As we continue our journey on the ethics of AI, I'm more concerned than ever about posting images of myself online. And quite frankly — you should be too.
On July 7, 2026, Meta launched Muse Image and quietly opted every public Instagram account into a system where anyone can @-mention your username in a prompt and generate new images of your face. Public profiles are available as references by default, unless users go change their reuse settings. Default. Not offered. Not asked. Default.
Now sit with what that actually enables. Someone types your handle into a prompt box. Out comes you ; at a rally you never attended, holding a product you'd never endorse, in a pose you never struck, wearing an expression you've never made. It doesn't have to be malicious to be devastating. It just has to be convincing. And your face is the thing that makes it convincing.
Here's what makes images fundamentally different from text. Train an AI on someone's writing and produce an essay, and you'd be hard-pressed to trace it back well-edited text leaves no fingerprints. The AI slop you had so much is just pure laziness. But faces don't work that way. A genetic ear shape. A signature smile. A birthmark at the corner of your jaw.
Same goes for artists. A visual style built over twenty years from the brush pressure, the color instinct, to the compositional habits you couldn't articulate if someone asked, that's a fingerprint too. You can't wipe it off and start clean tomorrow. Whether it's your face or your hand, the thing that makes it yours is exactly the thing the model is optimized to keep.
According to Variety the SAG-AFTRA urged all Instagram users, not just its members, to opt out and protect their likeness. When the actors' union sounds the alarm for everyone, that tells you how far this reaches.
"But we built it responsibly"
Sure. In a statement widely reported, Meta said it built the feature with controls and safety guardrails from day one, that private accounts under 18 are automatically excluded, and that adults with public accounts can opt out with easy-to-use control.
Read that middle clause again. Private accounts under 18. So a sixteen-year-old with a public account, which is to say, a sixteen-year-old who wants their art or their track times or their band to be findable, gets no automatic protection at all. Meta wrote the exclusion narrowly enough to sound protective and broadly enough to leave the kids most visible online exactly where they were.
As for "easy-to-use" — ha. PhotoWorkout found that the toggle sits four menu levels deep, and it only blocks future generations. Anything already made with your photos stays. The barn door closes after the horse has already been rendered in four art styles.
And you'll never know.
This is the fact I keep circling back to. WIRED noted it too: Meta's own help materials say users may not be notified when AI features create content using their Instagram content.
Not "notifications are coming." Not "we're evaluating." Just...... no.
That's not a privacy failure. A failure implies someone tried.
So where's the law?
Holland & Knight tracked a bill the Senate Judiciary Committee unanimously advanced called the NO FAKES Act on June 18, 2026. It's a bipartisan bill giving people control over AI-generated replicas of their voice and visual likeness. Unanimously! And yet a deepfake legislation tracker, StackCyber, notes it's still unclear whether it reaches a floor vote in either chamber. Total consensus, zero movement. Classic. I wonder whose lining their pockets.
Meanwhile, mePrism points out that US users have weaker legal footing than Europeans. Meta doesn't guarantee it will honor an objection at all. You can object; they can shrug. And Europeans aren't exactly safe either, since Malwarebytes notes Meta leans on GDPR's "legitimate interests" basis for AI processing. The land of the free is not so free anymore.
Which brings us to the money
Meta has been open about wanting AI-generated images of you flowing through the product itself. Business Insider reported the company began showing some users AI-generated images of themselves directly in their Facebook and Instagram feeds, describing it as a test of people's openness to seeing their likeness in posts they didn't create.
Your face isn't a byproduct here. It's inventory. Every promptable likeness is content that keeps someone scrolling, and scrolling is the entire revenue engine. The consent gap isn't sloppiness — it's the margin. Ask permission and adoption stalls; default everyone in and the supply is infinite and free.
And this is a pattern, not a Meta quirk. Malwarebytes also flagged that Twitch added an option to opt out of Amazon AI training two years after confirming that training had already begun. Two years. The playbook is consistent across the industry: ship first, disclose late, apologize never.
What you can actually do
Strongest option: switch your Instagram to private. Per the reporting, that keeps your photos out of the feature entirely.
If you need a public presence, like most of us, open Instagram, tap your profile, hit the three-line menu, find Sharing and Reuse, and toggle off both Posts and Reels. Meta says the option is rolling out gradually, so check back if you don't see it yet.
Then go check the settings on every teenager you're responsible for.
Do it today. Not because it undoes what's already out there. It doesn't. But because future you deserves a vote.
What do you think? At this point I'm not sure we can be surprised by anything anymore.
But maybe that's exactly what they're counting on.
This has a been A Geeky Production
