Meta turns off the Instagram feature that let users make AI deepfakes of public accounts

Meta has pulled a newly launched Instagram feature that let users generate AI images referencing any public account simply by tagging it in a prompt. The capability was introduced as part of Meta's rollout of its Muse Image AI model, which is designed to offer creative image generation directly within the platform. The backlash was swift, and the feature was disabled within days of its announcement.

The core concern was straightforward: as originally implemented, the feature required no consent from the account being referenced. Any public Instagram profile - whether belonging to an individual creator, a celebrity, or a brand - could be pulled into AI-generated imagery by anyone who tagged them. That raised immediate questions about potential misuse, including the creation of realistic-looking images of real people without their knowledge or approval.

Meta acknowledged the issue in an update to its original Muse blog post, noting that its intent had been to provide a useful creative tool and that it had underestimated how the feature would be received. The company did not offer a detailed timeline for any revised version of the functionality, leaving open the question of whether it will return in a modified form with opt-in controls or other safeguards for account holders.

The episode highlights a recurring tension in the rollout of generative AI features on consumer platforms - balancing creative utility against the rights and expectations of the people whose content and likenesses are involved. Meta has been expanding its AI presence across Instagram and its other apps, but this particular feature moved faster than its policy guardrails could support. How the company chooses to redesign or replace it will likely be watched closely by creators and platform regulators alike.