Meta Backs Down After Trying to Turn Your Instagram Feed Into an AI Playground
Ingreen / Shutterstock
Toggle Dark Mode
In a remarkably quick about-face, Meta is removing a highly controversial aspect of the AI image generation feature it announced less than a week ago, citing what we can only guess was overwhelmingly negative feedback from users.
The feature in question was part of Muse Image, a new image generation model designed to let Meta AI users create “high-quality visuals that [they] can download and share anywhere, including directly to [their] feed, story, or chat.”
On the surface, Muse Image seemed harmless enough — at least certainly no more so than similar tools offered by OpenAI and Google. However, Meta also snuck a disturbing little twist into the mix by turning every Instagram user’s public photos into fodder for Muse Image to use at will.
As potentially creepy as this sounds, it was actually much worse. The feature included every user’s public Instagram photos unless they explicitly opted out, and it wouldn’t even notify users if their content was used by Meta AI.
To be excruciatingly fair, it was never entirely clear how pervasive this image usage would be. When Meta shared the details on Muse Image, it implied that your content would only be used if something “@-mentioned” you by name in the Meta AI app. I’ll grant that’s a bit better than Meta’s chatbot randomly collecting photos from across people’s Instagram profiles, but it’s still an incredibly low bar, as you could still be drawn into some random person’s AI creations with no more than your Insta handle.
While it was possible to opt out of this, the process required a deep trip into the app’s settings and only worked if you knew about it in the first place. Other than Meta’s blog post — and all of the secondary coverage that resulted from its rather invasive decision — nothing on Instagram alerted users to the fact that their public photos were suddenly fair game for Meta AI.
This feature has understandably been less-than-well-received. While we’ll probably never know exactly how much feedback Meta got about it, we can assume it was enough, as the company removed that particular feature from Muse Image on Friday afternoon, three days after it was first unveiled.
The news first broke via a statement to Puck’s Dylan Byers, which was later added as an update to the original Muse Image blog post on Friday at 3:45 p.m. PT:
Earlier this week, we announced that one way for people to generate images in Meta AI is by @-mentioning public Instagram accounts that they want to reference. 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. We’ve heard the feedback that this feature missed the mark, so it’s no longer available.
Sadly, it likely wasn’t merely the hue and cry from everyday users in the trenches that got Meta’s attention. I’m not sure the company cares enough about that kind of feedback to reverse course at all — much less this quickly. However, when the Screen Actors Guild has to recommend its members opt out and public safety groups like the National Center on Sexual Exploitation are making calls to “REVERSE COURSE” in all caps because of the risks of an AI tool being abused, companies do tend to wake up and smell the coffee.
It also doesn’t help that we’ve already seen how egregiously AI can be misused. For the better part of January, Elon Musk’s Grok happily helped deviants create non-consensual explicit deepfakes by doing little more than similarly @-mentioning users on X and describing all sorts of compromising situations. We can only hope that Meta AI has slightly better guardrails than that, but completely preventing generative AI image tools from coloring outside the lines is more of an art than a science. The best way to avoid these types of controversies is not to play in those sandboxes at all.
