Meta has pulled an AI image-generation feature from Instagram days after launching it, following backlash over its use of public Instagram accounts as raw material for AI-altered images. The company said it "missed the mark, so it's no longer available." BBC

The feature, part of Meta AI's Muse Image tool, let users tag public-facing Instagram accounts and pull content from those profiles to generate or alter images. Meta had framed the tool as a way for users to "create images with friends and reimagine their Instagram photos," per its own announcement three days earlier. Meta AI The company says its intent was to build a useful creative tool while giving people control over whether their public content could be referenced by the model — control that, in practice, appears to have been contested by users whose accounts were tagged without their direct involvement in the decision.

Muse Image is Meta's most advanced image-generation model to date, built to follow instructions faithfully, edit with precision, and compose scenes from multiple source elements. Meta AI It sits alongside Muse Video, a companion video-generation model Meta unveiled in the same announcement, and draws on techniques Meta first previewed with its Emu Edit research model in 2023 — a generative system for instruction-based image editing that underpins much of the company's current Instagram AI tooling. Meta AI

The tagging mechanism is the specific element Meta withdrew. Instagram already offers an AI "restyling" feature in Stories that lets users transform their own photos and videos, and Meta AI separately runs a "Vibes" feature for editing images or animating video, accessible via meta.ai with a Meta, Facebook, or Instagram login. Meta Neither of those tools involved pulling content from other users' accounts by tag — the distinguishing, and evidently more contentious, design choice in the now-withdrawn feature.

The rollback illustrates a recurring tension in how platforms deploy generative AI against user-generated content libraries. Instagram's entire product is built on public and semi-public sharing, which makes it a rich corpus for training and prompting image models — but the same public-by-default architecture means any feature that lets one user invoke another's likeness or photos raises consent questions that a simple "your account is public" framing doesn't fully answer. Meta's own language about giving users "control over whether their public content could be referenced" suggests the company anticipated this friction, even if the mechanism it shipped didn't hold up under real-world use.

At launch, the tagging feature was limited to Instagram, but Meta has signaled broader ambitions: further AI features and integrations are planned for WhatsApp, Facebook, and Messenger, and Meta is separately developing an AI video tool for wider deployment. BBC That roadmap means the underlying questions raised by this withdrawal — about consent, tagging, and the boundary between a person's public presence and their exploitable data — are likely to resurface as Meta AI expands across its messaging and social products rather than being resolved by pulling one Instagram feature.

Meta has not said whether a revised version of the tagging capability is planned, nor detailed what specifically drove the backlash beyond its brief acknowledgment. The episode sits within a broader pattern across the industry of AI features shipping quickly and being retracted or modified once users encounter edge cases the design process didn't anticipate — a cycle that tests how much scrutiny companies apply before release versus after public reaction forces a reversal.