Meta Pulls Muse Image’s Tagging Feature Days After Backlash Over Using People’s Photos Without Consent
Meta disabled a controversial Muse Image feature three days after launch that allowed users to tag public Instagram accounts and generate AI images using those users' photos.
Meta reversed course on one of Muse Image’s core features just three days after launching it.
The tool, unveiled Tuesday as Meta’s first in-house AI image generator, had let users @-mention any public Instagram account and pull that person’s photos into an AI-generated creation, whether or not the tagged user had any relationship to the person doing the tagging.
In an update to its original launch announcement, Meta acknowledged the backlash directly: “Our intent was to provide a useful creative tool and to give people control over whether their public content could be referenced in this way.
We’ve heard the feedback that this feature missed the mark, so it’s no longer available.”
Hollywood Pushed Back Hard and Fast
The reversal followed pressure from the entertainment industry over the feature’s opt-out design, which included public Instagram accounts by default unless users disabled the setting.
Talent agency CAA, whose client roster includes Tom Hanks and Meryl Streep, reportedly said no one’s “name, image, likeness, voice or creative work should be used by any third party, including AI models, without clear, documented consent.”
SAG-AFTRA separately urged its members to dig into their Instagram settings and “take action to protect your likeness.”
Following Meta’s reversal, CAA praised the company’s response, saying it “commends Meta for its swift decision to remove the Muse Image feature,” calling consent-first design essential to responsible technology.
The episode closely echoes the backlash OpenAI faced over its now-shuttered Sora video app, which similarly launched with an opt-out likeness policy before being scaled back, underscoring.
A Second Problem: Meta’s Own Detection Tool Falls Short
The tagging controversy was not Muse Image’s only challenge.
A Reuters analysis found Meta’s AI detection tool, designed to verify Muse-generated images, failed to identify 55% of a 40-image sample after cropping, even though the tool successfully verified every uncropped original.
Meta’s detection system relies on an invisible Content Seal watermark embedded in each image the model generates, just as Google’s SynhID for probing AI-generated media.
When Reuters asked about the findings, Meta said the tool remains a preview and that while the watermark is built to survive routine edits, its signal can be lost under heavy cropping.
Meta’s Oversight Board raised the same concern in March, urging stronger detection tools to address the “proliferation of deceptive AI-generated content.”
Why Detection Gaps Matter Beyond Meta
Outside researchers say the limitation is not unique to Meta.
Siwei Lyu, a computer science professor at the State University of New York at Buffalo, said watermark-based detection can be effective when signals remain intact but can weaken or disappear after cropping, resizing, compression, or editing.
Meanwhile, both Google and OpenAI have issued similar caveats about their detection systems.
These technical vulnerabilities underscore the United Nations’ warning that AI is advancing faster than governments can oversee. And when baseline verification tools fail so easily, public policy simply cannot keep up.
Source: Meta scraps AI image feature days after launch following privacy backlash
