction will become increasingly important as AI generated images, videos and audio become harder to recognise with the human eye.
Why This Matters for Creators
The new setting could be particularly useful for professional creators.
A graphic designer may use AI to create an image and then edit it heavily before using it in a project. A video creator may generate a background with AI and combine it with footage recorded with a camera. A musician may use AI generated audio as part of a larger production.
In these cases, a visible watermark may not always be desirable.
Giving creators the option to remove it means they can decide how the finished work looks without necessarily losing the ability to prove its origin.
This is a major shift from the idea that every AI generated piece of content must have an obvious label that anyone can see.
But There Are Also Concerns
The change raises some reasonable questions.
If visible watermarks can be removed, people may find it harder to tell at a glance whether something was created by AI.
This could become a problem on social media, where people often see an image or video without checking where it came from. A person may share an AI generated image believing that it is real simply because there is no visible label.
This is one reason Google’s hidden systems matter.
The company is betting that machine based verification can provide a better long term solution than simply placing a large label on every piece of AI generated content.
Google Is Not Abandoning AI Transparency
In fact, Google’s broader strategy suggests the opposite.
The company has been investing heavily in tools that help people understand how digital content was created. Google says SynthID has now been used on more than 100 billion images and videos, while its verification tools are being expanded across Gemini, Search and Chrome.
Google is also working with other technology companies and industry groups to expand common standards for content provenance.
The goal is to create a system where people can eventually understand whether a piece of content came from a camera, an AI model or another editing tool.
This becomes more important as AI generated media becomes more realistic.
Google Is Also Open Sourcing Credentio
Google is taking another step by open sourcing a library called Credentio.
The library is designed to help developers add local validation for C2PA Content Credentials to their own applications.
This matters because AI generated content does not stay inside Google’s products. Images and videos are downloaded, edited and shared across many different websites and apps.
If verification only works inside one company’s platform, its usefulness will be limited.
Open standards and tools that other developers can use could make AI content verification much more widespread.
What Happens Next?
Google is expected to roll out the visible watermark setting over the coming days.
Users will be able to control the setting through the relevant media watermark option once it becomes available to their account and supported products.
The feature will initially cover Google’s Nano Banana, Omni and Lyria models through Gemini and Flow, with Google Search support expected later.
The bigger change, however, is likely to be the growing separation between what people can see and what computers can verify.
The Bottom Line
Google will let you remove the visible watermark from AI generated content, but that does not mean Google is abandoning AI transparency. The visible label can be turned off to give creators more freedom, while hidden SynthID watermarks and C2PA information can still help identify where supported content came from.
This could become the more practical approach as AI becomes a normal part of creative work. A visible watermark is easy for people to understand, but it can also get in the way when AI generated content is being used professionally. Hidden verification gives Google a way to protect transparency without forcing every creator to put an obvious label across their work.
The bigger question is whether these hidden systems will become a common standard across the internet. Google is already working with other companies and industry groups to make content provenance more widely supported. If that effort succeeds, the future of AI transparency may not be about putting a visible “AI generated” label on everything. It may be about giving people and computers a reliable way to check the history of digital content whenever they need to.