Half has retired a controversial feature of Muse Image which allowed you to generate images with theTO THE using photos posted by other people on Instagram as a reference. The system could thus depict a person in situations that had never actually occurred.
After the protests on lack of prior consent and about possible abuse, Meta admitted that the feature “missed the mark” and has removed just three days after launch. The provision, however, it did not affect the entire generatorwhich continues to be available within the company’s AI-based services, but only the mode that allowed photos from other public accounts to be used in creations has been eliminated.
How Muse Image by Meta worked
Muse Image is the first image generation model developed by Meta Superintelligence Labsthe company’s division dedicated to artificial intelligence. The tool allows you to create content starting from a textual descriptionedit existing photographs and combine several in the same composition.
The capacity linked to Instagram allowed for mention a public profile in the prompt using the @ and to use its images in generation. A user could, for example, ask Meta AI to create a scene in front of a monument or in a fictional place and add another person to it by indicating their account.
The system then used the photographs present in the profile to reproduce the subject’s face and appearance in the final result. There was no need to download them, upload them manually, or ask the owner for permission. These materials they were not necessarily used for training of Muse Image: They served as input for a single generation, similar to what happens when you load a photograph into an AI tool and ask it to be modified or reinterpreted.
The controversy, therefore, did not concern access to the images stored on users’ devices, but the possibility of transforming those already visible on the platform into new synthetic representations.
Because the function caused so many protests
The main problem was the consent. The novelty could involve public accounts belonging to adults, while private profiles and those of users under 18 were automatically excluded.
Anyone who did not want to allow this use of their posts and Reel had to intervene in the Instagram settings. The system was therefore based onopt-out: it was up to the person concerned to disable the option.
This mechanism has caused discussion because make a photo public it does not necessarily equate to authorizing its transformation with AI. An image accessible to all can be shared according to the rules of the social network, but its author may not expect to appear in a place, in a situation or with an appearance never personally chosen.
Meta claimed to have given its members control over its content. According to critics, however, authorization should have been requested earlier, without forcing people to independently seek the command to oppose.
Users were not notified if someone used their public Instagram photos
Another disputed point was the absence of a notification. The rules published by Meta indicated that the account holder would not be notified when someone called up their images through the platform’s AI tools.
The person involved could therefore have known nothing about the image created. The risk did not only concern public figures and entertainment professionals, but any user with an account visible to all. The instrument could of course be used innocuously, for example to create a fun composition with a friend. The ease with which it allowed the reproduction of a real person, however, also paved the way for less transparent uses.
Among the risks reported were: non-consensual modificationsharassment, impersonation and the production of misleading content. An image could have shown someone performing actions that never happened, or placed them in a false context, and then circulated outside the platform without clear indications of its origin.
This does not mean that Muse Image automatically allows you to make any illegal content. Meta applies security and moderation systems to their services, but the complaints concerned the possibility that these protections failed to prevent any abuse.
How could you disable the use of Post and Reel for AI images
Meta had provided a specific command for block employment of public posts and reels in AI-powered creations. To find it you had to reach the own profilepress the button ☰open the section Sharing and reuse and disable theoption dedicated to the company’s AI functions. The choice, however, occurred only after the initial activation.
With a system based onopt-inhowever, reuse would remain blocked until the account holder voluntarily agreed. In this way, permission would be requested before use and not taken for granted.
Because Meta retired the feature after just three days
Muse Image was presented July 7, 2026 and the negative reactions arrived almost immediately. The possibility of generating new images using photographs of real people has attracted the attention of users and agencies representing actors, creators and other entertainment professionals, particularly exposed to the risk of unauthorized digital replicas. Among the realities involved in the debate is also the Creative Artists Agency (CAA)which represents numerous public figures.
The about-face came on July 10th. Meta explained that he wanted to offer a creative tool, while at the same time giving people the opportunity to decide how to use their images. However, the comments received showed that the solution adopted did not meet public expectations.
The story highlights a distinction destined to become increasingly important with the integration of generative AI into social networks: making a photograph public does not mean authorizing any transformation of it. Being able to see content and having permission to create an artificial replica of the person depicted are two different issues. In the case of Muse Image, allowing users to say “no” was not considered sufficient, with critics saying consent should have been asked first.
