Google Will Start Labeling AI Images in Search Results

Google is strengthening content transparency with C2PA technology across its products, helping users identify generative AI content. Together with industry partners, more sustainable, interoperable solutions are being developed to increase trust online.

Share
Google Will Start Labeling AI Images in Search Results

Google has announced that in the coming months it will start labeling AI-generated and AI-edited images in its search results. This comes as a response to the growing popularity of AI image generators, which has made such images increasingly common, even in searches for real people.

As AI is used more and more in products and services to boost creativity and productivity, it's becoming increasingly important for people to understand how content is created and modified. To support this, significant investment is going into tools and solutions like SynthID, which provides information about a piece of content's origin and how it's changed over time.

Security-Enhancing Technology for Better Content Control

Provenance technology can reveal whether an image was taken with a camera, edited with software, or created by generative AI. This information helps users make more informed decisions about the content they interact with, strengthening media literacy and trust. As a member of the steering committee for the Coalition for Content Provenance and Authenticity (C2PA), Google has been working to further develop the technology for attaching provenance information to content. The latest version, Content Credentials 2.1, offers greater security and better protection against tampering, reducing the risk of data being altered or misleading.

Rolling Out the C2PA Standard Across Google Products

Over the coming months, the latest version of Content Credentials will be integrated into several of Google's products:

Search: If an image contains C2PA metadata, users can see whether it was created or edited with AI tools via the "About this image" feature in Google Images, Lens, and Search.

Ads: Google's advertising system is also starting to integrate C2PA metadata. This will scale up over time to improve policy compliance.

In addition, the possibility of showing C2PA information on YouTube when content is captured with a camera is being explored. More updates on this are planned later this year.

Google encourages other major players to do the same.

Establishing and signaling content provenance is a complex challenge, and there are no simple solutions that cover all content online. That's why it's crucial to collaborate with other players in the industry to develop sustainable, interoperable solutions. Google encourages more services and hardware vendors to adopt C2PA Content Credentials.

Google's work with C2PA complements its broader strategy for transparency and responsible AI development. One example is SynthID, an embedded watermarking technology from Google DeepMind, which will soon be introduced across more generative AI tools and media formats like text, audio, image, and video. Google has also joined several other coalitions and groups focused on AI safety and research, and has introduced a Secure AI Framework (SAIF).