Google Now Generates Images in AI Overviews: Zero-Click Comes for Brand Visuals

Since July 14, 2026, Google no longer just summarizes the web in its AI Overviews. It now creates images inside them too. These images don't say where they come from. They don't link back to any website. And the brands involved get no say.

Until now, a searcher got their answer as text, without clicking a single site. That's what people call "zero-click." Now it reaches images too. So keeping control of your brand's visuals becomes a real issue. Here's what changed, what the numbers already say, and the four concrete moves to stay in control.

What Google announced on July 14

On July 14, 2026, for Google Images' 25th anniversary, Google announced something new: AI Overviews can now create images. Its Nano Banana 2 model does the work. In plain terms, a search can show an AI-invented image at the very top of the page — with no site opened or even cited. The rollout is gradual. It hits English-language searches first, according to reports from Search Engine Land and Search Engine Journal on July 14 and 15.

The contrast with text answers is huge. When an AI Overview answers in text, it shows its sources. Those are links. Few people click them, sure, but they exist and you can measure them. An AI-created image shows nothing. No credit. No link back. No trace in your analytics tools. That's what the first analyses of the brand angle point out. Your product, your industry, or your category can be illustrated by AI. And it draws on what it learned from the web — including your own content. All of this without warning you, and without citing you.

Why zero-click goes even further

This didn't come out of nowhere. It adds to a system that already captures a big share of attention. Today, AI Overviews appear on 48% of Google searches. And when they appear, the top organic result — the one that isn't an ad — loses a lot of clicks: Ahrefs measured a drop from 0.039 to 0.016. That's a 58% fall for the best-placed result on the page.

Until now, zero-click was only about text. The searcher read the answer, then left. With images, it reaches the visual too. A searcher can see "what a product, a layout, or an idea looks like." And no brand supplied or approved that image. In some industries, the image drives the whole decision: online retail, home decor, travel, restaurants, real estate. For them, one more part of the buying journey now plays out without ever leaving the results page.

What this changes for your brand

The risk is real, and it hits three points. The first is accuracy. When AI illustrates your type of product, it can create an image that's rough, out of date, or plain wrong. And that's the image the searcher sees first. The second is credit. With no link or mention, all the money spent on your real images earns you nothing in this block: your product photos, your visual style, your creative work. The third is measurement. An AI-created image leaves no trace on your side. You can't know how many people saw it. This blind spot already existed for text. Now it extends to images.

Faced with this, one reaction seems logical: block Google's crawlers so they stop reading your pages. Bad idea. Our study of 100 French media outlets made this clear. A site that shuts the door on these crawlers disappears from AI answers. And it controls nothing in exchange. Staying silent doesn't take you out of the conversation. It just lets others speak in your place.

Getting cited remains the only proven protection

The right answer is to go on the offensive. The goal: become the source Google cites and reuses. It's also the only lever whose effect we can put in numbers. According to Seer Interactive, brands cited in an AI Overview earn 35% more organic clicks and 91% more paid clicks than brands left out of the block.

In other words, the same AI Overview that starves the absent of traffic gives more to the present. Getting cited by AI is no longer just a communications goal. It's the only proven protection against zero-click — for text and for images alike.

Four moves to stay in control

Here are four moves to make. They range from the most technical to the most everyday. Together, they raise your odds that Google leans on your real content — not a generic blend — to build its answers.

1. Complete your structured data. These are pieces of information tucked into your pages' code so machines can understand them. Three tags matter here: Organization, Product, and ImageObject. They clearly tie your real images to your brand, with license and credit that machines can read. This is work on your page template, not writing. And it's exactly the kind of information AI uses first to correctly identify a brand.

2. Treat your images like real content. Four things matter: a clear description for each image (the "alt text"), meaningful file names, credit and license info embedded in the file, and an up-to-date list of your images for Google. Every well-labeled official image gives the engine one more reason to show yours instead of inventing one.

3. Make every page easy to cite. Start with a clear, complete 40-to-60-word answer. State the source of each fact. Organize the page around questions. These fundamentals remain the front door: showing up in the text AI Overview is what earns the 35% of clicks Seer measured. The complete GEO guide explains the method in detail.

4. Track your presence in AI answers. Since June, Google Search Console has offered reports on generative AI. They show how often your pages appear in AI Overviews — but without the clicks or the searches. Add regular tracking of your citations on answer engines like ChatGPT or Perplexity. You'll see when your brand appears, and above all when it disappears.

To see where your site stands on these four points, the LightSpot audit runs 46 criteria across 3 pages for free. Among other things, it checks your structured data, how easy your pages are to cite, and AI crawler access to your site. Then it hands you the top 3 fixes to make first.