Google’s Circle To Search feature can tell you if an image was generated by AI

As Google introduces new models and tools for generating AI content, it also makes it a little easier for people to answer the question “was this created with AI?” » The company is expanding its SynthID AI detection system so that features such as Circle to Search and Lens are able to identify AI-generated and edited images.

SynthID is the watermarking system developed by Google that adds invisible metadata to content created or edited with its own AI tools. Last year at I/O, the company launched a dedicated SynthID detector and later integrated the functionality into the Gemini app. Now, Google is also extending these AI detection capabilities to Google Chrome and Search.

With the update, users will be able to use SynthID to assess the origin of an image via a number of methods. On Android, Google’s Circle to Search will be able to flag AI content. Google Lens and the Chrome version of Gemini will also be able to answer queries like “Is this AI generated?” » According to Google, these tools will be able to provide granular details about a given image. In one example shared by the company, Google says a given image was initially captured from a Pixel phone and then edited with AI-enabled tools in the Google Photos app.

Google

That said, Google won’t be able to provide such granular details about every potentially AI-altered image you’re likely to encounter. The company added content identification (a separate, industry-standard watermarking system) to the Pixel 10 line’s native camera app and is now extending the technology to its Pixel 8 and 9 models. But while Google has some very good insights into the provenance of images captured and edited through its own tools, it may not be as reliable when it comes to flagging content from other AI platforms.

On this front, Google says OpenAI, Kakao, and ElevenLabs have committed to integrating “SynthID technology into more of their AI-generated content.” In a blog post, OpenAI said its integration will start with images created with “ChatGPT, Codex or the OpenAI API.” Google is also expanding support for content IDs, which should also make AI detection a bit more universal. The Gemini app is expected to gain support for content credentials starting today, with integrations into Chrome and Search expected in the “coming months.”

Of course, no AI watermarking system is foolproof and there are myriad ways for users to evade detection tools and watermarks. But in an age where AI-generated content is becoming increasingly compelling and ubiquitous, making it easier for people to check out what’s in front of them is a useful first step.