Meta Scraps AI Image Feature After Privacy Backlash

Meta scraps AI image feature Muse Image just three days after switching it on. That is the headline. But the real story is how fast it happened. This was not a slow-burn controversy that regulators eventually caught up with. A product was launched on a Tuesday. It was buried by Friday afternoon, taken down by the very users it was built to impress.

 A Feature That Landed Wrong From Day One

Muse Image was meant to be a milestone. Built by Meta Superintelligence Labs, it was the company’s first proper image-generation model. It sat directly inside the Meta AI chatbot.

The pitch sounded simple. Type an “@” mention of any public Instagram handle, and the tool would pull from that account’s public photos. Then it would generate or remix images, sketches included. Meta called it a creative partner that “knows your world.” Users, however, called it something else entirely once they understood what that phrase actually meant in practice.

The Real Problem: Opt-Out, Not Opt-In

So what went wrong? Not the technology. The defaults.

Every public Instagram account belonging to someone over 18 was opted in automatically. No notification went out to account holders. Nobody asked first, and nobody was told afterwards. If you wanted your face out of other people’s AI-generated images, the burden fell on you to dig through settings and switch it off yourself.

Emmy winner Hannah Einbinder, known for “Hacks,” was among the first to call it out. She flagged the feature publicly on Instagram, telling followers it had been switched on without their knowledge, and urged them to opt out right away.

Hollywood and Privacy Groups Push Back

From there, the backlash moved quickly.

Talent agency CAA, whose client list includes Tom Hanks and Meryl Streep, went straight to Meta. Its argument was blunt: nobody’s name, image, voice or creative work should be usable by an AI model without documented consent. SAG-AFTRA joined in too, warning about nonconsensual digital replicas and urging members to disable the feature while it was still live.

Meanwhile, Privacy International, the London-based rights group, told the BBC this was simply the latest example of AI companies treating people’s images as raw material, consent or no consent.

Meta’s Response: A Quick Reversal

Meta did not dig in. By Friday, the company pulled the feature entirely.

“We’ve heard the feedback that this feature missed the mark, so it’s no longer available,” Meta said in its statement. The company added that its original intent was to give people control over whether their public content could be referenced this way. That intent, though, clearly did not survive contact with how the feature actually shipped. CAA welcomed the reversal, calling it a swift and responsible call.

A Second Privacy Rollback, Same Week

The timing gets messier from here. Around the same period, Meta also rolled back end-to-end encryption on Instagram direct messages. This was a feature the company had previously championed as standard across its apps.

As a result, Meta can now technically access message content, including photos, videos and voice notes, if it chooses to. That rollback landed just as users were already on edge about how their content gets used.

Regulators Are Watching Too

There is a regulatory thread here as well. In India, Electronics and IT Secretary S. Krishnan said the government stood ready to examine whether Muse Image complied with domestic law, should a formal complaint arrive. So even with the feature gone, this story may not be fully closed.

The Bigger Lesson for Meta

For Meta, the real lesson is not about the technology itself. It is about the rollout instinct. Opt-out by default, applied to other people’s likenesses, was never going to survive daylight. Three days proved that much.