Meta’s Muse Image tool was designed to turn Instagram into a generative AI playground, allowing people to create new images using photos from any public account.
The core problem was consent: adult users with public profiles were opted in automatically, with no upfront notice or explicit choice. Anyone could be remixed into AI art by strangers simply because their account wasn’t private. Reports showed Muse could generate images of people who had never interacted with the tool at all, including photos featuring children who obviously couldn’t consent to such reuse. To make matters worse, Meta’s own policy confirmed users would not be notified when their content was used in AI features, keeping the whole process largely invisible.
Creators, unions and privacy advocates quickly denounced the opt‑out model as an inversion of basic digital rights. Hollywood unions and talent agencies warned that Muse normalised non‑consensual manipulation of someone’s image and could undermine control over professional likeness and copyrighted work. Digital rights groups called the rollout a “privacy landmine”, pointing to existing harms from deepfakes and non‑consensual AI imagery elsewhere on the internet. Their argument was simple: protection should be the default, and any AI reuse of identity should require explicit, informed opt‑in.
Under pressure, Meta stressed that private accounts and users under 18 were automatically excluded from Muse, and that any public user could disable the feature with a few taps in Instagram’s Sharing and Reuse settings. Users could also flip their profile to private to lock themselves out of AI remixes entirely. But critics noted these controls were buried, easy to miss and did nothing to remove AI images already generated from someone’s posts. For many, this reinforced the sense that meaningful control arrived only after the data had already been exploited.
Within days of launch, the backlash forced Meta to pause and then remove the Instagram implementation of Muse Image, admitting the feature “missed the mark” on user expectations. The episode has become a case study in how not to roll out AI features on social platforms, especially when they touch identity and consent. It underscores a wider shift in user sentiment: AI creativity is welcome, but only when people remain clearly informed, empowered and in control of how their content trains or feeds the machine.
Read the original article: