Meta dropped a new AI tool today. It makes pictures. Real looking ones. If your Instagram account is public, you’re basically volunteering to be the subject. Fix it. Now.
The model is called Muse Image. It’s part of a family Meta calls Muse Spark, cooked up in their superintelligence labs. Meta’s blog claims it handles complex requests, stitches together composite photos, and edits existing stuff. You can try it right now in the Meta AI app, on Instagram, or WhatsApp. They plan to roll it out to Facebook, Messenger, and eventually advertisers too.
Mark Zuckerberg showed it off on his own feed Tuesday. He posted about thirty new AI effects for Instagram Stories. We got Zuckerberg clones. A 360 view featuring AI head Alexandr Wang. An exposure portrait of CTO Andrew Bosworth.
It’s not a new trick. OpenAI tried this with Sora last year, putting users into video scenery. People hated it. Celebrities hated it. It made deepfakes too easy. Meta’s tool brings the exact same problem. Same risk, different company.
Forget the AI slop for a second. Ignore that you’re training their systems every time you upload. Look at the settings.
If you are over eighteen and your account is public, anyone can “tag” you in a prompt. They can generate hyperrealistic images of your face. Deepfakes, effectively. Just by typing your name.
How to block it
I tested it. Wanted to see how easy it would be.
I grabbed the handle of a CNET colleague, Abrar Al-Heeti. Public account. I typed a prompt adding her username. One minute later, she was a pirate. Detailed. Convincing. Scary easy. I tried the same with my private account. Meta AI choked. Request failed. Privacy matters, apparently.
So, how do you stop it?
“Allow people to reuse your content…” is the toggle you want to kill.
Go to Instagram Settings. Then Sharing and reuse. Toggle off “Allow people to reuse your content on this account and with AI features at Meta.” You can tweak this for posts and reels separately.
If your account is private? You’re safe by default. No remixing. No AI generation from your pics. My colleague turned hers off immediately after we tested it.
There’s another layer, though. When you first let the app recognize you in Meta AI, it asks for permission. It might want a selfie or three uploaded photos. This is where you set boundaries. You can restrict who gets to use your likeness: only you, approved followers, mutuals, or everyone. Check Settings > Your likeness in the app to lock that down.
They make it sound harmless. A fun new effect. But once the gate is open, who decides what comes through? 🏴☠️
