Meta has launched Muse Image, an AI image generator developed by Superintelligence Labs, and quickly integrated it across Meta AI, Instagram Stories, and WhatsApp. The tool offers standard features: text-to-image generation, style transfer, and preset filters. But the stir isn’t about new AI tricks. It’s about how Muse Image uses photos from public Instagram profiles for AI-generated pictures without asking the original owners for permission.
Previously known internally as Mango, Muse Image is now available free for basic use, with a subscription required after hitting usage limits. Beyond creating images from text prompts, it lets users edit existing photos using text commands, create marketing visuals, and simulate how items might look in different interiors. One demo shows a used couch in a garage, and Meta plans to extend this functionality to Facebook Marketplace soon.
The most controversial aspect is simple: if an Instagram account is public, any other user can tag the photo’s owner and use that image to generate new AI pictures within Meta’s tools-no separate approval or notification is required. Users can opt out in settings, but the feature is enabled by default for public accounts. This ”opt-out after the fact” approach is at the heart of the backlash.
Meta is also rolling out new AI effects powered by Muse for Instagram Stories. Rather than a standalone AI tool for prompt geeks, this brings image generation and editing right alongside familiar filters. This move keeps Meta competitive with OpenAI, Google, and Adobe, whose AI generation is already baked into chatbots, editors, and productivity apps. Meta clearly wants to avoid falling behind in AI image generation technology.
Muse Image’s access to Instagram photos and user consent
The problem runs deeper than just a new filter button. Meta is once again navigating treacherous ground where user content feeds automated systems, with consent buried in settings. This issue dates back to the Cambridge Analytica scandal, when Facebook data was harvested for political profiling. Meta later paid a record $5 billion fine to the FTC in 2019.
In 2021, Meta disabled facial recognition on Facebook after lawsuits and regulatory pressure over how biometric data was collected and stored. Then it was about face templates; now it’s about AI-generated images from regular photos. The core complaint remains: Meta grabs sensitive data and only offers users a way to opt out after the feature is already live.
Other companies are taking different approaches to AI training data. Adobe’s Firefly AI is trained on licensed Adobe Stock content and public domain materials. Meanwhile, Stability AI, Midjourney, and DeviantArt have faced lawsuits from artists, while Getty Images sued Stability AI in both the US and UK. With Muse Image, Meta isn’t just defending its training data practices-it’s also under fire for how the finished AI products use real people’s photographs right now.
This isn’t just a local controversy. Bloomberg Intelligence estimates the generative AI market could hit $1.3 trillion by 2032. Companies are rushing to embed AI into every mass-market app as soon as possible. But consent protocols, user notifications, and control over how people’s photos are used tend to lag behind-and sometimes by years.
Meta, which owns Instagram and Facebook, is banned in Russia as an extremist organization. The future of Muse Image won’t hinge on the quality of its pictures but on regulatory and user pushback against this ”on by default” model. Without simplifying opt-outs and adding clear notifications, Meta risks a familiar scenario: a flashy launch followed by drawn-out privacy controversies far from the product spotlight.
* Meta owns the services mentioned here; it is banned as an extremist organization in Russia.