Privacy advocates are raising significant concerns following the rollout of Meta’s new Muse Image AI tool, which allows the software to generate images by drawing from the faces and content of public Instagram profiles by default. Released on Tuesday, the tool enables users to tag public accounts and pull source material from those profiles without notifying the users involved. Meta describes this as its most advanced image generation model to date.
Cybersecurity firm Malwarebytes highlighted the issue in a July 9 blog post, suggesting that the most effective way for users to protect their images is to switch their Instagram accounts to private. This, they argue, prevents strangers from utilizing personal profile photos as source material. For those who prefer to keep their accounts public, Meta provides an opt-out mechanism located in the “sharing and reuse” section of the app’s settings, where users can toggle off the permission for others to reuse their content.
Critics, including those from the privacy-focused company Proton, argue that the opt-out process is intentionally difficult to navigate. They contend that data sharing is enabled by default and that the settings are buried, often leading users to discover the feature only after public backlash occurs. Proton further warned that the on and off toggles appear nearly identical, making it easy for users to leave the feature active by mistake. Thorin Klosowski, a senior security and privacy activist at the Electronic Frontier Foundation, stated that this functionality should have been an opt-in feature from the start, as it represents a new use of photos that users never anticipated when they originally joined the platform.
While Meta claims the tool was built with strong safety guardrails, concerns remain regarding the protection of minors. The company noted that Muse Image users cannot tag accounts belonging to individuals under 18, and teens themselves cannot use the tool to reference other accounts. However, Meta has not clarified whether children featured in the public photos of adult users could still be processed by the AI. Proton warned that there is a risk of children’s faces being appropriated in this manner.
A spokesperson for Meta defended the tool, stating that it includes robust protections to block the generation of content that violates community standards. The company emphasized that adult users with public accounts can easily opt out and that they will take action against any content that breaches their policies. Users who encounter objectionable AI-generated images can report them by holding down on the image and selecting the “thumbs down” option.
Meta’s own promotional materials for Muse Image highlight its ability to compose elements from multiple input references, such as researching “the hottest women’s summer 2026 looks,” scanning shoppable links for linen dresses, and dressing the user up in them. While the company frames these as creative features, privacy advocates maintain that the ability for strangers to leverage personal accounts for AI generation poses a significant privacy risk that users are largely unaware of.
Privacy advocates argue that expecting users to proactively turn off this feature is unfair – and that “finding the setting is its own adventure”, as Malwarebytes writer Danny Bradbury describes.
Another example displays the image generator’s more complex capabilities: combining multiple images – such as a particular person, a particular make of a bike, and a visual style to emulate – in a single prompt, to customize the output with a granular level of specificity.
