Meta's Muse Image AI tool sparks privacy backlash

Meta released its Muse Image AI tool on Tuesday, allowing users to reference public Instagram profiles in prompts to generate new pictures. The system pulls the faces of people featured in those social media posts to customize the output. Users whose images are integrated into what Meta calls its “most advanced image generation model yet” receive no notification.

The feature activates by default for adult public accounts. While switching an account to private is the most comprehensive fix, public users who want to maintain their visibility must navigate to the “sharing and reuse” section of their settings to toggle off content reuse. Cybersecurity firm Malwarebytes noted that "finding the setting is its own adventure", and Proton warned that the toggle states look identical, making it "easy to leave one active by mistake".

This default-on approach to data sharing poses a distinct regulatory risk in Europe, where strict privacy laws generally require explicit opt-in consent before companies can repurpose historical data for new technologies. Proton criticised the design, noting that "public backlash becomes the main way users find out what happened to their content". Thorin Klosowski, a senior security and privacy activist at the Electronic Frontier Foundation, echoed this, stating: "This is the sort of setting that should absolutely be opt-in for Instagram users." He added that this is "a new use of the photos they’ve been posting publicly for years, and certainly wasn’t on anyone’s mind when they signed up for Instagram years ago."

Meta is pitching the tool as a commercial asset, demonstrating its ability to scan "shoppable links for linen dresses" or combine specific people, bikes, and visual styles into a single image. A Meta spokesperson stated the tool was built "with strong controls and safety guardrails from day one", automatically excluding private accounts and users under 18. The company noted it would "take action against any content that violates [their] community standards".

However, the company has not clarified if children depicted in photos on public adult accounts are protected from being used in prompts. Proton warned that minors featured in public photos risk having their faces appropriated. Malwarebytes highlighted the broader business vulnerability: the technology is appealing when applied to yourself, but "it’s less appealing when anyone else can do the same with your account." As Meta prepares to expand the generator to Facebook and introduce video capabilities, its aggressive push to monetise AI continues to test European privacy expectations.