Google is putting image generation directly into Search AI Overviews as part of a 25th-anniversary update for Google Images.
The July 14 announcement has two separate pieces. First, Google Images is getting a new browseable home on desktop in the US, with a real-time gallery tailored to a signed-in user’s interests. Second, AI Overviews in Search will be able to generate images from text prompts using Google’s Nano Banana model.
Google says the generated-image feature will roll out over the coming weeks in English, in regions that already support image creation in AI Mode. That makes this more than a birthday post for Google Images. It moves image generation into one of Google’s most visible answer surfaces.
Search becomes a creation surface
The product line between searching for an image and making one is getting thinner.
Google Images began as a way to retrieve pictures from the web. Lens then made an image itself a query. Multisearch let users combine images and words. The new AI Overview update continues that arc by letting a user ask Search for a visual that does not already exist.
That matters because AI Overviews already sit above normal results for many queries. If image generation becomes part of that surface, users may not need to open a separate creative app, visit a model provider, or choose between search and generation. A prompt inside Search can become the start of a visual asset.
The new Google Images home points in the same direction from the discovery side. Google says the gallery updates in real time and can be shaped by saved collections. For image-heavy planning, shopping, design, travel, fashion, and mood-board behavior, that gives Search a more app-like loop.
The rollout is still bounded
There are important limits in Google’s own post.
The new Google Images home is rolling out on desktop in the US, in English, for signed-in users. Image generation in AI Overviews is also rolling out over the coming weeks, and only where AI Mode image creation is already supported.
That means the story is not “everyone gets generated images in Search today.” It is that Google is testing how far Search can move from retrieval to creation while keeping the feature inside its existing regional, language, and AI Mode guardrails.
For publishers, marketplaces, and creators, that distinction matters. A generated answer surface may reduce some image-search trips, but it can also make visual research more sticky inside Google. The practical effect will depend on how often AI Overviews offer generation, how clearly generated media is labeled, and how easy it is to continue from an generated image into normal web sources.
