Google to Let Users Disable Visible AI Watermarks While Invisible SynthID Tracking Remains

TL;DR

• Google is adding a new toggle that lets users turn off the visible watermark on AI-generated images and videos from tools like Gemini, Imagen, and Veo.
• Even with the visible mark disabled, every file will still carry Google's invisible SynthID watermark embedded in pixels, audio, and metadata for AI detection.
• The move gives creators cleaner, more professional-looking outputs but raises new questions about copyright, misinformation, and how to maintain trust online.

A Quiet Change With Big Implications

Google is making its AI-generated content look a lot more human. In a subtle but significant update rolling out this week, the company is giving users the option to remove the small, visible watermark that currently appears on images and videos created with its generative AI models.

The watermark — a small "Made with Google AI" or Gemini sparkle logo typically placed in the corner of an image or video — has been a standard feature since Google expanded access to Imagen 3 and Veo. Until now, there was no official way to remove it without cropping or editing the file. That is about to change.

How the New Toggle Works

The new setting is appearing as a simple on/off switch in the settings menus for Gemini, Google Labs, and the Vertex AI platform for developers. When enabled, the visible watermark will no longer be automatically applied to new generations.

According to Google, the toggle will apply to outputs from its latest models, including Imagen 4 and Veo 3, and will be rolling out gradually to personal Google accounts and Workspace users over the coming weeks. Free users will see the option in the Gemini app's settings under "Media generation," while enterprise and developer users can manage it at the project level in Google Cloud.

Google notes that the setting only affects future generations — it will not retroactively remove watermarks from images and videos already created. Users who want a clean, watermark-free look for presentations, marketing assets, or social posts will no longer need to rely on workarounds.

SynthID: The Watermark You Can't See

Disabling what you can see doesn't mean disabling what Google can detect. The company emphasizes that its invisible watermarking system, SynthID, will remain active and mandatory on all AI-generated content, even when the visible badge is turned off.

Unlike a logo in the corner, SynthID works by embedding a digital signal directly into the data of the image, video, audio, or text itself. It subtly alters pixels, sound waves, or word choices in a way that is imperceptible to the human eye and ear but can be detected by Google's identification tools.

Google says SynthID is designed to survive common edits like cropping, compressing, filtering, or screen-recording, making it a more robust way to prove provenance than a visible mark. Anyone will still be able to check if a file was AI-generated by using Google's SynthID Detector tool or through emerging open standards like C2PA, which Google supports.

Why Google Is Making the Change Now

So why offer the choice at all? The decision appears to be a direct response to creator feedback.

Since the launch of more photorealistic models, artists, marketers, filmmakers, and everyday users have complained that the visible watermark made AI outputs look unprofessional and limited their commercial use. Competitors like OpenAI and Midjourney do not apply a prominent visible watermark by default, putting Google at a disadvantage for users who wanted clean, client-ready assets.

By making the visible watermark optional while keeping the invisible one, Google is attempting to strike a balance: give creators the polished, unbranded look they want without abandoning its responsible AI commitments. The company has framed the move as part of its broader push for "helpful and responsible" AI — empowering users while preserving tools to fight misuse.

What It Means for Creators and Copyright

For creators, the change is largely welcome. Photographers, designers, and video producers using Gemini and Veo will now be able to generate and share work without an obvious AI label distracting from the final product. This could make Google's tools far more appealing for professional workflows, portfolios, and advertising where aesthetics matter.

However, the update also complicates the conversation around copyright and ownership. A visible watermark was an immediate, transparent signal to viewers and clients that a piece of content was synthetic. Without it, creators will need to be more proactive about disclosing AI use themselves, especially in commercial contexts where authenticity matters.

For rights holders, the reliance on an invisible system means enforcement will depend entirely on detection tools rather than visual inspection, which could make casual infringement harder to spot.

The Trust Problem: Can You Still Tell What's AI?

The biggest concern is what this means for trust online. Visible watermarks, while imperfect, were one of the few friction points slowing the spread of AI-generated misinformation. With that visual cue now optional, the burden shifts almost entirely to invisible detection and media literacy.

Experts warn that most people will never use a SynthID detector before sharing an image or video they see on social media. While Google, Meta, and others are pushing for industry-wide adoption of provenance standards like C2PA — which adds tamper-evident metadata about a file's origin — adoption is still far from universal and metadata can be stripped.

Google maintains that an invisible watermark is actually more trustworthy than a visible one, precisely because it cannot be easily cropped out. The company argues that as generative AI becomes more realistic, relying on a small logo was never a sustainable defense against deception. The future, it says, is in cryptographic and AI-based detection that works even when nothing looks AI-generated at first glance.

Whether users, platforms, and regulators agree with that vision will determine how this quiet toggle reshapes the look — and the credibility — of the internet's next wave of content.

AndroGuider Team

Articles written by the AndroGuider team. We try to make them thorough and informational while being easy to read.