Meta's AI does not need your permission to use your face

Meta built a tool that turns your public Instagram into source photos for a stranger’s prompt, and it switched the tool on for you without asking. It is called Muse Image, it launched on 7 July, and it does something the old privacy settings were never designed to stop.

Here is the mechanic. You open the Meta AI app, type a prompt, and tag the @username of any public Instagram account. The model reaches into that account’s public photos, reads the person’s face, and drops them into whatever scene the prompt describes. The account owner is not asked. No notification lands on their phone. They find out if, and only if, they happen to see the result.

Every public account was opted in the day it launched. The switch that turns it off sits four menus deep, under a heading most people will never open. That was a choice. Meta could have asked first. It decided the default should be yes.

JB Branch, who runs federal AI policy at the consumer group Public Citizen, did not soften it: “Meta has once again chosen the creepiest possible path. People should not wake up to discover their face has become raw material for someone else’s AI experiment. Instead of asking for meaningful consent, Meta quietly defaults users into the system and buries the opt-out in account settings.”

You can switch off every setting on your own account and still be one friend’s public photo away from the model.

Your own account is the easy part. You can go and turn this off in two minutes, and you should. Open Instagram, Settings, Sharing and reuse, then switch off Posts and Reels under the line about reusing your content with AI features at Meta. Done. Except that toggle only governs the photos you posted. It does nothing about the photo your cousin put up from her wedding, or the group shot a friend shared from Saturday night, or the tagged picture your mum posted three years ago. Your face is sitting across accounts you do not control.

Their toggle decides your face, not yours. If the people who post pictures of you leave the setting on, and almost all of them will, because almost nobody has heard of this yet, your face inside their public photos stays fair game. You can lock your own profile down to nothing and still be pulled into an AI image through someone who loves you and never thought to check a menu. There is no clean fix for that. Privacy here is not an individual setting. It is a group project, and the group has not been told there is a project.

The talent agency CAA called on Meta to flip the whole thing to opt-in, arguing that “no one’s name, image, likeness, voice or creative work should be used by any third party, including AI models, without clear, documented consent.” Meta shipped the opposite.

So do the two-minute version for yourself today. Then do the harder thing. Send this to the people who post photos of you, your family, your close friends, the group chat that documents every dinner, and get them to switch it off too. One more thing worth knowing. Turning it off is not retroactive. Anything already generated before you flip the switch stays generated, and Meta does not delete it. The sooner the people around you turn it off, the smaller that window stays.

If you want the full opt-out walkthrough with the settings screens, Fast Company laid it out here. Fair warning, you will want to send it to your family before you finish reading it.