Particle.news

Meta Pulls Instagram Tagging in Muse Image After Consent and Detection Failures

Meta says Muse Image outputs carry an invisible Content Seal watermark that its preview detector can fail to read after heavy cropping.

Overview

  • Earlier this month Meta launched Muse Image with a feature that let users reference any public Instagram account by @-mention and effectively opted public profiles into being used.
  • Within about 72 hours the company removed the Instagram account-tagging option but kept the Muse Image model and other Meta AI features available.
  • Meta says every Muse Image output includes an invisible Content Seal watermark and also told Reuters the preview detector may lose the watermark signal after heavy cropping, while a Reuters analysis found the detector failed to verify roughly 55% of cropped images.
  • Creators, unions such as SAG-AFTRA, talent agencies and rights groups criticized the buried opt-out, saying technical measures cannot fix harms that come from nonconsensual reuse and loss of context for everyday photos.
  • The episode raises clear regulatory and reputational risks for Meta, highlights technical limits of watermark-based detection, and underscores calls for consent-first product design and stronger policy or legal remedies.