July 8, 2026

AI Incider

AI News. No Noise. Just Signal.

Meta AI Uses Your Instagram Photos Unless You Opt Out

2 min read
Meta AI can now turn public Instagram photos into AI images by default. Here is why the opt-out model alarms privacy experts, and how to opt out.

Your public Instagram photos may now be raw material for anyone’s AI art project, and Meta turned that on for you automatically. This week the company began letting users @-mention any public Instagram account in a prompt and generate new images built from that person’s real face, no permission required.

What Meta Changed

The feature rides on Muse Image, the first in-house image model from Meta Superintelligence Labs, launched July 7. Type a prompt, tag a public profile, and Meta AI pulls photos from that account to use as a visual reference, producing fresh images that carry the person’s likeness. Unlike Midjourney or DALL-E, which make you upload a photo first, Muse reaches straight into Instagram’s vast library.

The Meta AI Opt-Out Problem

Here is the part privacy experts object to: every public account is opted in by default. Meta sends no notification, asks for no consent, and adds no approval step. Someone could be generating dozens of AI images of your face right now, and you would have no way to know. Turning the setting off only stops future generations. Anything already made stays made.

Opting out is buried in settings. On your Instagram profile, open the menu, find “Sharing and reuse,” and switch off both Posts and Reels under the option to allow your content to be used with AI features on Meta.

Opt-Out vs Opt-In

The design choice matters. Meta uses an opt-out model, which quietly enrolls the largest possible pool of faces and puts the burden on users to escape it. Contrast that with OpenAI’s approach to likeness in Sora, where consent is a hard stop. Its Cameo feature requires you to record and approve your own likeness before anyone can use it, and you can revoke access at any time. In other words, OpenAI asks first, and Meta assumes yes.

That gap is the whole story. Both companies want your face to train and fuel their image tools, but only one makes you say no after the fact. As AI likeness features spread, expect regulators to focus less on what these systems can do and more on whether people agreed to be part of them in the first place.

Continue Reading…

Leave a Reply