As generative media becomes more advanced and easier to use, understanding where content comes from and whether it has been altered is more important than ever. Today, we’re expanding tools for content transparency and verification across Search, Gemini, Chrome, Pixel, and Cloud, while deepening our collaboration across the industry.
Expanding the technology even further
Three years ago, we introduced SynthID, an industry-leading digital watermarking technology that can embed signals invisible to the human eye into AI-generated content. Since then, SynthID has been built into our generative media models and products, watermarking more than 100 billion images and videos, as well as content equivalent to 60,000 years of audio.
Across an increasing number of generative media tools, we are adopting C2PA Content Credentials, an industry standard that shows how media was created and edited, whether or not AI was involved. The Pixel 10 is the first smartphone to add Content Credentials to photos in its native camera app. And in the coming weeks, we’ll extend this to video on Pixel 8, 9, and 10.
By using this technology at capture time, Pixel can record when the content was actually taken by the camera. In the age of generative media, we believe it is just as important to know whether a file was truly captured and left unedited as it is to know whether it was generated or edited by AI.
Offering more ways to verify content
Our goal is to make it easier to understand the content you see online. That’s why we recently added SynthID verification for images, video, and audio to the Gemini app. This feature has already been used 50 million times worldwide, and starting today, we’re gradually expanding it to Search. We’ll also roll it out to Chrome in the coming weeks.
In Search, you can verify the source of an image through features like Lens, AI Mode, Circle to Search, and Gemini in Chrome.
Just ask:
“Is this made by AI?”
or
“Is this AI generated?”
We’ve also added verification for C2PA Content Credentials, which lets you quickly check whether content is an original, unedited file captured by a camera, or whether it has been edited, and which tools were used. This feature starts rolling out today in the Gemini app, and will also come to Search and Chrome in the coming months. This builds on our work to label AI-generated content on YouTube and our development of Backstory with trusted testers to improve the speed and reliability of detection tools.
Working together across the industry
Because digital media moves across multiple platforms, industry-wide collaboration and the adoption of strong, interoperable tools are essential. As companies like OpenAI, Kakao, and ElevenLabs adopt SynthID for more AI-generated content, more content across the internet will carry these invisible watermarks. This builds on our decision to open source SynthID’s text watermarking technology, and our work with NVIDIA to watermark AI videos generated by the Cosmos world foundation models.
To help more organizations identify AI-generated media, we will offer a new AI Content Detection API on Google Cloud’s Gemini Enterprise Agent Platform. This will give businesses a powerful way to identify AI content generated not only by Google’s models, but by other leading models as well. They can use it to decide how to evaluate and manage media on their platforms, from feed ranking and backend workflows like fraud prevention in insurance, to user-facing features like fact-checking and labeling synthetic media. We’re starting with a limited number of trusted partners, and we’ll continue improving the API based on their feedback.
As members of the C2PA steering committee, we also continue to support a common global standard for source verification technology. This will allow the transparency tools built into our devices to work seamlessly across the platforms people use every day. For example, Meta, also a member of the C2PA Steering Committee, will begin showing Content Credentials for camera-captured media on Instagram. That means real photos and videos taken on a Pixel phone will soon be recognized and labeled as such when shared to Instagram.
For years, we’ve been working to add helpful context to the information people find online. Content transparency is a complex challenge, but we will continue advancing the technology and setting a high bar for the industry. Our goal is to give you the tools you need to determine the history of every piece of content you encounter.
