Meta shuts down the Instagram feature that used to let users create AI deepfakes of public accounts.

• Meta has shut down the feature that let people generate AI deepfakes parodying public figures on Instagram.
• The main reason is the backlash over rights violations and the risk of misuse, especially involving celebrities who never consented.
• This directly affects creators and public accounts who used the feature for joke content, as well as people who had their faces turned into deepfakes without knowing it.
• AI tool developers now need to go back and tighten up their guardrails before rolling out a feature like this again.

Note: We don’t have confirmed details on the exact date the feature was shut down. If you need to cite this, check Meta’s official announcement directly.

What the feature looked like right before it got the axe

The feature let users take the face of a celebrity or public account and turn it into a parody AI clip easily, right through the app’s interface — no external tools required.

The problem was that the system was too wide open. Anyone could grab someone else’s face and turn it into content without that person’s consent, and often without them even knowing they’d been used.

Meta has now shut this feature down, so what’s shown in screenshots is now evidence of something that actually worked in production — not an unreleased concept.

The day you see yourself saying something you never said

A colleague on our team once ran into a clip of themselves saying things they’d never actually said, surfacing in their feed from a public account they didn’t even recognize. The face, the voice, the lip movements — realistic enough that they had to scroll back and watch it twice.

The problem is there’s no way to know who grabbed the photo to make it, or how many people had already seen the clip before the person found out. What they felt in that moment wasn’t just anger — it was a loss of confidence in whether their own online identity still belonged to them.

That’s exactly why Meta had to make the call to shut this feature down. Even if it was intended as a fun content-creation tool, once it was opened up for anyone to grab a public account’s face and use it freely, the consent risk grew beyond anything they could control.

Where this feature sat in Meta’s roadmap

This deepfake feature was part of the generative AI push Meta has been driving all year — from the Meta AI chatbot to AI characters that let users build chat personas on Instagram/Facebook themselves. The idea was to make everyone an “AI content creator” as easily as possible, with no editing skills required.

The problem is this feature touched something different from Meta’s other AI tools: it used the “real face” of a public account as raw material, not a fictional character or a generic prompt. Given that, the faster Meta pushed adoption, the faster the consent gap widened. What was supposed to showcase Meta’s AI capabilities ended up opening the door to questions about accountability instead.

Before the shutdown vs. after — what changed

Before this, anyone with a public account could have their face turned into an AI deepfake through this feature by someone else, with no need to ask the account owner’s permission first. Simply having a public account was enough to qualify — there was no opt-in system letting account owners decline from the start.

After the shutdown, what’s left are generic AI editing tools that aren’t specifically tied to “a real face pulled from a public account.” Anyone wanting to make this kind of content has to fall back on older methods or other apps instead — which involves considerably more friction than before.

When this happens in real life

Picture a famous singer suddenly showing up in a clip promoting a cryptocurrency they never actually endorsed, circulating in a family group chat — the source being a face pulled from a public Instagram account that anyone could turn into a clip.

Or a product-review creator suddenly finding a fake clip of themselves pretending to review a competitor’s brand, forcing the brand they actually work with to spend time verifying which one is real.

Brands themselves have their own headache too — before paying an influencer, they now have to double-check whether a clip circulating in the market is genuine or an AI fake.

And for regular users like us, we’ve all shared a funny clip in a friend group chat without ever realizing it was a deepfake made through this feature — risking becoming an unwitting spreader of fake content.

What other platforms are saying about this

Now that Meta has shut this feature down, a lot of people want to know whether other platforms are stricter or looser about the same issue.

TikTok and YouTube already have policies requiring AI-generated content to be labeled, but they haven’t gone as far as shutting down a public-figure deepfake feature the way Meta just did. X (formerly Twitter) is even looser — there’s still no clear measure in place for controlling fake images of public figures.

Honestly, each platform is mostly moving in response to social pressure rather than any shared common standard. Users end up having to figure out for themselves how much they can trust each platform.

What’s gained and what’s lost from shutting this down

The clear win is that the risk of fake news and celebrity impersonation drops, especially for public accounts that previously had their faces used in fake clips without their knowledge. That kind of thing can seriously damage reputation and credibility — cutting it off at the source is a lot easier than trying to clean it up after the fact.

The loss is that some creative tools people used to make lighthearted parody content or test out new ideas disappeared along with it. Creators who used the feature responsibly, without deceiving anyone, got caught in the crossfire too — a handful of bad actors ended up shutting down a genuinely useful tool for everyone.

The extra cost Meta and users now have to pay

Shutting this feature down doesn’t mean the problem is over. Meta’s moderation team still has to chase down deepfakes leaking through other channels, since anyone determined to fake a celebrity’s face always has a workaround — whether through third-party apps or pre-made uploads.

Another cost that doesn’t show up in any metric is the trust that was already lost before the shutdown. Users who’d already had fake clips of themselves go viral are protected from new incidents going forward, but the shutdown doesn’t erase the old ones still circulating on other platforms.

The heaviest burden falls on regular users, who now have to judge for themselves which clips are real and which are fake. There’s no central verification system that can instantly say “this one was made by Meta AI” versus “this one is genuine.” In the end, users are left to do the screening themselves — a job that shouldn’t have fallen on them in the first place.

What to watch next

The open question is whether Meta will bring this feature back with tighter guardrails — things like identity verification before generation, or an unremovable watermark.

In the meantime, there are things you can do right now: set your account to private if you don’t need it to be public, turn on two-factor authentication to prevent account impersonation, and check your Instagram privacy settings periodically, since Meta tends to change these policies often without much advance notice.

Another thing worth watching is whether other platforms like TikTok or X will follow suit. If Meta was willing to walk this feature back first, it could put pressure on other companies to reconsider their own. This story almost certainly won’t end with Instagram alone.