Meta has launched Muse Image, its first in-house AI image generation model, with a feature that lets users pull a friend’s Instagram likeness into a generated photo, raising familiar questions about consent, deepfakes, and synthetic content in social feeds.
Built by the Alexandr Wang-led Meta Superintelligence Lab, Muse Image is now available in the US through the Meta AI app, Instagram, and WhatsApp. It powers over 30 new Instagram Stories effects and brings image generation directly into Meta AI chats.
The prompt-as-account feature. The most attention-grabbing capability: tagging a friend’s Instagram account in a prompt incorporates their likeness into the output. Meta says the same privacy controls governing how user photos are reused on Instagram apply. But the feature closely resembles the viral appeal, and subsequent policy headaches, of OpenAI’s now-defunct Sora app, which let users generate videos containing friends’ likenesses.
Muse Image uses “advanced reasoning to understand complex prompts, seamlessly blending multiple photos into high-quality creations,” according to Meta. The model can generate functional QR codes, render legible text within images, and accept conversational follow-up prompts for iterative edits. In the Meta AI app, users can draw or write on a photo to indicate desired changes.
Commercial ties. The model also integrates with Meta’s commerce ecosystem. Taking a photo of a room and asking Meta AI to redecorate can pull furniture listings from the web and Facebook Marketplace.
Free usage is available for “everyday creation,” with limits that reset after a period. Users who hit the cap are prompted to subscribe to Meta One for additional generations. A Muse Video model is in development.