The pushback worked. Only four days after Meta quietly launched a tool that let any user tag a public Instagram account and generate AI images from the person’s photos, the company pulled it. The feature, part of Meta’s new Muse Image generator, was gone. “We’ve heard the feedback that this feature missed the mark, so it’s no longer available,” the company said Friday in a statement that was notably short on specifics about what exactly the mark was or how badly they missed it. 

But hey, we should all take the win. It’s rare that a consent-by-default gets reversed in under a week. But understanding why it got reversed matters more than celebrating it. 

Let’s look at what actually happened. On July 7th, Meta launched Muse Images with a feature that allowed users to generate AI images by @-mentioning any public Instagram account. The photos and videos from that account would be pulled as reference material for the generation. The feature was automatically enabled on all public accounts. To opt out, users had to actively find the setting. That meant that participation was by default, and it was up to the user to find the right setting to opt out. 

The backlash was immediate once the public realized what Meta had done. SAG-AFTRA told members to protect their likeness and opt out, stating plainly that “a feature that encouraged nonconsensual digital replicas is unwise.” CAA, one of Hollywood’s most powerful talent agencies, pushed back hard as well, arguing that no one’s name, image, likeness, or voice should be used without clear, documented consent. Even casual Instagram users were vocally against the new feature, and Meta quickly folded. 

The lesson here is uncomfortable but important: it really is that deep. Adult content creators have been sounding this exact alarm about consent and stolen likeness for years. The technology to scrape, replicate, and distribute someone’s appearance without permission has existed for years. The harm of nonconsensual deepfakes, stolen identities, and AI-generated content using real people’s faces without their knowledge has been extensively documented. And when it was just adult creators, the response was silent. Platforms moderated the worst outputs inconsistently and called it solved. 

What changed last week wasn’t the validity of the concern; it was who voiced it. Once SAG-AFTRA and the CAA spoke up, Meta moved fast. After all, their clients’ faces are worth money to powerful people. Even though the consent argument is identical to what adult creators have been making, the platform’s response was not. 

It’s important to see how content policy actually works. Platforms respond to liability, and liability is calibrated by who is complaining. A major talent agency threatening legal action on behalf of A-list clients is a different kind of pressure than an OnlyFans creator reporting that her likeness has been stolen and replicated, even though the result is the same. 

But just because Meta rolled back this feature doesn’t mean the problem is solved. Anyone can still take any photo from the internet and run it through a generative AI tool. Removing it made it harder for people to do it, but there’s nothing being done to stop the problem. 

This isn’t the first time that a major AI company floated a feature like this and had to walk it back after being caught. It’s clear this isn’t an accident. These companies are testing how much we are willing to take. 

The statement from Meta about “missing the mark” shows that they don’t really care. They weren’t apologizing for not offering consent for this feature or what they would do with the data already collected. It was simply them trying to save face. 

These kinds of tests aren’t going away. And everyone, from adult content creators to A-list actors to casual social media users, needs to get loud with their arguments. Keep opting out of every setting that defaults to yes, and keep your guard up.