- Google announced Verify AI at Google I/O 2026
- It allows Gemini to identify if and how content was altered by AI
- Verify AI is being supported by Nvidia, OpenAI, and others
A year ago AI content was very easy to pick out from a lineup — now it’s approaching impossible. Thankfully, Google I/O 2026 has showcased a possible solution: Verify AI.
Google has already rolled out an AI detector in Gemini which relies on invisible-to-the-naked-eye watermarks in Google-made content (from the likes of Veo and Nano Banana) to determine if something is AI-made or not. Now this tool will be coming to its Circle to Search tool so you can more easily scan content you see in your day-to-day using your Android phone or Chrome browser.
It’s not just coming to more places, this AI detection is getting a boost too. Across its AI products Google says it’s adding “quantum credentials verification” with support for C2PA and SynthID. Practically, this means that when you now feed an image into Gemini’s AI detection tool you’ll get a much more specific breakdown of where it came from and how it was edited (if at all).
So if it was originally a picture taken with a camera but later edited with AI — such as adding a party hat and shades to your puppy on his birthday — Gemini will tell you this context, or it could say if an image was entirely created by AI. Though I should note the AI won’t be able to say exactly how a photo was altered, just if it was changed by AI at all.
Perhaps most importantly, Google says it’s collaborating with various other companies in the AI space — including Nvidia and OpenAI — to bring these verification credentials to content created by tools across the industry.
This means Google’s tools won’t just recognize its own AI content, but images and videos created by a wide variety of AI sources.
An endless back and forth
This is certainly news to celebrate. AI-generated content is making it harder and harder to believe what you see online across all aspects of our lives.
On the more extreme end, you have people creating fake videos to disparage public figures or create public outrage with footage of events that never happened, but I’ve also found online shopping is getting harder — I’ve been browsing sites like Etsy to pick up some last-minute items for my upcoming wedding (like a cake topper and guest book) and so many of the listings seem to show AI images that are impossible to trust.
My hope is that we don't soon see the rise of tools that can crack Google’s AI detection, scrub away the hidden watermarks, and once again obscure the origin of AI-made images and videos. With promises of “quantum credentials verification,” I’m hoping this will be a tough nut for bad actors to crack, but we’ll have to wait and see how this game of cat and mouse progresses.
