- You can now identify AI-generated images in the Gemini app
- Google uses SynthID technology to identify any AI image created using a Google app
- Google will also maintain a visible watermark on all AI images created on the free and Google AI Pro tiers
AI-generated images, at least those made using Google apps, just got easier to spot. From today you can upload an image that you’re unsure about into the Gemini app on your mobile and simply ask it, “was this image generated by Google AI?”
The technology that enables this feature is called SynthID. Google says it is starting its implementation on images, but will expand it to audio and video soon.
A release from Google says: “In addition to SynthID, we will maintain a visible watermark (the Gemini sparkle) on images generated by free and Google AI Pro tier users, to make images even easier to detect as Google AI-generated.”
Although Google adds that, “Recognizing the need for a clean visual canvas for professional work, we will remove the visible watermark from images generated by Google AI Ultra subscribers.”
Uploading the image into the Gemini app is easy. All you need to do is tap the + icon at the bottom of the screen, then choose either Files or Gallery, depending on where the file is located (Gallery here means Photos, for the iOS users amongst us). Once you’ve uploaded the image, just type in a prompt like “is this image AI generated?” and tap the > icon to send it.
Is it effective?
Google introduced SynthID in 2023. Since then, over 20 billion AI-generated pieces of content have been watermarked using it. When you ask Gemini if an image you’ve uploaded was made with AI, it will check the SynthID watermark, then use its own reasoning to return a response that gives you more context about the image.
I’ve tried using Gemini to verify if an image I created using the new Nano Banana Pro was AI-generated, and it identified the invisible watermark, although the visible Gemini watermark – the sparkle – at the bottom right of the image was a dead giveaway, anyway.
But how does Gemini fare with AI images that aren’t created by Google? Obviously, when I uploaded an image created using ChatGPT (which doesn’t use a visual watermark) into Gemini it couldn’t locate a SynthID watermark (because there isn’t one) and could only make its best estimation of whether the image was AI from visual clues. It decided it was, but I didn’t feel like it was particularly sure.
I tried it with a random image I found on the Internet and it said “This image is not made with Google AI, but it’s not possible to determine if it was generated with other AI tools.”
My overall impression is that Google’s efforts to help detect AI images are a step in the right direction, but until there is a digital watermark system that is widely adopted by all the AI image generation platforms, and aimed at consumers, it remains a flawed system at best.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.
