Meta's first 'superintelligence' lasted three days. The company is explaining why.

Meta's first product from its Superintelligence Labs survived less than 72 hours in public. The company pulled the 30-filter Instagram image editor, branded "Muse Image," on July 11, three days after launch, after backlash over a default-on feature that let users generate AI images tagging public Instagram accounts without permission.

"Superintelligence" is Meta's marketing label for the frontier-AI group it has been hiring and acquiring around, not a claim that the system itself reasons at human level. The first consumer product from that lab was a feature, not a model launch in the conventional sense: a set of lighting and composition filters plus an @-mention tool for AI-generated birthday cards, group-trip memes, and similar images that could pull in strangers' Instagram handles by default.

SAG-AFTRA publicly condemned the default-on behavior, calling anything less than a clear opt-in "unacceptable" and "an utter miscalculation of public sentiment," per The Register's write-up. Meta's pull statement confirmed the framing: "We've heard the feedback that this feature missed the mark, so it's no longer available." Several of the filters were credited to Instagram creators whose aesthetics Meta had used to amplify the product, a laundering move that turned the rollback into a creator-relations story as well as a consent story.

The mechanism is the news, not the model's capability. The collapse was a consent design failure wearing frontier-AI branding, and the rollback statement is one of the few public postmortems the lab has issued.