MENLO PARK – The photos you shared publicly on Instagram can now be ingredients in someone else’s AI-generated image, and Meta does not need to tell you first.
Meta launched Muse Image on Thursday, a feature built into its Meta AI assistant that lets a user type a prompt, reference a public Instagram account with an @-tag, and receive a generated image incorporating the visual style and content from that account’s public posts. The feature comes from Meta Superintelligence Labs, the AI research division the company consolidated in June, and is rolling out first to Instagram users before extending to Facebook, Messenger, and WhatsApp.
What Meta calls “bringing creative visions to life” is what SAG-AFTRA called an emergency on the same day of the launch. “Take Action To Protect Your Likeness,” the union wrote Thursday, instructing its members to opt out before Muse Image began processing their images without explicit consent. The union’s response arrived within hours of Meta’s announcement, illustrating how quickly likeness rights have become a frontline labor issue in the AI product cycle.
The technical architecture is straightforward. A user enters a prompt in Meta AI, mentions any public Instagram account using its @handle, and the system draws on that account’s visible photos to generate a custom image in a matching aesthetic. Meta also offers 30 preset prompts, packaged creative scenarios that can be applied to any @-mentioned account with minimal text input. The company frames the feature as a tool for fans, brands, and photographers who want to reference a particular visual style.
What the feature is not is opt-in. Any Instagram user who maintains a public account becomes a default participant in Muse Image. An opt-out mechanism exists, but it requires navigating to account settings and finding a control that Meta has not featured prominently in its public launch materials. For the large majority of users who will not encounter SAG-AFTRA’s alert or read technology press coverage, their photos are available by default.
That distinction – opt-out rather than opt-in – is the central dispute. The EU found that Meta breached the Digital Services Act over addictive design features on Facebook and Instagram, a ruling that arrived just days before the Muse Image launch. Regulators across Europe and in several US states are drafting rules addressing AI-generated likenesses, but no comprehensive federal standard exists in the United States. Meta’s Muse Image arrives into a legal environment where the rules are still being written.
CBS News reported Thursday that “Muse allows other Instagram users to use photos from public accounts to make AI images without explicit approval or knowledge of the account owner.” Meta has not responded with clarification beyond its launch materials. The company relies on existing terms of service, updated in recent years to encompass AI-related rights, to cover the practice. Users who agreed to those terms as a condition of platform access did not necessarily understand they were consenting to serve as source material for AI image generation tools available to any other user.
Privacy and digital rights organizations have argued for years that AI companies employ consent frameworks that assume participation unless a user actively refuses. The burden falls on the person whose content is being used, while the commercial benefit accrues to the platform. Meta’s approach to Muse Image follows that pattern precisely.
The scope of the feature is broader than it might initially appear. Muse Image is not limited to verified creators or recognisable accounts. It works with any public Instagram account: a teenager’s food photography, a nurse’s travel pictures, a local architect’s project portfolio. The @-mention system requires a user to know or find the target account’s handle, but Instagram is designed for discoverability. Most public accounts are reachable with minimal effort. The person whose images are being incorporated receives no notification that their content is being used.
Meta Superintelligence Labs, formed by pulling researchers from across the company’s AI divisions, has moved quickly on consumer-facing generative tools that leverage the volume of content already hosted on Meta’s platforms. Instagram alone holds years of user-generated imagery at scale. Meta has previously restricted off-site data controls in moves critics characterised as weakening user privacy protections ahead of expanded AI deployment. Muse Image continues that trajectory.
What Muse Image produces at scale – in content and in regulatory response – remains to be seen. The feature is still rolling out; the effects on creators with large followings are visible, but the effects on private individuals with smaller public accounts, who never anticipated commercial use of their images, will accumulate quietly. The burden of saying no belongs to the user. The value of saying yes belongs to Meta.
