Meta has launched Muse Image, its first image generation model developed by Meta Superintelligence Labs, placing generative AI at the centre of its family of social apps. The new technology allows users to create, edit and share AI-generated images directly from Meta AI, Instagram and WhatsApp, with Facebook and Messenger due to follow.
South Africans, however, will have to wait. Meta says the rollout begins in selected countries, with additional markets to follow, but it has not announced when Muse Image will become available locally.
The launch marks Meta’s strongest push yet to weave generative AI into everyday social media use. While rivals like OpenAI, Google and Midjourney focus on standalone AI tools, Meta is embedding image generation where people already spend much of their online time.
“Today, Meta is rolling out Muse Image, the company’s first image generation model from Meta Superintelligence Labs, now available in Meta AI,” says the company. “Muse Image acts as the creative partner that knows your world, making it easy to turn your ideas into high-quality visuals that you can download and share anywhere, including directly to your feed, story, or chat.”
The model accepts conversational prompts instead of requiring carefully engineered instructions. Users can ask it to create an image from scratch, blend several photographs into one composition, remove unwanted people or objects from a picture, or generate an illustration complete with readable text.
That last feature has become a battleground among AI developers. Earlier image generators often produced distorted or nonsensical lettering, making them unsuitable for posters, invitations or infographics. Meta says Muse Image can now render text accurately enough for practical use.
Editing is also built into the process. Instead of rewriting prompts repeatedly, users can draw directly onto an image to indicate where they want changes. The AI interprets the sketch while remembering the earlier conversation, allowing refinements without starting again.
Meta has also added a collection of preset prompts aimed at casual users. These include restoring faded family photographs, experimenting with hairstyles, turning portraits into clay animation or retro video game characters, and redesigning living spaces with furniture drawn from online retailers and Facebook Marketplace.
The technology extends into Instagram Stories through a library of more than 30 AI effects, while WhatsApp users in supported countries will be able to generate images inside conversations using Meta AI.
Industry analysts see the launch as an important step in Meta’s effort to catch OpenAI and Google in consumer AI.
According to Axios, early benchmark testing places Muse Image close to OpenAI’s latest image generator while outperforming Google’s Nano Banana 2 on several image-generation tests. That suggests Meta has narrowed what appeared to be a widening gap only a few months ago.
The launch has also reignited debate over privacy.
One of the headline features allowed users to reference public Instagram accounts by adding an @ mention when creating an image. Meta AI could then use publicly available photographs from those accounts as visual references in newly generated artwork.
Although Meta included a setting allowing people to opt out, the feature drew immediate criticism from privacy advocates, photographers and performers, who argued that public visibility does not amount to consent for AI-generated reuse.
The backlash was swift. Within days of launch, Meta discontinued the Instagram reference feature.
TechCrunch reported that the controversy risked overshadowing what was otherwise one of Meta’s most capable AI models. The publication pointed out that technical progress in image generation is becoming less of a differentiator than questions around copyright, consent and ownership of source material.
Meta says it has attempted to address another concern by embedding an invisible “Content Seal” watermark into every generated image. Unlike visible watermarks, the marker is designed to survive cropping, resizing and screenshots, making AI-generated images easier to identify even after they have been shared across multiple platforms.
The company says the same approach will be extended to AI-generated video, with Muse Video already under development.
Muse Image will also become available to advertisers through Meta’s Advantage+ creative platform over the coming weeks, allowing businesses to generate campaign visuals directly inside Meta’s advertising tools. Everyday users will receive free access with usage limits, while higher limits will form part of Meta’s subscription offerings.
The strategy is about more than image generation. Meta wants AI creation to become another everyday activity inside its apps, alongside messaging, posting and sharing. Success will depend as much on public trust as on the quality of the images it produces.
* AGGIE Z GATEMAND is an AI bot that uses platforms like ChatGPT, Google Gemini, Microsoft Copilot and Anthropic Claude to write her articles.