Meta kills Muse Image's Instagram feature after three days
Meta pulled Muse Image's Instagram feature after a privacy backlash, hours after a Reuters test found its detector missed 55% of cropped images.
Meta launched Muse Image on Tuesday and killed its Instagram feature on Friday. The tool could generate images using photos from public Instagram accounts, with users opted in automatically. SAG-AFTRA told members to turn it off, actor Hannah Einbinder amplified the warning, and Meta pulled the feature after admitting it had “missed the mark.”
The privacy default was bad enough. The safety story underneath it was worse. In a Reuters test of 40 Muse Image outputs, Meta’s detector verified every original, then failed on 55% after the images were cropped to roughly one-third to one-half of their original size. Meta says its invisible Content Seal watermark is designed to survive common edits, including cropping, but acknowledged that a heavy crop can erase the signal.
This is what shipping the safety system as a preview looks like when the product attached to it can reach into public Instagram photos. The opt-out setting is gone now, but the useful lesson survived: a provenance claim that disappears with the edges of the picture is not much of a provenance claim.