OpenAI, Kakao Adopt Google's SynthID Watermark

In a video posted to YouTube, Google DeepMind announced that OpenAI, Kakao, and Lavin Labs are adopting SynthID, Google's watermarking technology for AI-generated content. SynthID embeds an invisible signal inside image, video, and audio files, allowing machines to later verify whether that content was made by AI. Until now, Google had only applied this technology to its own products, like Gemini and Nano Banana. What stands out this time is that outside companies — including direct competitor OpenAI — are starting to adopt the same standard. "Millions of people are already using the SynthID detector in the Gemini app," Google DeepMind said in the video.

To break it down, SynthID is a watermark that's invisible to the human eye but embedded directly in a file, letting machines later determine whether an image, video, or audio clip was AI-generated. Back on August 14, Google had already said that while the visible watermarks on Gemini and Flow outputs could be turned off, the invisible SynthID signal and C2PA metadata would remain intact regardless.

Content Credentials Show a Photo's History

Google also said it's expanding Content Credentials verification across more of its products. This feature indicates whether a photo or video was captured by a camera or generated by AI, and it also flags whether the content was later edited using AI tools. One example shown in the video involved a photo taken with a Pixel camera and later touched up in Google Photos, with Gemini identifying both the capture device and the editing history. Rather than just labeling something AI or not, the system traces each stage the content has passed through.

Checking AI-Generated Content Directly in Chrome

Google said it's building SynthID and Content Credentials verification directly into Chrome's search features. Users can circle an image on screen with Circle to Search, or right-click an image and ask, "Was this image generated with AI?" to get an answer. This capability was previously available only inside the Gemini app — now it's built into the browser itself.

How to Use It

The expanded content-verification feature is available directly in Chrome.

1. Use the Circle to Search gesture over the image you want to check, or on desktop Chrome, right-click the image.
2. From the menu that appears, ask "Was this image generated with AI?"
3. Chrome reads both the SynthID watermark and the Content Credentials data together, returning an answer that indicates whether the image was AI-generated, camera-captured, or later edited with AI tools.

Google didn't specify which countries will get the feature first or whether it will extend beyond Chrome to other browsers. That said, as outside companies like OpenAI and Kakao begin adopting SynthID, images they produce are increasingly likely to be checkable through the same Chrome workflow.

Editor's View

The fact that Google is bringing a watermark it once kept in-house to a direct competitor like OpenAI signals an ambition for this to become industry-wide trust infrastructure rather than a single company's feature. Competition in AI image generation remains fierce, but there's a clear logic at play: aligning on shared standards for verifying whether output is real or fake benefits everyone in that layer. For OpenAI, embedding SynthID in its own image outputs also adds one more layer of defense if deepfake controversies arise.

Placed alongside the August decision to let visible watermarks be turned off, the pattern becomes clearer. Google is loosening restrictions on the visible watermark, letting users remove it, while simultaneously expanding verification of the invisible SynthID signal and Content Credentials all the way into Chrome search. Visible markers become optional, while the invisible tracking layer keeps getting more thorough. Compared to the previous approach — where checking whether something was AI-generated meant uploading a file to a separate verification site — this is moving toward a world where a single right-click in the browser does the job.

It's also worth noting that Kakao, a Korean company, is among the early adopters. If SynthID gets built into Kakao's image and video generation services, Korean users gain one more channel for verifying content authenticity. When disputes arise in Korea over whether a news photo or an image circulating on social media is genuine, getting into the habit of doing a first-pass check via Chrome's right-click feature isn't a bad practical move. That said, limitations remain: the absence of a watermark doesn't guarantee something isn't AI-generated, and content from generation tools that don't use SynthID won't be caught by this method at all.

In the coming weeks, more generative AI companies are likely to announce adoption of SynthID or Content Credentials. Being early to establish verification standards could also put a company in a stronger position once regulatory discussions on this front move forward.