Meta rolls out Muse Image, its first Superintelligence Labs image model
Developed over time first seen 3 months ago
Meta has launched Muse Image, the first AI image-generation model developed by its Superintelligence Labs division, which now powers the image-making tools across the Meta AI app, Instagram and WhatsApp, with Facebook and Messenger to follow. The model is part of Meta's new Muse family of AI models, which is replacing the company's Llama lineup, and it signals the direction of the division led by Alexandr Wang, whom Meta hired last year. A notable and potentially controversial feature lets users "@ mention" other Instagram accounts in prompts so the model can incorporate those people's likenesses into generated images.
According to Meta, tagging a username allows Meta AI to draw on that account's public photos to build a visual, though users can control how others reuse their content for AI. Wang described Muse Image as "agentic", saying it works with the Muse Spark large language model to reason through prompts, search the web and plan before generating, and he also teased a forthcoming Muse Video model. Other capabilities include transforming images via suggested prompts, designing invitations and postcards, redesigning rooms from a photo pulled off Facebook Marketplace, and editing images by drawing directly on them; the model will also power 30 new AI effects coming to Instagram Stories in the US before a wider rollout.
- Meta's first Superintelligence Labs image model, Muse Image, is now live.
- Users can @ mention Instagram accounts to add their likeness to images.
- Muse family replaces Llama; a Muse Video model is coming next.
New here? Start with this
Meta has set up a division called Superintelligence Labs, led by Alexandr Wang, who Meta hired last year to spearhead its push into artificial intelligence. Muse Image is the first image-generation tool built by this team, and it now runs the picture-making features inside the Meta AI app, Instagram and WhatsApp, with Facebook and Messenger due to get it soon. It belongs to a new "Muse" family of AI models that is taking over from Meta's previous Llama models.
The tool lets people type a description and get an AI-generated image, but it also has an unusual feature: users can "@ mention" another Instagram account in their prompt, and the model will use that person's public photos to include their likeness in the picture. Meta says account holders have some control over how their images are used this way, though the feature raises questions about consent and how people's likenesses can be used online.
This matters because it shows how deeply AI image tools are being built into everyday social media use, affecting how billions of Instagram, WhatsApp and Facebook users create and share content. It also marks a first major product from Wang's new division, offering an early sign of the direction Meta's AI efforts are heading, including plans for a future video-generation model.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Meta's @ mention feature represents thoughtful innovation that respects user agency and enables consensual creative collaboration. By allowing people to incorporate others' public photos into personalised creative projects, the company democratises sophisticated image generation across its massive user base. The inclusion of user controls—enabling people to manage how their likenesses are deployed—demonstrates responsible design that recognises public visibility whilst protecting individual choice. The agentic reasoning capabilities make the model genuinely useful for legitimate creative purposes, and this represents meaningful technological progress that drives innovation forward.
The case against
The @ mention feature raises significant consent and privacy concerns that user controls alone cannot adequately address. Using someone's public photos to generate their likeness into entirely new contexts without explicit, informed consent represents a meaningful violation of autonomy and dignity. The feature creates substantial risk of non-consensual deepfakes, defamatory imagery, and targeted harassment, with public figures and women particularly vulnerable to manipulation and abuse. Public visibility does not constitute permission for AI-driven likeness use, and placing the burden on users to preemptively block such uses shifts responsibility from the platform to potential victims rather than addressing the underlying ethical problem at scale.
More coverage
Read the full article at the source →
Originally published by The Verge as “Meta’s new Muse Image model can pull other Instagram users into AI photos”.