AI slop is everywhere. It is strewn across the internet, stuffed into social feeds, and even popping up in real-world printed materials. If you have a good eye or you’re simply wise to bad imagery, you may be able to spot these on your own. If not, Google has an AI slop detector designed to make it easier to tell fake images from real ones.
The tool is called SynthID Detector. Why SynthID, you ask? The moniker is borrowed from the namesake of Google’s invisible digital watermark embedded into every image created with Nano Banana, the image generation tool found in Google Gemini. Every time you make an image with Gemini, it contains invisible digital noise that tells other apps and services that the image was made with AI and that it came directly from Google’s platform.
Google’s AI image slop scanner could be helpful when you encounter AI-generated photos online, but it is not foolproof.
Since you can’t see SynthID details in an image with the naked eye, these photos can proliferate online spaces without anyone knowing they’re not real. Google hopes to fix that, though, with SynthID Detector, which can decipher the origin of some AI images — emphasis on some.
To use SynthID Detector, go to the webpage and sign in with your Google account. Then find an image that you suspect to be AI, and drag and drop it into the search box. If successful, the page will instantly display the AI origin of the image, and it can even tell you if the photo has been edited further from its original version.
SynthID Detector/Zach Laidlaw
You can scan images, videos, and audio files across a broad range of formats that you would typically encounter on the web so that all your bases are covered. Unfortunately, there are some other limitations to keep in mind before you rely on SynthID Detector too much.
As a Google-made service, SynthID Detector is designed first and foremost to detect images with Google’s proprietary SynthID watermark embedded inside. Right now, that includes images generated with Gemini, along with a couple of other partners, such as Nvidia, OpenAI, and Kakao, an AI company from South Korea that you have never heard about.
That’s it.
RELATED: Google's data-devouring AI wants everyone's personal information
solar22/Getty Images
Google is working to expand SynthID support into other AI image generation companies, apps, and services. For instance, Apple Playground, the image generation app that comes with newer iPhones, will soon be compatible. However, the vast majority of image creation tools will slide under SynthID Detector … well … undetected. And there are a lot of them: Midjourney, Adobe Firefly, and the open-source versions of Stable Diffusion, Flux, and DALL-E 3, just to name a few.
So in other words, Google’s AI image slop scanner could be helpful when you encounter AI-generated photos online, but it is not foolproof. If the image comes from a non-SynthID-partner, any queries will show up empty.
SynthID Detector/Zach Laidlaw
Another area where SynthID falls short is in detecting real images. Because the system is designed to search only for SynthID watermarks, it can’t decipher when no watermarks are present at all. Therefore, real images are treated just like AI-generated images from non-SynthID-partners. It would be nice if SynthID Detector could at least tell when a real photo is real, but in this aspect, it is virtually useless.
As AI becomes more prevalent, society will likely need more tools like SynthID to help us discern factual content from fiction. Unfortunately, while SynthID Detector is a good first start, it shows just how severely lacking AI detection is when working with non-centralized standards for AI creation, labeling, and distribution. To truly be beneficial to mankind, detection tools need to either adopt robust scanning features or AI needs standardized policies, neither of which are likely to happen soon.
Like Blaze News? Bypass the censors, sign up for our newsletters, and get stories like this direct to your inbox. Sign up here!