Trained to Imagine
Rendering
Creative Intelligence
AI Video / Image Models

Meta launches Muse Image, and users push back over use of their photos

Source: TechCrunch ·

Meta released Muse Image, a free AI image generator in the Meta AI app, Instagram Stories, and WhatsApp. One feature lets users tag anyone with a public Instagram profile and turn that person’s photos into new AI images; the owner is not notified and has to switch it off in settings.

TTI Takeaway

The consent argument has left the courts and reached the feed. Until now the consent fight was fought by rights societies and publishers — GEMA, the studios, the news outlets. Here it is about ordinary people’s faces, used by strangers, with opt-out as the default, and users objected on their own behalf. For anyone commissioning work, it is a reminder to ask early not just what a tool can produce, but whose likeness it can pull in without asking.

Analysis

Meta’s image model matters less for what it can draw than for where it lives. Muse Image, built by Meta Superintelligence Labs and code-named Mango internally, launched free inside the Meta AI app, Instagram Stories and WhatsApp. That puts a capable generator directly into the place where most people’s photographs of themselves already sit, and the backlash shows how quickly the consent question becomes personal once that happens.

The specific flashpoint, as TechCrunch reported it, was a feature that lets users take another Instagram user’s pictures and turn them into new AI images simply by tagging them, as long as that person’s profile is public. Meta’s policy states that people may be able to create content with your Instagram content using its AI features and that you will not be notified when they do. Meta says users have control and can disable this in settings, but the default is opt-out. TechCrunch placed the reaction in the context of Meta’s record, including the $5 billion FTC penalty in 2019 after Cambridge Analytica and the shutdown of Facebook’s facial-recognition system in 2021. Muse also offers prompt-based editing, custom ad creation and a Marketplace-linked interior-design use, is free for everyday creation with a subscription beyond a usage limit, and Meta says Muse Video is already in development.

For anyone producing campaigns, two issues follow. The first is likeness. Talent releases for social and influencer work were written for a world where a person’s public photos could be reposted, not regenerated into new scenes by strangers. Producers and talent representatives will reasonably want to know whether a performer’s account has the setting switched off, and whether a campaign that invites user participation could end up generating images of people who never agreed to appear in it. The second is advertiser exposure. Muse is pitched partly as an ad tool, and the October story about Meta’s commercial terms, which make advertisers solely responsible for anything its AI features create or change in their ads, including features on by default, shows where the liability is being placed. Brands using Meta’s native generation should treat its outputs as their own deliverables, with the same review and sign-off as anything a vendor hands over.

There is also a regulatory angle for European work. Since August, the EU AI Act requires deployers to disclose deepfakes, and a tool that makes realistic images of identifiable, tagged people sits close to that line. How Meta labels Muse outputs, and whether that labelling survives a repost, is not addressed in the reporting and is worth asking about before a European campaign relies on it.

The open questions mostly concern defaults and scale. An opt-out setting protects only those who know it exists, and the arrival of a video model would raise the stakes considerably: a still image of a tagged person is one thing, a moving, speaking clip is another. Our view is that the useful signal here is not the model’s quality but the reaction to it. Ordinary users objecting on their own behalf tends to move platform policy faster than court filings do, and the way Meta adjusts the feature will say a lot about how consumer-facing generation tools handle real people’s images from here.