Meta launched Muse Image on Tuesday, July 8. It is an AI image generator built by Meta's own AI research division. Most of what it does is unremarkable: type a description, get an image back. But one feature caused immediate problems.
Anyone could tag a public Instagram account in a prompt, and Muse Image would use that person's photos to generate new images of them. The account owner was not asked. They were not even told it had happened. Every adult with a public account was enrolled by default, with no notification that the feature existed.
Meta's position was that people could opt out by digging through settings. But as one privacy group put it, Meta had "buried the opt-out in account settings" while defaulting millions of people into a system they had never agreed to. That is a meaningful distinction. Opt-out means the burden falls on the person whose face is being used. Opt-in means the company has to ask first.
The backlash came from several directions at once. SAG-AFTRA, the actors' union, told all its members to turn the feature off. Talent agency CAA, which represents some of the most recognizable names in entertainment, contacted Meta directly and issued a public statement: "No one's name, image, likeness, voice or creative work should be used by any third party, including AI models, without clear, documented consent." Safety organizations flagged the obvious risk of the feature being used to generate fake explicit images of real people.
Meta pulled the feature by Friday. Three days after launch. The company said it had "heard the feedback" and that the feature was "no longer available."
This is worth paying attention to for reasons that go beyond celebrities. The underlying question Meta tried to answer with an opt-out default is one that every platform is currently wrestling with: who controls what happens to your image once it is posted publicly online? Meta's answer, at least briefly, was that posting publicly meant consenting to AI use unless you actively said otherwise.
That answer conflicts with how privacy law works in Europe and several other markets. Under GDPR, the standard that covers the EU and, by extension, any company operating there, using someone's personal data requires their prior consent. An opt-out default does not meet that standard for many types of data processing. The same logic applies in Brazil, Argentina, and parts of Asia. Meta almost certainly could not have run this feature in those markets in the first place.
The bigger signal here is not that Meta made a mistake and corrected it. Companies launch things and pull them back. What matters is the direction they were heading: using the photos that billions of people have already posted as raw material for AI, with the default set to yes.
For any business operator whose staff, executives, or brand accounts maintain a public presence on Instagram or any other major platform, the practical step is straightforward. Review the sharing and reuse settings on your public accounts. The specific feature is gone, but platforms will keep testing variations of this model. The default is unlikely to favor the account owner.
The consent battle between platforms and users over AI-generated images has not ended. Meta's retreat this week is a pause, not a resolution.