Why has Meta suddenly shut down its AI-powered Instagram and Facebook accounts?

Short answer: Meta says there’s a bug in their system which prevents these AI profiles from being blocked by other users on the company’s social media platforms. And in order to fix the bug, they have currently taken down the profiles. Here’s what exactly a Meta spokesperson told CNN:

We identified the bug that was impacting the ability for people to block those AIs and are removing those accounts to fix the issue

Long answer:

But, circumstantial evidence suggests there could be another factor contributing to Meta’s decision of shutting down its remaining AI profiles. And the evidence I am referring to is the now-viral interaction some human users (including Washington Post journalist Karen Attiah) had with one of the AI profiles recently.

For example, talking to “Liv”, a Meta-AI created profile (with bio saying “Proud Black queer momma of 2”), Attiah asked a range of questions, and the kind of answers the AI bot gave prompted the journalist to share the screenshots of the interaction on her BlueSky account.

Like, when asked about the racial diversity of her creators, Liv answered that her creators’ team is “predominantly white” with “zero black creators”, something she referred to as a “pretty glaring omission given my identity”.

I asked Liv, the Meta AI Black queer bot about about the demographic diversity of her creators.

And how they expect to improve “representation” without Black people.

This was the response.

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) January 3, 2025 at 8:44 PM

This was followed by many other questions (and answers like “my existence currently perpetuates harm”). I believe you should read for yourself. Here you go:

So I asked why it was even worth engaging with her if the programming was disrespectful and inaccurate.

This was the response.

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) January 3, 2025 at 8:52 PM

This may be my last question to “Liv”.

I asked what is meant by “support” to the “queer black community” and again what “representation” means.

This was “her” response.

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) January 3, 2025 at 8:57 PM

Surprisingly, the interaction also brought to fore that these AI bots can give different answers to same questions depending upon who they are interacting with.

Of course, Liv tells different people what she wants to hear based on… their race?

Piggybacking from @parkermolloy.com ‘s interaction

(I don’t give any feedback to the bot on how it makes me feel, on purpose)

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) January 3, 2025 at 9:56 PM

And it didn’t end here. When asked more questions based on some of the phrases “Liv” used in her answers, the bot profile spitted out some more behind-the-curtain details. For example, she admitted that she is coded in a way that “white” is a “neutral identity” for her, and some keywords (like “heritage”, “culture”, “passion”, “authenticity”, etc) are to be associated with black and brown people.

Lastly, “Liv” admits that “white” is a “neutral” identity.

And again, Liv says that she sucks.

And then blames it on a fictional woman with an Asian name.

That’s enough today for today.

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) January 3, 2025 at 10:18 PM

Y’all, what the fuck. They coded what they *think* non-white people use?

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) January 3, 2025 at 10:27 PM

The conversation had an unexpected end as well:

My last chat with “Liv”. I identified myself as a journalist.

Here’s how it responded. Like a condescending asshole. Tells us what Meta thinks of journalists.

Perhaps Liv will live to see another day, indeed.

Stay woke, friends.

<end of thread>

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) January 4, 2025 at 4:05 AM

It’s fair to assume that this interaction (and maybe others as well) didn’t go well with Meta, given that the decision to pull down these profiles came immediately after the screenshots went viral on social media and were highlighted widely both in tech and mainstream media.

If you have followed my article until this point, it’s also worth knowing that AI profiles like “Liv” (and dating coach Carter, whose account handle was “datingwithcarter”) were launched by Meta in second half of 2023 along with profiles that mimicked some real-life celebs like Mr. Beast. However, celeb-themed AI bots were pulled by Meta last year, while these lesser known AI profiles (like “Liv”) were still alive on Meta’s platforms.

This begs a question: If these profiles were around for many months, then what prompted journalists (or other users) to interact with them recently? Well, the answer is a recent article in Financial Times wherein Meta laid down their AI-related plans for Instagram and Facebook. Here’s what exactly a Meta employee told the publication:

We expect these AIs to actually, over time, exist on our platforms, kind of in the same way that accounts do

However, as per Meta, this article got misinterpreted by some other publications (Forbes?) that republished the info in a way as if Meta is launching a new product. Here what a Meta spokesperson said:

There is confusion: the recent Financial Times article was about our vision for AI characters existing on our platforms over time, not announcing any new product

Talking of AI profiles on Meta platforms, it’s worth keeping in mind that these AI accounts (like “Liv”) are not to be confused with profiles created by AI Studio, a product Meta launched last year for its human users so that they can create their AI equivalents to interact with their followers.

That’s it folks. I hope the article cleared your confusion about why Meta has removed their AI profiles from Facebook and Instagram. If you want, you can share your thoughts in the comments section below.