Meta launches Muse Image in Meta AI for chats & apps

Meta has launched Muse Image in Meta AI, the first image generation model from Meta Superintelligence Labs.

The model is now available through Meta AI and is already used in creative features across Meta's apps. It also powers 30 AI effects for Instagram Stories and image generation in direct chats with Meta AI on WhatsApp in selected countries.

According to Meta, Muse Image lets users create visuals from conversational prompts. The system is designed to handle more complex requests and combine multiple photos into a single image for download or sharing in chats and social feeds.

The rollout adds another image tool to Meta's consumer AI push as large technology groups compete to make generative software part of everyday messaging, social media and photo editing. Rather than offering it as a standalone service, Meta is embedding the software in products with large existing audiences.

Prompt tools

Users can ask Meta AI to generate images of historical landmarks, produce functional QR codes and place clear text inside visuals. The system can also create infographics and other images in a requested style while keeping written instructions legible.

A presets panel inside Meta AI offers suggested prompts to help users get started with common tasks. These include restoring old family photos and trying different hairstyles, with the results available for sharing across Meta's apps.

Meta also described a room redesign feature. Users can photograph a room and ask Meta AI to redesign it with products sourced from the web or Facebook Marketplace, applying a specified style to the space.

Editing workflow

Users can edit generated images directly by tapping a markup icon and circling or annotating changes on the image itself. The system keeps the wider conversation in context, allowing users to revise style choices or add elements without restarting the process.

Meta said Muse Image uses multi-step planning when processing prompts. Working alongside Muse Spark, another Meta AI model, it can plan image layouts, pull in real-time context from the web and blend visual references.

This approach is intended to handle more layered requests, such as inserting a pet into a painting or combining a selfie with a holiday picture. Meta said the model is designed to align the resulting image more closely with what a user requests.

Public content feature withdrawn

Meta also said a feature announced earlier in the week has been withdrawn following feedback. It had allowed users to generate images in Meta AI by mentioning public Instagram accounts.

Meta had presented the option as a way to reference public content for creative use while giving users control over whether their material could be tagged through an opt-out setting. The feature is now unavailable.

The withdrawal highlights the sensitivity around AI systems trained on or linked to user-generated content, particularly when those systems sit inside mass-market social platforms. Questions of consent, attribution and control have become central as companies look for ways to personalise AI outputs with material drawn from real people and public posts.

Wider rollout

Meta said Muse Image will expand to more countries and to Facebook and Messenger. It also plans to make the model available to advertisers through Advantage+ creative.

Standard image creation with Meta AI is free, while more extensive creation options sit behind Meta subscription plans. Meta is also developing Muse Video as another visual generation product.