Google expands image-remixing tool Whisk to 100+ countries

Prompt engineering for images is often a massive pain. You spend twenty minutes describing lighting, camera angles, and art styles, only for the model to spit out something completely off-mark. Google thinks it has found a way around the keyboard.

Google just expanded Whisk, its experimental AI image remixer, to more than 100 countries. Originally unveiled last year as a US-only experiment, the web-based tool is now accessible globally across both desktop and mobile browsers.

Ditching text prompts for visual Lego blocks

Unlike Midjourney or standard image generators where you type out long paragraphs, Whisk is built almost entirely around image-to-image prompting.

Basically, the tool asks you for up to three reference images to define three distinct building blocks: the subject, the scene, and the style. You drop a photo of your pet for the subject, a photo of a Tokyo alley for the scene, and a vintage comic panel for the style.

From there, Google’s latest Imagen 3 model takes over, blending the visual cues into a coherent new image. Text prompts haven't completely disappeared, but they take a back seat, acting purely as optional notes to tweak specific details within those parameters.

Still an experiment, with notable gaps

While 100+ countries sounds like a massive rollout, several major digital markets are conspicuously missing from the map.

Users across the United Kingdom, the entire European Union, India, and Indonesia still can't access the platform. Google is also keeping Whisk firmly categorized as an "experimental" project under its Google Labs banner, meaning it's primarily a tech showcase rather than a finished commercial product.

Why it matters

We think Whisk gets one crucial thing right: visual ideas are best communicated visually. Forcing people to translate colors, textures, and vibes into text tokens has always felt like an unnatural workaround. By turning the prompt into a drag-and-drop mood board, Google dramatically lowers the friction for casual remixing and rapid concept brainstorming.

It is also a clever tactical move for Google’s Imagen 3. Instead of fighting OpenAI on text instruction nuance, Google is packaging its model into a fun, low-pressure sandbox that anyone can mess around with on their phone.

The catch is where this hits a legal wall. The absence of the EU and the UK strongly hints at ongoing copyright and regulatory headaches. When you make it effortless to lift someone's style and someone else's subject with zero attribution, commercial creatives will hesitate to touch it. For now, it is a brilliant visual toy, but its path to a professional production tool remains blurry.