Meta says it pulled Muse Image because the backlash got too loud to ignore. It never said what happens to the photos the tool already touched.
Instagram launched Muse Image on Tuesday, July 7. By Friday, July 10, it was gone. Three days is not a long life for a feature Meta had spent months building, but it was long enough for a huge slice of public accounts to become raw material for an AI model that never asked permission first.
How Muse Image Worked, and Why It Backfired
Here is how it worked while it was live. If your Instagram account was public and you were over 18, anyone could type your username into a Muse Image prompt and Meta AI would pull your posted photos as a visual reference to generate something new: a different outfit, a different scene, your face grafted onto a moment you never posed for. You were opted in by default. Turning it off meant finding a setting most users never knew existed, according to reporting from TechCrunch and Forbes.
That default is what set off the backlash. SAG-AFTRA, the union representing film and TV performers, told its members to change their settings immediately and publicly criticized the rollout. CAA, one of Hollywood's largest talent agencies, went further and raised the issue directly with Meta, according to Variety. Privacy advocates piled on too, arguing that an opt-out model puts the burden on the person being copied rather than on the company doing the copying.
Meta folded fast. "Our intent was to provide a useful creative tool and to give people control over whether their public content could be referenced in this way," the company said in a statement reported by Newsweek. "We've heard the feedback that this feature missed the mark, so it's no longer available."
The Photos Left Behind
Pulling a feature from an app is not the same as deleting what it built while it ran. Reporting from PhotoWorkout found that opting out only stops future image generations. It does nothing to the images Muse Image already made using your photos before you flipped the switch, and Meta has made no commitment to retroactively delete them. The button vanished from your feed. The model that learned from your face over those three days did not forget just because you flipped a switch.
That's the real story here, and it's quieter than a splashy launch and an even splashier product death. Meta's servers processed a huge volume of public Instagram accounts as training material for a generative model, and nothing in the company's public statement says that material has been purged. The tool is off. The data pipeline that fed it is a separate question, and Meta hasn't answered it.
This is not a new pattern for the company. Meta has spent the past two years training its AI systems on public Facebook and Instagram posts across the European Union and beyond, drawing regulatory scrutiny nearly every time. Muse Image was the most visible version yet of a habit Meta has already normalized: treat public posts as free training data first, and treat privacy as a settings toggle to bolt on once people notice.
Frankly, the speed of the reversal should worry you more than it reassures you. A company that ships an opt-out default on a facial-generation tool, then reverses course only after a talent agency calls, was never asking whether this should exist. It was asking whether anyone would notice before launch. This time, they noticed in three days.
What Regulators Do Next
What happens next depends on regulators as much as on Meta. The Federal Trade Commission and privacy authorities across the EU have both taken an interest in how platforms train generative models on user content without explicit consent, and Muse Image is exactly the kind of case that invites a formal inquiry. Until one lands, the honest answer to what happened to the photos already pulled into Meta's systems is that nobody outside the company knows, and Meta isn't volunteering it.
Also read: The UN's AI Summit Showed Robots Racing While Governments Argue Over Rules • Researchers hacked a quantum neural network on real trapped-ion hardware and gutted its accuracy • Sam Altman Calls Elon Musk Homeboy in Their Latest Fight Over AI Infrastructure
